EMERGENCY MANAGEMENT IN SOCIAL MEDIA GENERATION

Transcription

1 EMERGENCY MANAGEMENT IN SOCIAL MEDIA GENERATION Deliverable 3.1 Usage Patterns of Social Media in Emergencies Christian Reuter 1, Thomas Ludwig 1, Therese Friberg 2, Matthias Moi 2, Rajendra Akerkar 3, Sylvia Pratzler Wanczura 4, Alexis Gizikis 5, Tony O Brien 5 University of Siegen 1, University of Paderborn 2, Western Norway Research Institute 3, Fire Department of Dortmund 4, European Emergency Number Association 5 June 2014 Work Package 3 Project Coordinator Prof. Dr. Ing. Rainer Koch (University of Paderborn) 7th Framework Programme for Research and Technological Development COOPERATION SEC : The impact of Social Media in emergencies

2 Distribution level Due date Public (P) 30/06/2014 (M3) Sent to coordinator 30/06/2014 No. of document D3.1 Title Status & Version Work Package Usage Patterns of Social Media in Emergencies Final 3: Analysis of Social Media in Emergencies Today and Tomorrow Related Deliverables D3.4, D3.5, D3.6 Leading Partner Leading Authors University of Siegen Christian Reuter, University of Siegen (Section 1, 2 and 6.1, Questionnaire, Overall Internal Review) Thomas Ludwig, University of Siegen (Section 1, 2 and 6.1, Questionnaire) Therese Friberg, University of Paderborn (Section 3 and 6.2) Matthias Moi, University of Paderborn (Section 2.3, 3 and 6.3) Rajendra Akerkar, Western Norway Research Institute (Section 4 and 6) Sylvia Pratzler Wanczura, Fire Department of Dortmund (Section 5 and 6.4) Alexis Gizikis, European Emergency Number Association (Section 5 and 6.4) Tony O Brien, European Emergency Number Association (Section 5 and 6.4) Contributors Marc André Kaufhold, University of Siegen (Section 2 and 6.1) Gasper Bizjak, Bert Brugghemans and Loek Pfundt, Federation of the European Union Fire Officer Associations (Questionnaire, Annex A) Thomas Spielhofer, Tavistock Institute (Overall Internal Review) Reviewers Keywords Reynold Greenlaw, Oxford Computer Consultants Social Media, Current Use, Usage Patterns This project has received funding from the European Union s Seventh Framework Programme for research, technological development and demonstration under grant agreement no

3 Table of contents List of Figures... I List of Tables... II Glossary... III 1 Introduction Abstract Purpose of the Document Target Audience Literature Review: Social Media in Emergencies Social Media and its Definition Overview of Cases in Literature Media Patterns Usage Patterns Citizens to Citizens (C2C) Authorities and Citizens (A2C and C2A) Systematization Role Patterns Types of Users in Social Media Role Patterns in Literature Systematization Summary Literature Review: Existing Data Quality Problems and Ways of Structuring Information Data Quality General DQ Problem Patterns Approaches for Resolving DQ Problems in Social Media Data Challenges in Structuring and Querying Information Review of Related Research Projects EU Framework Projects Other Projects... 25

4 4.3 Summary Interview Study: Social Media in Emergencies Methodology of Data Collection Participants Methodology and Details of Question Asked Collection of Results Results Organization Profile (Questions: A1 A4, A10) Type and Size of Organization PSAP Models Participant Position and Functionalities Risks, Vulnerabilities and Operational Size (Questions: A5 A8) Operational Tasks (Questions: B1 B3) Communication Channels (Questions: B4 B8) Current Use of Social Media (Questions: D1 D14) Potential Use of Social Media (Questions: D15 D23) Research on Social Media (Questions: D24 D27, F5) Challenges and Benefits of Using Social Media (Questions: E1 E3) Future Plans for Using Social Media in Emergencies (Questions: F1 F4) 43 6 Summary Social Media, Emergency Use and Role Patterns Data Quality Patterns Related Research Projects Interview Study References Section Section Section Annex A: Interview Questions Annex B: Identification of Interviews (for internal use only)... 59

5 List of Figures Figure 1: Crisis Communication Matrix [ReMP12], adapted concerning the terminology... 8 Figure 2: Types of users in the real, digital and official hemisphere Figure 3: Model 1: Emergency Response Organizations handling emergency calls (Source: EENA) Figure 4: Model 2: Filtering stage 1 PSAP and resource dispatching stage 2 PSAPs (Source: EENA) Figure 5: Model 4: Data gathering by stage 1 PSAP, resource dispatching by stage 2 in an integrated control room (Source: EENA) Figure 6: Model 5: Emergency Response Organizations independent PSAP (Source: EENA) I

6 List of Tables Table 1: Overview of Cases in Literature... 6 Table 2: Classification of Social Media by presence/media richness and self presentation/self disclosure [KaHa10]... 6 Table 3: Usage Patterns Table 4: The referenced EU Framework projects based on common aspects of social media analysis Table 5: The referenced other projects based on common aspects of social media analysis Table 6: Organization Profile (Source: May 2014 Survey of authorities) Table 7: Characteristics of the analyzed interviews (Source: own elaboration) Table 8: Operational Tasks (Source: own elaboration) Table 9: Participant involvement in four stages of an emergency (Source: own elaboration) Table 10: Communication Channels for inbound messages (Source: own elaboration) Table 11: Communication Channels for outbound messages (Source: own elaboration) Table 12: Communication Channels for broadcasting warning messages (Source: own elaboration) Table 13: Current use of Social Media (Source: own elaboration) Table 14: Benefits of Social Media use (Source: own elaboration) Table 15: Difficulties of using Social Media (Source: own elaboration) II

7 Glossary Abbreviation EmerGent EMS EMC Expression Emergency Management in Social Media Generation Emergency Management Services Emergency Management Cycle A2A Authorities to Authorities (section 2.4) A2C Authorities to Citizen (section 2.4) C2A Citizen to Authorities (section 2.4) C2C Citizen to Citizen (section 2.4) V&TC Volunteer and Technical Communities (section 2.2) DQ Data Quality (section 3) OASIS CAP OASIS Common Alerting Protocol (section 4) GIS Geographical Information System (section 4) R&D Research & Development (section 4) PSAP Public Safety Answering Points (section 5.1) III

8 1 Introduction 1.1 Abstract This deliverable will present the results of our analysis the use of Social Media before, during and after emergencies. It is based on a comprehensive study of the Social Media related behavior of authorities (among EMS) and the general public in past emergencies. It provides insights into the current use of Social Media in emergencies; this includes: a discussion of a wide range of recent papers on the use of Social Media in emergencies (section 2) leading to usage patterns, a typology of the use of Social Media in crisis management (section 2.3) and role patterns, the different types of Social Media users during an emergency (section 2.5), examples and typologies of data quality problems as well as approaches for resolving these problems (section 3), an overview of the various other related projects of relevance to EmerGent (section 3), results of own interviews with authorities about their current and prospective use of Social Media (section 3). 1.2 Purpose of the Document The aim of this document is as follows: Identify and map several types of users, in terms of motivation, attitudes and behavior of both authorities and the public, based on past emergencies and a stakeholder analysis. Provide an accompanying analysis of the current use of Social Media based on different role patterns both from the citizen and the authorities perspective and both from the producer and consumer perspective (information provider and/or information dissemination), which lead to use cases of Social Media in emergencies. In order to archive this purpose the following tasks have been performed: We have reviewed related work on Social Media in emergencies in order to identify usage patterns as well as role patterns (section 2). Furthermore we have reviewed related work on data quality problem patterns in emergencies (section 3). A third review summarized related research projects (section 4). Based on these foundations we will present the results of our interview study, which has been conducted in 7 European countries and 8 different cities and provided insights in current use of Social Media, which is sometimes not covered in research (section 5). Finally we will summarize our findings. 1.3 Target Audience Researchers in Social Media for emergencies Authorities interested in Social Media 1

9 2 Literature Review: Social Media in Emergencies This section summarizes the use of social media in emergencies. First we will define the term social media and present various applications that fall under this definition and can be distinguished (section 2.1). Then we will give an overview about documented cases of social media use in research papers (section 2.2). Based on these foundations we are going to present media patterns (section 2.3) as well as to analyze the findings of these cases towards both usage patterns and role patterns: First, and based on a distinction of citizens (C) and authorities (A), we distinguish different usage patterns (C2A, C2C, A2C, A2A) and present the findings from literature related to different usage patterns (section 2.4). Then, and based on the same cases, we summarize role patterns identified in literature and classify them in an own and novel form (section 2.5). Finally we will present a summary (section 2.6). 2.1 Social Media and its Definition For more than one decade Social Media has been used in crisis management [ReMP12]: already after the terrorist attacks of September 11 (2001) wikis, created by citizens, were used to collect information about missing people [PaLi07]. In a study on the information search behavior during the forest fires in southern California in 2007, it was found that affected people communicated via mobile phones and used the Internet to search for information and to trigger off any kind of communication, and to read blogs, news sites and forums [SuPS08]. During the last years this behavior even increased: Social Media are widely used by citizens collaboratively coping with a crisis (see various cases in section 2.2). Social Media the most generic term is defined as: Group of Internet based applications that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of User Generated Content [KaHa10] User Generated Content can be seen as the sum of all ways in which people make use of Social Media [KaHa10]. Various classes of social software applications can be distinguished [EbGH08]. The following descriptions are based on [ReMP12]: Wikis aim at a collaborative accumulation and creation of information and knowledge. They are useful to collect knowledge of a topic based on one s own research. In a crisis, this activity may be done by people who are not seriously affected by the crisis. Blogs support the publishing of information and self expression through an individually owned journal. They are not useful as a fast response as they often contain longer personal entries. Micro blogging is an alternative. The most prominent application is Twitter. It allows entries limited to 140 characters, similar to text messages. Users can publish messages (tweets) on their site and tag words (#hashtag) within a message. With the help of these tags, certain topics can be found. It is possible to address other users with notation. It is also possible to publish tweets by sending text messages using a mobile phone. Due to the tweet s text message like structure, tagging, the ease with which tweets about specific topics can be found, its huge dissemination with over 600 million users worldwide and its mobile usability, Twitter is a significant social medium for crises. 2

10 Social Network Sites (SNSs) are defined as: web based services that allow individuals to (1) construct a public or semi public profile within a bounded system, (2) articulate a list of other users with whom they share a connection, and (3) view and traverse their list of connections and those made by others within the system. The nature and nomenclature of these connections may vary from site to site. [BoEl07]. SNS were originally used for relationship management, self expression and communication, but now these applications are incorporating more and more functionality related to blogging and the exchange of information. An example for a social networking service that offers good opportunities for crisis communication is Facebook. In addition to networking with people, contacting organizations, companies or celebrities is possible. Each profile includes a wall on which people can leave messages or where the owner of the profile can post their own status messages. The entries can include text, links, photos, videos and other users can add comments. On the wall a dialogue between individuals and organizations can take place. Facebook provides a client for various mobile devices making it possible to upload profile information from almost everywhere. People with older devices can receive updates through text messages and in turn update their own pages by sending their own messages via text message. Lastly, the wide acceptance of Facebook (there are about 1300 million active users worldwide) makes it the largest online community. Social sharing enables the user to generate and categorize digital content. Photo and video communities are the most interesting for crisis communication. Flickr for sharing photos and YouTube to share video clips are the most popular platforms. Both services allow the indexing of uploaded content using self selected tags describing the content and making a targeted search possible. Flickr offers several ways to interact with the website; one option is a website optimized for mobile devices, another is uploading photos via e mail. YouTube offers a mobile version of the page, which supports the recording and uploading of clips by using a mobile phone. Flickr also lets users geo tag pictures that can then be placed on a map, or be displayed with additional image information using EXIF data. The selection of the most important representatives is based on the MAU Value (Monthly Active Users), which counts unique, active users over 30 days [Ball13]. Users are considered to be active if they are unique and have logged in at least once in the month. We furthermore focus on systems which are well known in Europe. An additional restriction is the minimum value of 100 million. This provides the following list [ReSc14]: (1) Facebook (1.15 billion MAUs) (2) YouTube (1 billion) (3) WhatsApp (350 million) (4) Google+ (327 million) (5) Tumblr (300 million) (6) Twitter (240 million) (7) LinkedIn (184 million) (8) Instagram (150 million). 3

11 2.2 Overview of Cases in Literature The majority of research so far focuses on crises and emergencies in the USA and deals with Twitter [ReMP12]. Table 1 summarizes the referenced literature of section 2 in a structured way by providing an overview of studies with regard to their reference, the related case or scenario, a brief overview of the scientific contribution and a keyword. The cases are sorted by the year the event took place. They have been identified while searching in Google scholar for the keywords social media, emergency, disaster, crisis. However, due to the amount of studies, only the most appropriate (related to the project) have been selected in order to provide an overview. Reference Case Contribution Keyword [LiPS08] [EnRe14] [ShPS08] [HuPa09] [QuWW09] [Sutt10] [HeZa10] [LaSh11] 2004 Indian Ocean tsunami 2005 Hurricane Katrina, 2010 volcano Eyjafjallajökull in Iceland 2007 Southern California wildfires 2008 hurricanes Gustav and Ike 2008 Sichuan earthquake 2008 Tennessee River technological failure 2009 attack on four police officers in Lakewood, Washington 2009 Los Angeles Fire Department Photo repository sites were used by citizens to exchange information. Indicates that the perceived credibility of Social Media information is less than of printed, official online or televised news and information from family, relatives or friends. Photo repository sites were used by citizens to exchange information. Depicts differences between the use of Twitter in crises and the general use. Outlines that people gather and synthesize information. Outlines the phenomena of broadcasting. Shows the ability of Twitter to organize and disseminate crisis related information. Public Information Officers highlight the importance of the information evangelist within organizations. Photo Sharing Information Credibility Backchannel communication Microblogging Information synthesis Broadcasting Types of Tweets Information Evangelism [StPa10] 2009 Oklahoma Fires Highlights the role of retweeting. Collective Intelligence [VHSP10] 2009 Red River Floods Highlights broadcasting by people on the ground as well as activities of directing, relaying, synthesizing, and redistributing. [Birk12] [StPa11] [Star13] [ReMP12] 2010 Bornholm blizzard 2010 Haiti earthquake 2010 Haiti earthquake 2010 mass panic at the Love Parade music festival in The emergence of two Facebook groups in Denmark shows that the geographical location and self selection into groups create different views of a crisis situation Was analyzed with the help of translators and reveals the phenomenon of digital volunteers. Examines collective intelligence as transformations of information through activities like relaying, amplifying, synthesizing and structuring information in the wake of disaster events. Outlines the need for duplex communication. Situational Awareness Emergent Groups Digital Volunteers Collective Intelligence Crisis Management 4

12 Reference Case Contribution Keyword [NaVS12] [StPa12] [Wile14] Germany, 2010 volcano Eyjafjallajökull in Iceland 2010 San Bruno Californian gas explosion and fire disaster 2011 Egyptian uprising 2011 Great East Japan Earthquake Illustrates that sentiment analysis (analysis for identifying and extracting subjective information by using language processing and liguistic approaches) with emotions performed 27% better than Bayesian Networks alone. Shows how the crowd expresses solidarity and does the work of information processing through recommendation and filtering. Emphasizes the use of Twitter to provide emotional support and mentions the problem of widely publishing obsolete or inaccurate information and the unequal distribution of useful information. [PBWH12] 2011 Norway attacks The notion of peripheral response has been developed in relation to emergent forms of agile and dialogic emergency response. [Jenn12] 2011 San Diego / Southwest Blackout [DeHP12] 2011 Shadow Lake fire The availability of Social Media illustrates that contrary to expectations, the cell phone system did not have the expected availability, and as a result, users had a difficult time using Social Media to status/contact family and friends. Shows the deployment of trusted digital volunteers as a virtual team to support a incident management team. [ReHP13] 2011 Super Outbreak Distinguishes groups of twitterers, such as helpers, reporters, retweeters, and repeaters. [WMAR13] 2011 Tunisian revolution Social media linked the young activists with actors in other cities. [YCLL13] 2012 hurricane Isaac Leads to knowledge which classification algorithms work best in each phase of emergency. [HDPA14] 2012 hurricane Sandy Shows that few departments used online channels in their response efforts and that communication differed between fire and police departments and across media types. [Amer12] [FAAB13] [KaRe14] [BeLG13] 2012 Online and Telephone Survey 2013 European Flood in Germany 2013 European Flood in Germany 2012 Analyses the Social Media use of government States that 25% of the participants will download an emergency app and 12% of the general public used Social Media in crises. Confirms the potential of Twitter as a distributed social sensor but at the same time highlights some caveats in interpreting immediate results. Identifies challenges (1) clarity and representation of relevant content, (2) moderation and autonomous work, (3) feedback and updates in interaction relationships and (4) integration of technologies and interaction types of designing Social Media for volunteers in emergencies and depicts the role of moderators. Presents an account type typology containing high level organizational accounts, accounts for formal functions and roles, formal personal accounts and affiliated personal Sentiment Analysis Information contagion and diffusion Commuters Peripheral response Availability of Social Media Trusted Volunteers Volunteers Political Four Phases of Emergency Management Officials Social Media use Aid Organization Visual Analytics Moderators and Design Challenges Emergency Response 5

13 Reference Case Contribution Keyword [CMPB14] [Gorp14] authorities 2013 Investigates the current tools, work practices and ad hoc collaboration of distributed digital volunteer teams 2014 Investigation of V&TCs and aid organizations 2.3 Media Patterns accounts. Identifies design implications for integrating the activities of distributed volunteers, facilitating connections between different tools and tasks, enabling team members to share their actions and using current work of volunteers to train machine learning algorithms. Categorizes Volunteer and Technical Communities into software platform development communities, mapping collaborations, expert networks and data aggregators and identifies six barriers of collaboration with aid organizations: limited resources, the management of volunteers, different levels of engagement,the level of commitment by V&TCs, different ways of working and the aid organizations limited knowledge about the V&TCs expertise. Table 1: Overview of Cases in Literature Digital Volunteers Volunteer and Technical Communities (V&TCs) Independently from the concrete implementation, Social Media can be categorized in several ways. Focusing on the presence/media richness and self presentation/self disclosure [KaHa10] could be a first basis to categorize different types of Social Media. Social presence / Media richness Low Medium High Self presentation / Selfdisclosure High Low Blogs Collaborative projects (e.g. Wikipedia) Social Network Sites (e.g. Facebook) Content communities (e.g. YouTube) Virtual social worlds (e.g. Second Life) Virtual game worlds (e.g. World of Warcraft) Table 2: Classification of Social Media by presence/media richness and self presentation/selfdisclosure [KaHa10] The most promising categories regarding Social Media in emergencies are defined by the categories in the middle where the social presence / media richness is medium. They can be used independently from the self presentation aspects. Users are virtually always online in their communities and communication is asynchronous which simplifies the spread of information. The second important category is build up by Social Media within content communities. They have a low self presentation, because the content and not the users are in focus. That means that Social Networks like YouTube (Video) or Flickr (Pictures) are very powerful in how they distribute content regarding their focused media type. By interconnecting these two categories an even more powerful content and media sharing with high self presentation is made possible, though care must be taken not to create personally identifiable data through 6

14 making these connections. Past events like Hurricane Sandy 2012 or Haiti earthquake 2010 have shown the advanced usage of Social Media from those categories [OKR10], [OAR13]. But there are some challenges, which are sometimes opportunities that have to be taken into account. For instance, the integration of Social Media into workflows of EMS is a very ambitious task, because many problems like data quality concerns, reliability of data, etc. exists if you look into the details. Bontcheva and Rout have summarized some interesting characteristics of data streams from Social Media from a data centric / information processing point of view in [BoRo12]. The main challenges and opportunities to process Social Media data are: Short messages: Most messages are very short. It s often hard to identify the sense without looking on the context. Noisy Content: The spelling in Social Media is often unusual (for instance, cya meaning see you or 2moro which stands for tomorrow ) Temporal: In addition to linguistic analysis, social media content lends itself to analysis along temporal lines, which is a relatively under researched problem [BoRo12]. Social Context: In order to interpret messages, feelings etc. correctly you have to make use of the social context (E.g. connection to other users, personal interests, previous posts) User Generated: Since the consumer and the producer of content are the users itself, there is a rich source of implicit and explicit information available about the users (f.i. interests, location, age, etc.) Multilingual: Produced content is very multilingual. E.g. less than 50% of the tweets are in English [CWT13]. 2.4 Usage Patterns In terms of the communication matrix for Social Media in crisis management [ReMP12], four different use patterns for using Social Media in such a situation can be distinguished (Figure 1) depending on a distinction of (a) authorities and the (b) public as the (i) sender and the (ii) receiver of information. The reason why authority was chosen in is that (especially with Social Media), citizens may be communicating with somebody that is not an EMS (e.g. local police, county council). Based on this, different use patterns (Figure 1) can be distinguished: 7

15 Citizen/Public Authorities to Citizens (A2C): Crisis Communication Citizens to Citizens (C2C): Self Help Communities Receiver Authority Authorities to Authorities (A2A): Inter organizational Crisis Mgmt. Citizens to Authorities (C2A): Integration of Citizen Generated Content Authority Citizen/Public Sender Figure 1: Crisis Communication Matrix [ReMP12], adapted concerning the terminology Citizens and Citizens (C2C) (top right): On the public level, citizens and volunteers communicate with each other, via conventional media like telephones or Social Media such as Twitter or Facebook. Volunteerism in crisis situations is not a new phenomenon: Nearly 40 years ago [QuDy77] as well as [StQu85] characterized these emergent groups as private citizens who work together in pursuit of collective goals relevant to actual or potential disasters but whose organization has not yet become institutionalized. According to [Quar84] the essential conditions for the emergence of such groups are (a) a legitimizing social environment, (b) a perceived threat, (c) a supporting social climate, (d) a network of social relationships, and (e) the availability of specific (immaterial) resources. According to some studies citizens react in a largely rational way rationally to crisis situations, rarely panic, are not helpless, and do not loot [HeRu04]. They are instead capable of taking part in a large amount of rescue and response work. [ReHP13] distinguish between activities in the real and the virtual world: real emergent groups [StQu85], which usually act in the form of neighborly help and work on site, and virtual digital volunteers [StPa11], who originate from the Internet and work mainly online. Authorities and Citizens (A2C and C2A): Crisis communication (A2C) (top left corner in Figure 1) or public alerting is a second case for the use of Social Media during emergencies. Often authorities inform citizens using Social Media sometimes for one directional communication, at other times also taking answers and comments into consideration. The integration of citizen generated content (A2C) (bottom right) is supported by various tools. Acknowledging the affected citizens as a responsible actor in crisis management raises the potential of benefitting from citizen generated content, for example to illustrate problematic situations through photographs taken with mo 8

16 bile phones. The perceived unreliability of such information is a significant obstacle in exploring such opportunities (see Section 3). This could be alleviated by crowdsourcing strategies to confirm the trustworthiness of information visible on a picture [ReMP12]. Authorities and Authorities (A2A): The inter organizational collaboration (bottom left corner in Figure 1) of authorities is often not supported by Social Media such as Twitter or Facebook. Social media can help to improve inter organizational awareness and informal processes. [WPKH09] examined the potentials of online social networks with EMS students: Sharing information, communication and networking were the most popular features. They also show that possible concerns against those systems may be information integrity, user identification, privacy and technology reliability. In the InfoStrom project an inter organizational collaboration platform for authorities has been created [PRLL13]. Furthermore, authorities may use Social Media for internal communication. However, in this review, this pattern will not be explored in details as it does not directly involve citizens Citizens to Citizens (C2C) This section summarizes Social Media usage patterns of C2C communication and gives an insight into the types of information, such as situational updates or emotion related information, and activities in the sphere of information. While there are issues in the unorganized behavior in or chaotic use of Social Media among citizens, many studies illustrate implications, recommendations and challenges in designing social software for C2C communication and collaboration that should be considered in the development of technological artifacts within the EmerGent project. Thereby the distribution and growing use of mobile phones during emergencies suggest the consideration of cross platform solutions. Many contributions cover the communication among citizens. The use of Twitter has been analyzed scientifically in the context of various crises such as the 2008 hurricanes Gustav and Ike leading to the observation of differences between the use of Twitter in crises and its general use [HuPa09]. The 2008 Tennessee River technological failure outlines the phenomena of broadcasting emergency relevant information via Twitter, exceeding the boundaries of locally limited networks and raising emergency awareness [Sutt10]. The analysis of the 2008 Sichuan earthquake outlines that people gather and synthesize information [QuWW09], while the case of the 2010 Yushu earthquake in China shows that people use microblogging to seek information about the status of the emergency or people, e.g. obtaining situational updates or searching for missing people [QHZZ11]. The classification into and relative distribution of opinion related (33%), situation update (25%), general related (18%), emotion related (16%) and action related (4%) tweets gives an insight into the different ways citizens use Social Media in emergencies. A study of the 2009 Red River Floods highlights activities of broadcasting emergency information by people on the ground as well as of directing, relaying, synthesizing, and redistributing [VHSP10]. They analyzed how Twitter can contribute to situational awareness the entire available knowledge which can be integrated into a coherent picture to assess and response to a situation [ReGK07]. The comparative study indicates that tweets including situational updates or geo location information are more likely to be retweeted than those with other event related information. Concerning the EMC it can be seen that citizens used Social 9

17 Media for preparatory activities previous to the rather predictable floods whereas for the spontaneous wildfires the activities, for the most part, started in the response phase. Studies on the 2009 Oklahoma Fires highlight the role of retweeting [StPa10]; the 2009 attack on four police officers in Lakewood, Washington shows the ability of Twitter to organize and disseminate crisis related information [HeZa10]. The Twitter usage during the 2010 Haiti earthquake was analyzed with the help of Tweak the Tweet translators and reveals the phenomenon of digital volunteers [StPa11] outlining that they can converge to strongly intertwined networks where each participant seems to increasingly identify himself with the associated role. Further correspondent Social Media data was later on examined regarding collective intelligence defined as an universally distributed intelligence, constantly enhanced, coordinated in real time, and resulting in the effective mobilization of skills [Lévy94] as transformations of information through activities like relaying, amplifying, synthesizing and structuring information in the wake of disaster events [Star13]. Another study on Haiti suggests that whenever there is greater uncertainty reduction needed, there will be a larger amount of collaboration on the platform, while uncertainty is caused by redundant information and mistakes due to the chaotic unorganized online behavior of volunteers [VaOR13]. A case study of a blizzard on the Danish island Bornholm in December 2010 [Birk12] explored two Facebook groups, which were created for different motives and backgrounds in the course of this event. The comparative case study revealed that the two Facebook groups constructed different views on the emergency and could also be distinguished by the geographic position and purpose of their members. Members of one group focused on managing the emergency, the other group focused on obtaining situational information. A study on the 2011 Egyptian uprising shows how the crowd expresses solidarity and does the work of information processing through recommendation and filtering by structuring and disseminating information from the affected area [StPa12]. While a study on the 2011 Great East Japan Earthquake emphasizes the usefulness of Twitter to provide emotional support, it also mentions the risks of widely publishing obsolete or inaccurate information as well as the unequal distribution of useful information [Wile14]. A study on the 2012 hurricane Sandy shows that citizens handle those activities that are unlikely to be done by official emergency services like recover lost animals [WhPA14]. A survey about the availability of Social Media during the 2011 San Diego/Southwest Blackout illustrates that the cell phone system did not have the expected availability, and as a result, users had a difficult time using Social Media to status/contact family and friends [Jenn12]. A poll of the American Red Cross states that 75% of the population will contact friends to see if they are safe, 37% will purchase supplies or will seek safe shelter and 25% will download an emergency app [Amer12]. Only 12% of the general public used Social Media during crises mainly for seeking information about weather, traffic and damage caused as well as sharing weather information, reassurance of safety and feelings about the emergency. Based on a timeline and qualitative analysis of information and help activities during the Super Tornado Outbreak in April 2011, [ReHP13] suggest a classification of twitter users in different roles, which are named helper, reporter, retweeter and repeater (additional details are provided in section 2.5). The timeline analysis further indicates that, while the preparedness and response phases contained warning and crisis tracking activities consisting of a high number of retweets, virtual self help communities start their relief activities in the recov 10

18 ery phase along with a relatively increasing number of external resource links. Furthermore, this study has identified (a) the integration of ICT for volunteers into existing networks, (b) the fostering of voluntary groups coming into existence, (c) connections between virtual and real activities, and (d) interfaces to official authorities as design requirements for volunteerrelated software. Based on interviews with authorities and digital volunteers, [CMPB14] suggest (a) the coordination and integration of voluntary activities, (b) the connection between different tools and tasks as well as (c) the possibility to share own activities in order to generate learning effects for spontaneous and less experienced volunteers. Additionally, [KaRe14] examined the use of Social Media during the 2013 European Floods in Germany identifying challenges to support digital and real volunteers in achieving (a) clarity and representation of relevant content, to facilitate processes of (b) moderation and autonomous work as well as promoting (c) feedback and updates in interaction relationships and to (d) integrate technologies and interaction types Authorities and Citizens (A2C and C2A) This section illustrates the benefits of the cooperation between citizens and authorities through Social Media. Volunteer work might reduce personnel issues, increase the reach of information gathering and contribute to situational awareness, but also indicates problems of legality, barriers of collaboration, mobilizing resources to scan huge amounts of information and increasing risk through hasty actions of citizens in emergencies. There are different approaches and tools for the analysis of citizen generated content. Thus, the EmerGent project should aim on amplifying benefits and reducing problems of interaction via Social Media, address the need for duplex communication and store law compliant research data for longterm analysis. Besides the use of Social Media for communication among citizens, the role of authorities as producer or receiver of information has also been studied. The 2010 mass panic at the Love Parade music festival in Germany as well as the 2010 volcano Eyjafjallajökull in Iceland outline the need for duplex (two way) communication between emergency services and citizens and for the seeking of information about people [ReMP12]. A comparative study of police units in the 2011 London riots discusses he benefits and challenges of instrumental and publicincluding expressive communication approaches through Twitter [DeBK13]. While the expressive approach, for example, allows close relations to the public and increases the possible reach, it requires high maintenance and has issues on easily overstepping boundaries like auditing the legality of information publication and learning when to engage and how to resource it. A survey on Social Media in the emergency management field describes the penetration of Social Media in state, county, and local authorities as incomplete, where many aspects are still new to authorities [SaWT13]. The main barrier identified is the lack of personnel to take on additional responsibilities, although the largely untapped resource of digital volunteers could help to alleviate some personnel issues. [KFSY11] analyzed the use of Social Media by government officials in Virginia and identified huge requirements concerning the overwhelming amount of data. [StSt10] recommend the use of structured crisis specific Twitter hashtags to increase the utility of information generated during emergencies and to facilitate machine parsing, processing and re distribution. They propose the Tweak the Tweet microsyntax which requires a main event, a main category, a location and multiple additional 11

19 tags. Furthermore collaboration among humanitarian aid organizations and Volunteer and Technical Communities (V&TCs) was analyzed in an exploratory study, categorizing the latter into software platform development communities, mapping collaborations, expert networks and data aggregators [Gorp14]. It further identifies six barriers of collaboration with aid organizations: (a) limited resources, (b) the management of volunteers, (c) different levels of engagement, (d) the level of commitment by V&TCs, (e) different ways of working and (f) the aid organizations limited knowledge about the V&TCs expertise. In a study on the 2011 Shadow Lake Fire, [DeHP12] showed the positive effects of the integration of trusted digital volunteers for the support of a Type I incident management team, which is a federally or state certified team with superior training and experience in the USA, by providing an approach to integrate Social Media channels into authorities efforts. The scenario based survey of [EnRe14] indicates that the perceived credibility of Social Media information is less than of printed, official online or televised news and information from family, relatives or friends. The 2009 case study of Public Information Officers (PIO) of the Los Angeles Fire Department highlights the importance of the information evangelist, a person who promotes the use of new forms of media and technology, in this case Social Media, within authorities to achieve an effective organizational utilization of Social Media [LaSh11]. A study on 2012 hurricane Sandy also shows that communications differed between fire and police departments and across media types [HDPA14]. They suggest making online media streams more listenable for on the ground emergency managers, new features and/or tools are needed that allow emergency managers to better track, respond to, and document public information. For that reason the study by [Pohl13] summarizes existing frameworks and tools developed in the context of crisis related (e.g. Twitcident or Tweak the Tweet ) and noncrisis related (e.g. Twitinfo) research work to analyze Social Media or to include new functionalities into the Social Media usage for crisis management. The comparison reveals that there are systems for (a) different applications, (b) considering one or several Social Media platforms for monitoring, (c) especially developed for crisis management and (d) performing different kinds of analysis: monitoring, event detection, sentiment analysis. A study about the 2011 Thailand flooding disaster describes the authorities actions taken to correct the mistakes caused by the emerging risks of the chaotic use of Social Media [ChCR12]. [PBWH12] also showed this negative impact in their study about the 2011 Norway attacks where the notion of peripheral response has been developed in relation to emergent forms of agile and dialogic emergency response. But another study postulates the emergence of Social Media as boundary objects from a collective action perspective to bridge cognitive, relational and social boundaries among entities involved in emergency response [TiRa13]. This is an interesting effect of socializing and group building during Social Media and shows that citizens develop their own dynamics during emergencies which should be considered during EmerGent. A recent contribution on the 2013 European Flood confirms the potential of Twitter as a distributed social sensor but at the same time highlights some caveats in interpreting immediate results [FAAB13] Systematization Different Social Media technologies also have different features which make them more effective for different emergency management objectives. For example, Facebook is passive in that 12

20 if a person does not have a smart phone, updates are not automatically sent, whereas Twitter can send messages to anyone with a cell phone who has subscribed to the service. This makes Twitter a more effective Social Media tool for some cases in emergency communication (e.g. sending emergency alerts) than sites such as Facebook. Technologies such as Facebook and YouTube may be more effective in the pre disaster stage or in communication. Our suggestion is to distinguish between citizen (C) and authorities (A) as two groups and to combine them to A2A, A2C, C2A and C2C communication. As long as emergent does not focus on A2A we do not have to consider it. Furthermore a distinction between the four phases, prevention, preparedness, response and recovery seem reasonable in order to distinguish between different usage patterns: A2C C2A C2C Prevention A2C Prevention C2A Prevention C2C Prevention Preparedness A2C Preparedness C2A Preparedness C2C Preparedness Response A2C Response C2A Response C2C Response Recovery A2C Recovery C2A Recovery C2C Recovery 2.5 Role Patterns Types of Users in Social Media Table 3: Usage Patterns The research regarding types of users in Social Media started with identifying individual roles and succeeded with the suggestion of role typologies. [HuPa09] identified information brokers who collect information from different sources to help affected citizens. In a second step [StPa11] recognized the action of remote operators as digital volunteers who progress from simple internet based activities like retweeting or translating tweets to more complex ones, e.g. verifying or routing information. To further differentiate potential user roles, [ReHP13] distinguish between activities in the real and the virtual world: real emergent groups [StQu85], which usually act in the form of neighborly help and work on site, and virtual digital volunteers [StPa11], who originate from the Internet and work mainly online. Based on a timeline and qualitative analysis of information and help activities during the Super Outbreak in April 2011, [ReHP13] suggest a classification of the twitter users in different roles, which are named helper, reporter, retweeter, repeater and reader. As part of the investigation of Social Media activities during the European Floods 2013, [KaRe14] revealed the role of the moderator as an enhancement of the previous role typology: Role Patterns in Literature The following list presents terms that authors have used to describe different (overlapping) social media users in crisis: 13

21 Emergent Groups [StQu85] are private citizens who work together in pursuit of collective goals relevant to actual or potential disasters but whose organization has not yet become institutionalized. Digital volunteers [StPa11] is described as an element of the phenomena popularly known as crowdsourcing during crises. Voluntweeters [StPa11] are digital volunteers in the twitter sphere. Community scouts [ReMP11] are proposed amateur first informers to overcome the perceived unreliability of Social Media information for authorities. Trusted digital volunteers [DeHP12] were used during the 2011 Shadow Lake fire in virtual teams to inform a Type I incident management team about Social Media activities. Helpers [ReHP13] provide emotional assistance and recommendations for action, offer and encourage help and are involved in virtual and real activities. Reporters [ReHP13] integrate external sources of information, thus providing generative and synthetic information as a news channel or eyewitness. Retweeters [ReHP13] distribute important derivative information to followers or users. They correspond with the already identified information broker by [HuPa09] Repeaters [ReHP13] generate, synthesize, repeat and distribute a certain message to concrete recipients. Readers [ReHP13] are passive information catching participants who are interested in or affected by the situation. Moderators [KaRe14] establish supportive platforms for real and virtual activities, mediate offers of and requests for assistance, mobilize resources and integrate information of media, organizations and authorities. While the previous role descriptions and models address the public use of Social Media, [BeLG13] examined the utilization of Twitter by authorities and suggest an account typology containing "high level formal organizational accounts", "accounts for formal functions and roles", "formal personal accounts" and "affiliated personal accounts": High level formal organizational accounts are used to formally inform the public about ongoing events in a unidirectional way of communication. Accounts for formal functions and roles distribute information about certain entities, retweet other civil security actors and maintain a bidirectional communication. Formal personal accounts disseminate role specific information and references of official work or actual topics varying between strict professional and social use. Affiliated personal accounts are used for an expressive dissemination of information, personal opinions, reflections and social conversation. [ZKTE14] distinguish between a) communication with family and friends, b) situational knowledge acquisition and c) organization of help and support as usage patterns of Social Media during crises. While communication with family and friends is a common activity to share thoughts and contents in the daily routine, it provides a communication channel among affected relatives to relieve phone lines for emergency response in exceptional situations. Eyewitnesses might support the situational knowledge acquisition through the fast distribution of information in terms of texts, links, pictures or videos across Social Media. The increas 14

22 ing usage of mobile phones strengthens the citizens capabilities to support situational awareness. Furthermore Social Media provide potentials for spontaneous organization of help and support in crises. Volunteers and affected citizens, for instance, created Facebook groups or transferred information onto Google Maps without the influence of relief or crisis management organizations Systematization Figure 2 seeks to categorize types of actors, who somehow participate in an emergency situation, into three different areas: The (a) real (non digital), which contains citizens facing the emergency on the ground, (b) the virtual, where distant citizens and volunteers utilize Social Media technologies to inform about or aid in the emergency management and (c) the official, consisting of involved authorities entities. However, these three areas are connected: authorities from the official, for instance, need to solve emergencies in the real, while they also need to gather eyewitnesses information from the virtual. Affected citizens and emergent groups are a defined phenomenon non digital world, while voluntweeters, repeaters, retweeters and readers strictly act in the virtual. Helpers might be involved in real and digital help activities with a varying focus as they can descend from the real or virtual. Reporters integrate information from the real, virtual and official. Moderators mainly act in the virtual, but can be involved in or initiate real activities and interact with or get information from official authorities. While organizational accounts disseminate formal information into the digital world, affiliated personal accounts are used to publish more personal information. Roles and functions as well as formal personal accounts primarily act as official responders in the real world. 15

23 Repeater Affected Citizen Helper Reader Emergent Groups Real Reporter Virtual Retweeter Moderator Voluntweeter Official Personal Affiliated Personal Roles and Functions Organizational Official authorities Figure 2: Types of users in the real, digital and official hemisphere 2.6 Summary This section provided a definition of social media Group of Internet based applications that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of User Generated Content [KaHa10] (section 2.1). After presenting an overview about documented cases, where social media has been used in crisis (section 2.2) and presented media patterns (section 2.3), we derived usage patterns related to citizen (C) and authorities (A) as two groups and to combine them to A2A, A2C, C2A and C2C communication (section 2.4). Finally we considered role patterns, which we have structured according to real and virtual activities as well as authorities and citizen. Within this structure we have implemented the different role patterns found in literature (section 2.5). 16

24 3 Literature Review: Existing Data Quality Problems and Ways of Structuring Information Data on Social Media before, during and after emergencies has a high potential to strengthen the ties between authorities and citizens. For example, the Chi Chi earthquake 1999 in Taiwan and Hurricane Katrina 2005 in US have proven that current telephone, radio and televisionbased emergency response systems are not capable of meeting all of the community wide information sharing and communication needs of residents and responders during major disasters [HCH10]. For both sides (authorities and citizens) there are many concerns regarding the quality of Information criteria (like reliability, correctness or relevance). Further, authorities especially have to be able to handle the huge amount of available data in time. This chapter focuses on existing general data quality 1 (DQ) problems. After the definition of DQ (section 3.1), patterns to classify DQ problems and the resulting consequences are described (section 3.2). In a next step general challenges of Social Media data are defined, followed by challenges in structuring and querying information (section 3.3) and challenges and opportunities of information streams (section 3.4). 3.1 Data Quality In general, poor DQ can create chaos [SLW97, p.38] and the more data sources are involved the more problems might appear. To understand what a DQ problem is, a definition is given: A DQ problem is defined as a situation in which the content or medium of information does not meet the requirements of its producers, administrators, or users. It can be expressed in terms of information attributes that are clustered according to their origin, consequence or responsibility. [Eppl06, p.45] It must be understood that DQ problems encompass much more than incorrect information values ([WaSt96], [WSF95]). They also include production problems and errors, technical problems with storage and access, and problems caused by the changing information needs of consumers [SLW97]. The last issue becomes increasingly important in the context of Social Media as more and more different people are involved. DQ in the context of Social Media is quite different from traditional DQ in information systems evaluation [ERGB12], given the characteristics of SM, such as: wide accessibility, permanence, global audience, recentness and ease of use ([AgYi10], [Baez09]). Social Media has extended knowledge creation borders across organizational boundaries, therefore unlike traditional information systems, managers have no control to influence quality of the information obtained [KaRa12]. Hence many DQ frameworks for traditional information systems ([Redm96], [AlMa99], etc.) fail to address quality problems caused through many to many mappings. This mapping refers 1 The terms data quality and information quality are often used interchangeably by various authors ([SLW97], [ZhGa00], [KSW02]). 17

25 to information created in different contexts to support the requirements of different activities and perspectives at the same time [CPD09]. Moreover, DQ problems and especially their consequences can be pervasive [WLZ01], costly [EpHe04] and even disastrous [FiKi01], especially in the context of emergencies. EMS often have to make decisions in the heat of the moment'', on the basis of incomplete information and ambiguous intelligence about the unfolding events [JPH99]. Therefore as a first step it is necessary to be aware of existing and occurring DQ problems and their consequences to determine and implement support. 3.2 General DQ Problem Patterns To derive, develop and chose methods in order to improve DQ, occurring DQ problems must be understood. In order to structure existing DQ problems, they can be grouped from several perspectives (based on a literature review and experience): they can be categorized according to their origin (i.e., what causes the problem?) or according to their consequences for the information consumer [Eppl06]. These two different ways of categorizing the causes of DQ problems along with possible way of overcoming them are described in further detail below. DQ problems according to their origin (cp. [Eppl06]): Irrelevant Information as a result of selection, filtering, and profiling problems o Description: One possible cause for DQ problems can be that the information provided is not targeted at the community that is supposed to use it. In other words, a problem arises because information is addressed to the wrong audience. This can result in irrelevant, incomplete, or simply not useful information for the information consumers. o Solution: align information better to the needs of the user community Low quality/incorrect information as a result of source or authoring problems o Description: Another cause of DQ problems may simply be that the information producers create bad information that is incorrect, inconsistent, or outdated. So the cause of this problem is not that the information is relayed to the wrong users, but that the information in the first place is incorrect or unreliable. o Solution: improve the training for the information producers and enforce certain minimal quality standards Sub optimal content management processes as a result of workflow and management problems o Description: In this case, the problem occurs because information is provided in the wrong manner. The information may be correct and targeted to the needs of the information consumers, but it may not be provided in the right way or using the right process. o Solution: improve the coordination and the workflows in the content management processes Unreliable infrastructure as a result of hardware and software problems o Description: Another cause of DQ may be problems associated with the infrastructure used to produce or share information. If the hard and software on 18

26 which any information system is built is not reliable, the information may not be accessible or secure. o Solution: improve the hard and software in order to make it more reliable and efficient DQ problems according to their consequences for the information user [Eppl06]: Information overload: as a result of not being able to identify the right or most relevant information o Description: If an information consumer cannot identify (find and access) the relevant information, then this may be because it is difficult to get an overview on the available information because it is too dispersed, too vast, or because it is incomplete, inconvenient or impossible to access. o Solution: Sources need to be integrated or aggregated (for example, through portals, maps, directories, summaries, etc.) Misjudgment: Information cannot be judged or evaluated o Description: Once the information has been found, it has to be evaluated by the user. If an information consumer cannot evaluate, judge, or trust a piece of information this may be due to inaccurate or inconsistent statements, or due to an insecure platform on which the information was found. o Solution: To increase the trustworthiness of information its validity and background must be checked and clearly indicated Misinterpretation: Information is understood or interpreted incorrectly o Description: If an information consumer can evaluate the credibility of an information source, but he or she cannot interpret a piece of information itself, this could be the result of lacking clarity or simply because the information is not correct or does not refer to its sources. It may also be because the information is not kept current and not maintained and hence the user no longer knows what is still relevant and what has become obsolete. o Solution: To foster a better understanding of the information, a context must be provided that allows for an adequate interpretation of a piece of possibly isolated information Misuse: Information cannot be used or applied o Description: Finally, if the information consumer is not sure how to apply the information he or she has been provided with, then the resulting misuse could be the consequence of a wrong information format, a wrong timing or a wrong application context. o Solution: To avoid a misuse of the information or its non use, it must be reorganized for easy application, that is to say it must be provided in a (interactive) format that can be directly used by the information consumer These definitions can help to better frame, delineate and analyze DQ challenges in the further work of EmerGent. When the DQ problems are considered and taken into account, methods to assess and improve DQ can be better derived, developed and chosen. 19

27 3.3 Approaches for Resolving DQ Problems in Social Media Data Whereas some structured typical DQ problems can be resolved through data cleansing algorithms, data profiling programs, stabilization algorithms (e.g., phonetic manipulation and error correction), statistical process control, or dictionary matching routines (see e.g. [SLW97]), Social Media data used for emergency management in all phases must satisfy requirements for high standards of reliability, resiliency, security, and availability that guarantee access to sufficient capacity on a day to day and emergency basis [KKKS11]. Hence, DQ problems from Social Media can often not be solved through automated processes as they contain un or at least semi structured data. Social Media has extended knowledge creation borders across organizational boundaries; therefore, unlike traditional information systems, managers have very little or no control to influence the quality of the information obtained [KaRa12]. As mentioned above, DQ in the context of Social Media is quite different from traditional DQ, given the characteristics of Social Media, such as: wide accessibility, permanence, global audience, recentness and ease of use ([AgYi10], [Baez09]). DQ problems from Social Media require [Eppl06]: fundamental analysis of business issues, a change in work practices and process redesigns, an analysis of the involved information community and its expectations and skills, an evaluation of the relevant knowledge domains and their attributes, as well as a rating of the content management process and infrastructure. During a crisis or emergency, information can come from a variety of sources including computer systems, eyewitnesses at the scene, Social Media and subject matter experts [LWMR13]. Authorities need to quickly decide if this information is credible and if they should act on it. Less experienced decision makers may have a more difficult time determining if information is credible than more seasoned emergency management decision makers [Klei99]. Poor DQ can affect the course of the disaster situation and potentially create additional risks to lives and infrastructure. [LWMR13] With the increased use of Social Media, it is easy for rumors to spread quickly. Tweets can have additional information attached to them such as geographic information and pictures. Keywords or hashtags are used to organize messages. Tweets are not always particularly credible. An analysis of tweets over a 49 day period noted that only 17% of the postings contained credible information [GuKu12]. Retweets from credible sources can still lead to rumors since it is possible for these accounts to be hacked. Other problems with data from Social Media are e.g. the following [Comm13]: Limited understanding: Many EMS remain more comfortable with traditional media, and not all are aware of the potential advantages of Social Media as a tool for alerts and warnings. Some of this is surely a result of generational or cultural differences. Familiarity and comfort with Social Media for EMS can be expected to grow as training opportunities are provided and newly hired employees bring with them a greater familiarity with social media. Loss of control: When they use Social Media, EMS cannot control which information Social Media users share, which raises concerns that they might lose control of messaging or face civil liabilities if misinformation is shared. By contrast, in the traditional 20

28 command post style of information dissemination, long standing relationships between the press and authorities provide some sense of control over what information is disseminated as well as well understood opportunities to disseminate corrections as needed. It is the belief of some officials that with Social Media, misinformation may spread more rapidly and continues to spread even after a correction is issued. [KKKS11] emphasizes the need to ensure information quality within emergencies, in order to validate the collected information for decision making. With the abundance of information on Social Media, this becomes central to achieving an effective and efficient outcome in emergency response. With the empowerment of the general public and the abundance of information on Social Media, fostering DQ is central for decision makers to achieve an effective and efficient outcome in the emergency response [Jens12]. Social Media clearly has strengths and presents opportunities in emergency management, but there are also (new and/or increased) risks and challenges. [HCH10] mention the following six challenges of using Social Media in emergencies: 1. Access to internet either because of geographical location or because of individuals low information technology literacy 2. Disasters can destroy the required technological infrastructure and interrupt services 3. In time of crisis and chaos, accuracy can be compromised 4. The scalability of the social networking sites to handle the workload (If not capable of handling workload, negative impact on efficiency and information sharing) 5. Privacy concerns 6. Social media will never be the only solution to disaster preparedness and relief 3.4 Challenges in Structuring and Querying Information To strengthen the tie between citizens and authorities via Social Media, someone has to deal at least with data quality issues regarding information overload (See section 3.2). While trying to use Social Media completely unfiltered, for instance without tool support, authorities have to deal with information streams that have high volume of data with low information content. E.g. in 2011 Twitter had more than 100 million active users, posting 230 million tweets a day 1. One year later, the number of active users had doubled 2. This also means, that during an emergency the general usage of Social Media is getting more and more important. Hence, to handle the vast amount of data, information from Social Media streams has to be transformed into a low volume but a rich information model. In other words, EMS and citizens must be enabled to find the right information on the different channels in time as well as to provide relevant information to the right channels in time. But querying the information channels directly does not scale anymore for, besides the huge amount of data received every second, the complexity of the queries needed to filter the waste out in order to get the useful information in time is very high [Piro09]. 1 active users (assessed June 2, 2014) 2 (assessed June 2, 2014) 21

29 The world looks at a disaster, often willing to help, but needing an accurate picture. [ ] Humanity s effectiveness will be much increased if relevant data streams are made available [W3C07]. Therefore information from Social Media streams has to be structured in a useful way as a prerequisite. The most promising approach is to structure and enrich information semantically, because it simplifies the way to create context aware queries. [BoRo12] summarized it as follows: Semantic technologies have the potential to help people cope better with social mediainduced information overload. Automatic semantic based methods that adapt to individual s information seeking goals and summarize briefly the relevant social media, could ultimately support information interpretation and decision making over largescale, dynamic media streams. On the technological level, several technologies to model or query data semantically are already there. Query languages like SPARQL 1 can be used to query semantic data stores. Data can be modelled on standards like RDF 2 or OWL 3. Also there are frameworks or solutions available, providing powerful forward and backward reasons to infer logical consequences from data like Prova 4, Jena 5 or CWM 6. Another challenge is to look on the data, derive patterns to identify useful information and contrast them to workflows of the authorities in the four phases of the EMC. Extracting useful data from Social Media as well as spreading useful information to Social Media is the most important goal. 1 sparql query/ (assessed June 10, 2014)

30 4 Review of Related Research Projects Social media offers an opportunity to communicate the crisis situation to other citizens or to authorities even though mobile phone or emergency services may be encumbered. During the past few years, various studies were performed focusing on various aspects of Social Media in emergency management underlining its unceasingly growing importance in this area. Several projects are developing and implementing systems, tools and algorithms performing Social Media analysis. This section recaps prominent EU Framework (section 4.1) and other project works (section 4.2) using Social Media analysis in the context of emergency management. The projects are listed in alphabetical order according to their acronym. 4.1 EU Framework Projects Project Name Project Objectives Social Media /Data Source Alert4All 1 (FP7 Security, ) COSMIC 2 (FP7 Security, ) CrisComScore 3 (FP7 Security, ) CRISMA 4 (FP7 Security, ) Alert4All aimed at improving the effectiveness of alert and communication towards the population in crises management. COSMIC is identifying the most effective ways in which new technologies and applications are being used by citizens and governments. The project developed an audit instrument as a tool for ensuring effective crisis communication strategies and implementation. The CRISMA project will develop a simulation based decision support Twitter, blogs YouTube, Twitter, Facebook News media Data from sensors Approaches Classification, Support Vector Machine classifier Classification Text processing Priorisation, Optimisation of response, counter Visualization components Map List, recommenda tions and best practices Text messaging GIS based visualization, Real time environment Filtering mechanism Keyword, tags Topic, information Topic, information Information Potential relevance to EmerGent Usage of results, especially on how citizens trust information from authorities through different communication channels. Mapping the use of current technologies in crises and also mapping the use of emerging applications. Usage of findings on the potential roles and ethics for citizen participation in emergency response. The auditing instruments for effectiveness will be one possible measurement method in the analysis phase. The auditing of decisions in crisis management will be one possible measurement method in project.eu

31 Project Name Project Objectives Social Media /Data Source ESS 1 (FP7 Security, ) IDIRA 2 (FP7 Security, ) INDECT 3 (FP7 Security, ) isar+ 4 (FP7 Security, ) REACT 5 (FP6 Sustdev, ) system, for modelling crisis management, improved action and preparedness. The ESS project developed a common information management and communication platform for supporting the management and coordination of emergency operations. IDIRA project is focusing on the interoperability of data and emergency procedures in response to large scale disasters. The project is developing threat detection tools and generation of data mining and information retrieval applications. isar+ project delivered the guidelines that enable citizens using new online and mobile technologies to actively participate in response efforts, through the provision, sharing and retrieval of emergency information. REACT has enhanced data by providing associated services that are able to semantically analyse and cluster Real time sensor data (thermal, video etc.) Geo and attribute data, integratio n of sensor data Weblogs, chats, news reports SMS, Twitter, Videos Data from callers and automati Approaches measures and preparedness Spatial localization, Data fusion Text classification, map, lists Relationship mining, machine learning methods for behavioural profiling Multivariete testing, Design of experiments Semantic analysis Visualization components al data visualization Filtering mechanism Potential relevance to EmerGent the analysis phase. Map, lists Time, query Analysis of state of theart technologies for crisis discovery and management and application of existing data fusion methods for developing a data fusion and mediation system. Georeferenced Visualisation map Event model Topic, time, information Keyword, query Tags, time, location GIS based interface Location, keyword, time, semantic Methods and technical interoperability standards developed here will influence the integration aspects. Consideration of methodologies and algorithms for data & event processing. Its approach to the dynamics between citizens and authorities in crises, allowing the leverage of authorities levels of shared awareness and performance, benefiting from citizens published information. Usage of the OASIS CAP Protocol to allow interoperability between Emergency Services together with the TCP segmentation offload 1 project.eu/ project.eu/ ist.net 24

32 Project Name Project Objectives Social Media /Data Source Approaches Visualization components Filtering mechanism Potential relevance to EmerGent environmental and crisis management information. c systems protocol for a common data ontology. SocialSensor 1 (FP7 ICT Networked Media, ) SocialSensor is developing a framework for enabling real time multimedia indexing and search across multiple Social Media sources. Facebook, Youtube, Flickr, Tumblr, Google+, Instagram Clustering of geodata and visual descriptors Map, timeline, lists Sorting, zooming Scalable mining and indexing approaches that taking into account the content and social context of social networks. WeKnowIt 2 (FP7 ICT, ) The system built in this project was based on interviews performed considering emergency management practitioners. Pictures, Videos, Text Text processing, Clustering Map, timeline Tag, time The results of the layer of social interaction and the massive user feedback layer will be considered as one of the inputs for metrics. Table 4: The referenced EU Framework projects based on common aspects of social media analysis 4.2 Other Projects Project Name Project Objectives Social Media /Data Source Crisees 3 (Developed at the University of Glasgow, UK, ) Disaster (EC CIPS Action, ) Emergency Situation Crisees has developed a monitoring tool for Social Media streams. Disaster 2.0 (D2.0) project has explored how EU governments can potentially use Web 2.0 applications and Semantic Technologies in disaster response. The emergency situation awareness Youtube, Twitter Twitter, Facebook, Ushahidi Twitter Approaches Extension to Sentiment Semantic technologies Aggregation, Text Visualization components Filtering mechanism Potential relevance to EmerGent Map, lists Time, query Extracting information from Social Media. Filtering information related to event. Visualising information on maps. Map, list Map, tag cloud, Tag, location, keyword Slices time, Use of results in which public utilise web 2.0 and web 3.0 technologies during disasters. The research results on text classification to

33 Project Name Project Objectives Social Media /Data Source Awareness Platform 1 (CSIRO, Australia, ) SensePlace2 2 (U.S. Department of Homeland Security, ) TEDAS 3 (Developed at University of Illinois at Urbana Champaign, ) Tweak the Tweet 4 5 (University of Colorado, ) Twitcident 6 (CrowdSense Dutch national applied research institute and technical university Delft, ) TwitterBeat 7 (Cyber Infrastructure & platform analysed tweets. SensePlace2 developed a system for filtering Twitter messages. TEDAS project developed an event detection system for Twitter. Tweak the tweet was a crowdsourcing platform. The key difference to other platforms is that this system 'works with the existing Social Media infrastructure'. Twitcident project developed a toolbox for filtering and analyzing information from Twitter streams during crisis situations such as fires, storms or other types of incidents. The TwitterBeat analyzes huge amounts of textual data uncovering the Twitter Twitter Twitter Twitter Twitter Approaches clasification, Keywords Named entity recognition Classification & rules, spatial and temporal analysis Trust grammar, parsing Classification, rules Sentiment on topic and location Visualization components Filtering mechanism Potential relevance to EmerGent timeline traffic, tag identify the impact of the incidents identified. Map, timeline Map, timeline Map, timeline Map Time, tag cloud Location, keyword, time Predefined hashtags Keyword, facets Use of different search and filtering facilities to browse through a huge amount of tweets by considering the extracted information. Results to detect and analyze events by exploring rich information from Social Media. Parsing algorithm applied to extract information given in combination with hashtags. This extracted information useful to perform keyword based filtering. Results related to facetsearch interfaces, i.e. on facets extracted in the previous steps, which helps the user to browse through the data to gain a better overview. Map Zooming Approaches on sentiment analysis can be used to identify the mood after a disaster, for 1 and Services/emergency situation awareness.aspx TEDAS

34 Project Name Project Objectives Social Media /Data Source Approaches Visualization components Filtering mechanism Potential relevance to EmerGent Geospatial Information Laboratory, University of Illinois at Urbana Champaign, ) sentiment. e.g., for crime prevention. Twitris 1 (Kno.e.sis Center, Wright State University, Ohio, ) Twitris provides situational awareness by monitoring an event on Twitter at both micro and macrolevels. Twitter, SMS Event discriptors Map, Tag clouds Spatial temporal theme Results on context based semantic integration of multiple Web resources and expose semantically enriched social data to the public domain. Table 5: The referenced other projects based on common aspects of social media analysis 4.3 Summary The projects listed above show that there are already systems available: for diverse applications (i.e., focusing on real time streaming and offline analysis); for using one or several Social Media platforms for monitoring; that are specifically developed for emergency management; for executing different types of analysis such as monitoring, event detection and sentiment analysis. The main advances in the field of enhancing the use of Social Media in emergencies will be: to enhance analysis and examination of the applicability of Social Media as a novel channel for alerting citizens. to enrich filtering mechanisms in the user interface. to improve the approach on how the data streams can efficiently be analyzed in emergency situation. to enhance methods and tools concerning Information Mining and Information Quality. to improve semantics driven analysis of a wide variety of Social Media content. to provide technical support for collaboration in large scale emergency relief efforts, by safeguarding interoperability, coordination and support among authorities and citizens

35 to enhance use of Social Media for sustainable collaboration between authorities and citizens, and further utilize Social Media to elevate the safety and security of citizens in emergency situations. to use measurement methods not only for communication from authorities to citizens but also the other way. to improve threat detection approaches and enhancement of privacy related issues. to utilize Social Media ubiquity to search, locate and effectively communicate between citizens and authorities. to boost an increasing information flow between authorities and Social Media as a whole, not only ony interoperability level. to achieve improved impact metrics to better disseminate results especially for emergency relief. to develop user interface for EmerGent apps supporting emergency response, taking the requirements for flexibility into account through applying an efficient approach. to develop online clustering approaches that will help increase situation awareness during emergencies and will provide near real time briefing material for decision makers. 28

36 5 Interview Study: Social Media in Emergencies 5.1 Methodology of Data Collection In order to collect information about the different types of authorities users and their motivation and attitudes in the use or potential use of Social Media in emergencies, a number of interviews with authorities professionals were conducted. As D3.2 also required direct stakeholder involvement for data collection, the interviews conducted in the prospects of Emer Gent focused on collecting information that related both to D3.1 and D3.2; this deliverable documents the results relating to D3.1 only. Here, it should be stressed that the following section is a kind of technical report, where the most important outcomes will be highlighted and summarized. These first results are the prerequisite and starting point for the development of the EmerGent concept, a basis for the next steps within the WP 3, i.e. Task 3.2: Analysis of processes and interoperable tools within EMS Task 3.3: Analysis of interfaces to mobile devices and major social network providers Task 3.4: Requirements gathering and definition Task 3.5: Analysis of potentials of citizens involvement in the EMC Task 3.6: Analysis of potentials using social media for EMS in the EMC Participants The selection criteria focused on collecting responses from a wide range of authorities and EMS within the available time frame. Interview participants were selected from across the networks of consortium partners and from a variety of Public Safety Answering Points ("PSAPs") and emergency dispatch centers. Here, the aim was to cover all types of authorities relevant for the project, i.e. Fire, Emergency Medical, Governmental authorities and Police. While the interviews collected responses from authorities across different countries, the selection criteria were not explicitly based on geographical location. Similarly, use of social media was not a prerequisite for participating in study, but on the contrary responses from both authorities using and not using social media were welcome. The interview covered different topics, structured in seven sections: A Introduction to the organization B Information about the PSAP and/or emergency dispatching center C Technical information about the PSAP and/or emergency dispatching center D The use of Social Media in emergencies E Challenges and benefits of the use of Social Media in emergencies F Future plans Due to the nature and wide range of the topics covered, for example operational and technical topics, in some cases more than one person from the same organization contributed to the interview. 29

37 5.1.2 Methodology and Details of Question Asked At the beginning of the task, it was agreed to focus on qualitative rather than quantitative data, and in this respect, the interviews followed a semi structured format. The freedom provided by the nature of this format, against a format strictly adhering to an interview transcript, is deemed to provide the opportunity for more direct communication between the interviewer and the interviewee. This approach allows for a more in depth exploration of the issues being covered. This and so provides a more rich set of results due to ideas and topics brought up during the interview. Additionally, this format allows the interviewer to ask questions in ways more appropriate to each interviewee based on his or her background. All topics were decided on well in advance and agreed by all parties involved in the interviews or tasks that depend on them. A guidance document was prepared to ensure that interviewers explored the same topics in a similar way. The guidance document contains a framework of questions outlined below, with free answers or in some cases multiple choice answers or rating scale answers, under each of the seven sections outlined in Annex A Collection of Results Personnel of the project partners conducted interviews during May 2014, in face to face or telephone sessions. The results were usually collected offline and were later added by each interviewer to an online tool for consistency and further analysis. The results are provided in the following sections anonymously and without an indication of the country of the interviewee. 5.2 Results This section provides a detailed analysis of the information collected during the interviews and describes the current and potential future use of Social Media in emergency management, by taking into account participant perceptions and experiences so far. The results are divided into nine sections. Overviews of the collected responses are presented in tables. Conclusions and important highlights of the analysis of results are provided within each section. The first few sections are focused on presenting the profile and operational characteristics of the respondents and their organizations, while the remaining sections focus on Social Media use. 30

38 5.2.1 Organization Profile (Questions: A1 A4, A10) Table 6 provides an overview of the participants organizational profile in terms of the PSAP model they follow and their operational level. Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 PSAP Yes Yes Yes Yes No Yes No No Yes Yes Yes PSAP Model n/a 2 n/a n/a Operation level loc loc reg loc nat loc, reg loc, nat loc = local; reg = regional, nat = national loc, reg, nat loc loc loc Table 6: Organization Profile (Source: May 2014 Survey of authorities) In total, 11 interviews were conducted with participants from 7 European countries and 8 different cities, both represented within the consortium and outside it. The exact locations of the interviews are not provided in this deliverable, in order to ensure the anonymity of the respondents. A wide range of authorities, such as Police, Fire, Emergency Medical and governmental authorities participated in the study, in addition to respondents who were responsible for directly receiving and answering emergency calls (PSAPs). The conducted interviews involve organizations that operate on all administrative levels at local, regional and national level, while most of them operate locally Type and Size of Organization The interviews collected information from a variety of organization types, extending to most services involved in emergencies, such as public safety answering points, dispatching centers, police, fire brigades and medical / ambulance services. The case of service providers not directly responsible to handle emergencies, such as a Police Information Service, was also studied as such types of providers are directly related to emergencies, collaborate during emergencies with the emergency services and also use Social Media in their daily activities PSAP Models A list of five different models described by EENA [WWW01], was used as the common basis for reporting and comparing the conducted interviews. While this list is not exhaustive and does not cover the entire call handling model or all possible models available worldwide, it was found sufficient for this study, because it helped quickly and easily refer to the major characteristics of operation amongst the respondents. The majority of PSAP model of operation was represented in the study. The participants from PSAPs followed four models: Model 1: Emergency Response Organizations handling emergency calls Model 2: Filtering stage 1 PSAP and resource dispatching stage 2 PSAPs 31

39 Model 4: Data gathering by stage 1 PSAP, resource dispatching by stage 2 in an integrated control room. In model 4 all involved EMS use the same infra structure (hardware / software), in contrast to model 2, where EMS use the different infra structure and interoperability can be a major issue. Model 5: Emergency Response Organizations independent PSAP Each model is depicted in the following figures. Figure 3: Model 1: Emergency Response Organizations handling emergency calls (Source: EENA) Figure 4: Model 2: Filtering stage 1 PSAP and resource dispatching stage 2 PSAPs (Source: EENA) Figure 5: Model 4: Data gathering by stage 1 PSAP, resource dispatching by stage 2 in an integrated control room (Source: EENA) 32

40 Figure 6: Model 5: Emergency Response Organizations independent PSAP (Source: EENA) Participant Position and Functionalities Participants in the interviews came from different professional backgrounds, including operations, communications, technical etc. Since in most cases, more than one participant contributed to each case, 19 professionals with senior positions participated in this study, including: Senior specialist in crisis communications Head of the Communication Centre Head of the Centre Commander PSAP leader Technical team leader Head of training department Their responsibilities in their organization included dispatch planning, field support, logistics, communication management, team coordination, crisis management, technical administration, exercise preparation etc. This data is provided to help understand the perspectives on the use of Social Media in relation to the operational / organizational responsibilities. It was important to validate, that all participants, regardless of their position and background, acknowledged the importance of Social Media in emergencies, considered it an interesting topic that they like to follow and agreed that it cannot be overlooked despite the difficulties currently encountered Risks, Vulnerabilities and Operational Size (Questions: A5 A8) The following table aims to highlight among others the likelihood and intensity of bi lateral communication needs, i.e. from citizens to authorities and vice versa that results out of the characteristics of population, given risks, hazards and vulnerability as well as the current calls / missions. 33

41 Case 1 Case 2 Case 3 Case 4 and 5 Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 Region Population 1) > > Amount of Commuters 1) medium 200 n/a high high high high high medium Risks, Hazards and Vulnerability High population density X X X X X X Large industrial sites X X X X X X X X Harbors X X X X X X Railways X X X X X X X X X X Tunnels X X X Major motorways X X X X X X X X Earthquakes X X X X Storms / bad weather conditions X X X X X Water catchment area X X n/a n/a Flooding X X X X X X X Cultural heritage, old cities X X X X n/a n/a X X X Touristic attraction X X X X X X X Cultural Events X X X Calls / Missions Fire n/a n/a Urgent and non urgent medical ² ) 7.000³ ) n/a n/a n/a / n/a n/a Technical response n/a n/a 828 n/a n/a n/a Burglary, Sexual assault, homicide n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a 1) in thousands; ² ) ambulance; 3) 3 out of 17 ambulances, total for city of case 2 is Table 7: Characteristics of the analyzed interviews (Source: own elaboration) 34

42 5.2.3 Operational Tasks (Questions: B1 B3) All participants share the task of responding to the public, or have a direct role in responding to emergency incidents. Case 5 is an exception to this as it holds a higher level role in emergency management. With the exception of case 7, most of the participants are involved in notifying the public and providing information. Similarly, case 10 is the only organization in the study, not directly responsible for receiving emergency calls. Receiving urgent calls from 112 Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 Yes Yes Yes No No Yes Yes Yes n/a No Yes Receiving direct urgent calls Yes Yes Yes Yes No Yes Yes Yes Yes No Yes Notification Yes Yes Yes Yes No Yes No Yes Yes Yes Yes Providing information Yes Yes Yes No No Yes No Yes Yes Yes Yes Table 8: Operational Tasks (Source: own elaboration) In some cases, additional roles and responsibilities were reported such as: Case 8: Direct contact with the army, weather monitoring, direct contact with all municipal services (divers, gas service, electricians, water service, transport service, construction service, hunters, mountain rescue, etc.), logistic and IT support, reports for the Major Case 11: Responsible for managing the fleet of police cars Case 1: Receiving non urgent medical calls via (non emergency related ambulance service); In case of alarm phases: alerting all necessary services, means and persons alerting the authorities having jurisdiction alerting the hospitals concerned; Act as an counterpart for the city government in case of information procurement; Providing information for the district government by dealing with e.g. major case incidents; Collecting information about the available number of beds in intensive care units in the local hospitals to assure a fast allocation of casualties Case 5: Due to the higher level role of Case 5 in the emergency management ecosystem, their responsibilities differ from the previous participants and include: Gathering information for coordination of actions on national level in all crisis management phases; Responsible for civil planning, critical infrastructure protection, threats analysis and monitoring; Coordination of trainings and exercises in a scope of crisis management Table 9 identifies how each participant is involved in the four stages of the emergency management cycle. 35

43 Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 Prevention n/a No Yes n/a n/a Yes No No Yes Yes Yes Preparedeness n/a No Yes n/a n/a Yes Yes No Yes Yes Yes Response n/a Yes Yes n/a n/a Yes No Yes Yes No Yes Recovery n/a No n/a n/a n/a No No No Yes No Yes Table 9: Participant involvement in four stages of an emergency (Source: own elaboration) Communication Channels (Questions: B4 B8) Communication channels are presented according to the purpose: 1. Receiving input messages from citizens 2. Sending messages to citizens 3. Broadcasting warning messages Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 Telephone calls X X X X n/a X X X X X X SMS n/a X X X Paging n/a n/a X Webform 1) X n/a X X Radio Traffic X n/a X X X Social Media n/a X X 1) website response Table 10: Communication Channels for inbound messages (Source: own elaboration) Other communication channels reported: Case 1 and 3: fax (deaf and persons hard of hearing) Case 2 is planning to implement SMS and channels in the near future, as it is important to widen the ways to reach the 112 PSAP. Case 8: WAP, Video Call The communication channel most frequently reported and used as a source of input messages from citizens is the telephone. Another frequently reported channel is SMS, which in some cases is also used by people with hearing disabilities. Social media is usually not reported as an inbound message source, with the exception of two cases. In the first case, it is reported that Social Media is so tightly integrated in our daily life and communication channels, that is too important to ignore nowadays and it would be unthinkable not to respond to it. In the second case, although social media is not officially acknowledged as an inbound message source, messages are received, read and certainly taken into account. In both these cases and in the 36

44 context of this study, social media is used as a communication channel for inbound information in practice, while officially it is not acknowledged or promoted as an information channel. Telephone calls Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 X X X X n/a X X X X X X SMS X n/a X X X Paging X n/a X X n/a X X Webform 1) n/a X Radio Traffic X n/a X X X Social Media n/a X X X 1) website response Case 7 Case 8 Case 9 Case 10 Case 11 Table 11: Communication Channels for outbound messages (Source: own elaboration) The purpose of sending messages to citizens follows the same patterns as with receiving messages and finds the telephone as the most commonly used channel. All other channels are similarly used as for inbound messages. Case 1 Case 2 Case 3 Case 4 Radio X X X X X X X X Local TV X X X X SMS Broadcast Social Media X X X X Sirens X X X Special warning 1) Offline media & press Case 5 Case 6 X X X X X Online media X X X X X 1) systems like KatWarn or AmberAlerts Case 7 Case 8 Case 9 Case 10 Case 11 Table 12: Communication Channels for broadcasting warning messages (Source: own elaboration) Authorities appear less reluctant in using Social Media for broadcasting messages. Most participants use the radio for broadcasts, although it was commented that there has been a shift from the press and the radio to the online channels. The use of their websites, other online media websites and Social Media provide the opportunity to inform the public very fast and allow the message to spread rapidly. In some cases, Social Media, such as Twitter, were previ 37

45 ously used for communicating with the media, but have proved to present an excellent opportunity for providing public information Current Use of Social Media (Questions: D1 D14) The following table gives an overview about the state of the art if and what kind of Social Media is present at the institutions / authorities that were part of the first survey. Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 None X X X X Twitter X X X X X X X Facebook X X X X YouTube X X Flickr LinkedIn X X Tweedeck X* X Hootsuite X* *individual key players Table 13: Current use of Social Media (Source: own elaboration) Here, it is visible that most of the analyzed case studies use Social Media (except case 4 and 5) for different purposes that will be described below Potential Use of Social Media (Questions: D15 D23) The analysis of the first stage interviews highlighted that, in general, Social Media is seen as an important tool for the future, among others for gathering of information, interpretation of incidents and communication to / with the public. It is seen as one of the fastest ways to disseminate information about prevention or to inform the public with warnings or notifications before, during and after a major incident. But not only this: Social Media supports a bidirectional communication and can support all stages of an EMC: starting with identification of potential hot spots and informing or warning the public about certain facts regarding an incident (prevention or mitigation), informing the public about certain warnings or directions regarding an incident like spread of hazardous substances (preparedness), informing the people about forthcoming evacuation out of exposed areas, and reduce or avoid the possibility of emergency escalations (response), inform the people about medical aid and further behavior (recovery). The intensity and degree of information exchange concerning relevant phases of the EMC making use of Social Media is diverse. Case 1 for example distinguishes between the foci: operation and general. Concerning the operation, the focus is set up on the receipt of relevant information for operation. This information is used for e.g. necessary intervention concerning the process of communication (e.g. information provided by e.g. media) and thereby control and regulation of information (e.g. instructions concerning behavior, revision and correction 38

46 of wrong information). In general it concerns public relations, information about training positions and advertisements. Even if the use of Social Media is seen as important there are some restrictions concerning especially the exchange (gathering and mining) of data, not only due to the limited resources but also limited tools and authorization. Case 10 differentiates the responsibility according to the dimension: in special cases and big emergencies the responsibility lies with the police communications department. In other simpler or pure information cases everyone in the dispatching center is responsible for SM use. Due to missing technical possibilities and tools ( internet lock out ) in the case of for example case 1 only the staff of the press office (fire department) are able and allowed to gather data (here data mining / gathering is done only if at all when this is connected to a clear operation). A similar situation is seen in case 2, where the communication service local (government) and communication service (fire service) is only allowed to spread data (responsibility belongs to the experts in the service) or case 6 the dispatching center (responsible there: dispatch officer, IT staff) or as in case 11 the officer in command. Nevertheless, some of the analyzed cases already use (more or less) Social Media tools. Case 2 uses the Twitter platform, Hootsuite [WWW02], Storify [WWW03], Social Mention [WWW04] and aims to use Coosto [WWW05] in future (see below). Here, the observation of current communication processes is used. Some other examples are crowd sourced maps including OpenStreetMap, Google Crisis Map or crowd sourced information management, crowd sourced pictures, information about smoke spread is used by e.g. case 2. Case 10 was said to use several tools (Hootsuite, Tweetdesk [WWW06] and Coosto) when preparing for a big event but not in urgent emergencies. During an emergency one or two people are gathering the data. Case 11 did not use any specific tools or applications. In one case, the use of a specialized Twitter service, Twitter Alert [WWW07] was reported. Twitter Alert was launched in late 2013 and aims to enable authorities to keep people well informed by providing accurate and important information during an emergency. It is open to be used by authorities, government services and NGOs worldwide. At the time of writing this deliverable, only organizations from Australia, Brazil, Ireland, Korea, Japan, Spain, the United Kingdom and the United States of America have joined the service. When a Twitter user signs to receive an authority s Twitter Alert, the user receives a direct notification whenever a tweet is marked as an emergency alert. Notifications appear on the web interface of Twitter and they are styled differently from regular Tweets to indicate they relate to an alert from one of the signed authorities. Notifications can also be delivered via SMS and in the case of Twitter applications in the various mobile operating systems such ios or Android, the related applications will produce push notifications. Beside the limited human and technical resources, the financing of using new Social Media is quite problematic and constrains its use. In some cases (case 1, 4, 5) there is no financing for Social Media. Nevertheless, in other cases (2, 3, 6, 10, 11) a budget is available or could be provided (local, regional, own budget) to guarantee this task, even if it is not part of the regular budgets and not marked as a use for Social Media. The strategy for implementation into the organization differs through the case studies. Some case studies have no clear plans (like case 1); other organizations (case 3) are experimenting with the use of Social Media in case of an emergency. There have been experiments devel 39

47 oped throughout the use of Coosto but these experiments were sporadic and without reciprocal reconciliation. Others (case 2) have clear expectations and define the first steps how to use and implement SM. Case 2 is running experiments with the active role of citizens during emergencies: they use Social Media on a daily basis to monitor and to gather information about incidents. Up to now this is performed manually, but in the near future it is aimed to implement software to automate this process and to implement Social Media in the dispatching rooms of the fire service and police (in the police, Social Media monitoring through specific software is already used to improve intelligence). In case 6, fire officers and dispatchers are using Social Media without a clear and determined goal. This way it is present in their daily routines, however none of the staff have received any training on how to use it so far. They are planning to provide warning messages via Twitter and Facebook in incidents where more than two fire brigade units are present at the spot or during incidents with hazardous materials. Nevertheless and as mentioned before, there might be some problems with integrating it with the existing software. As any new software has to be checked and verified by the PSAP director, this might take a while (in future, there might be a chance to integrate it into the existing software). While there are no formal or documented implementation strategy or guidelines in case 10, search for information and monitoring of Social Media is done continuously and in parallel with all normal activities. This is similar to the situation in case 11, where the information is used (one person in the PSAP is responsible for posting messages, all people are responsible for reading messages), but no formal workflow, method, or guidelines exist and no special tools are used. Case 2 aims at a distinction between expert Social Media users and basic users: Basic users are for example PSAP operators who just have to get information from Social Media when it's relevant. For these users, case 2 is focusing on buying software that will allow it to build a dashboard to show the right information when it is available. Expert users are the ones who control the dashboard and choose the search protocol. They are also the users that provide crisis communication on Social Media, as there is a requirement to be very familiar with Social Media language to be able to communicate in the most appropriate way. At this moment, implementation is still a problem as a lot of the necessary knowledge is only held by experts in the organization Research on Social Media (Questions: D24 D27, F5) The importance and position of new Social Media will increase in future. This is stated in the cases analyzed, even if most interviewees are not active in the field of research (e.g. cases 6, 7, 8 and 11). Therefore a further use is in principle welcome and recommended. Nevertheless due to missing technical and time resources an implementation in case 1 is currently not possible. Case 2 plans to use Social Media for e.g. gathering information, communication to the public, cooperation between services (example Google Hangouts to communicate between the gold and silver command). There are concerns to follow research and development on Social Media as stated by case 2 in the field of monitoring tools (e.g. implementation of Coosto), continuous evaluation of Social Media use (what works and what doesn t), strengthen the cooperation with universities and the team D5, a newly formed team of public communication specialists who help each other during crisis ( team D5 have a specific section on Social Media monitoring). 40

48 Case 11 states that possibilities of Social Media are currently being considered, even there are no formal plans for using it. Here, the verification of information is very important as well as the possibility of an account being hacked is considered a serious risk, especially if at the same time there is a serious emergency Challenges and Benefits of Using Social Media (Questions: E1 E3) The following table shows an overview about the benefits of Social Media use not only for the public but also for authorities. Better public warning Better reachability of citizens Improvement of the situation awareness for authorities Improvement of the situation awareness for citizens Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 yes yes n/a yes n/a yes yes yes yes yes yes yes yes n/a yes n/a yes yes yes yes yes yes yes yes n/a yes n/a yes yes yes n/a n/a yes yes yes n/a yes n/a yes yes yes n/a n/a yes Table 14: Benefits of Social Media use (Source: own elaboration) In times of an incident, inhabitants want to be informed by the authorities in the same way as they daily gather their information. That is fast and dependent of the involvement and curiosity of the inhabitants. Case 11 states that Social Media has taken away some of the distance between the public and the Police, resulting in the police coming closer to citizens. Case 10 highlighted that people know that they get valid information from emergency services and that this information can be trusted. Here, trust is seen as a prerequisite for an appropriate relationship between local / regional authorities and the society. Trust has a key role in dealing with given risks and communication and should be regarded as fundamental for risk interpretation and awareness of the public between real and perceived risks (interpretations of risk differ according to individual and social contexts). It has to be kept in mind that public decision making that is based only on the factual scientific dimension of risk leads to distrust, not taking into account the socio cultural dimension, which includes how a particular risk is viewed when values and emotions are concerned (e.g. whether a risk is judged acceptable, tolerable or intolerable by society is partly influenced by the way it is perceived to intrude upon the value system of society). In addition it contributes to the vulnerability of institutional settings as well as affected individuals. Case 3 stated that when the government takes a position in the Social Media networks they also will be able to deliver an interpretative perspective and handling perspective to the inhabitants in the area of the incident. This causes effects on the resiliency of a community, because only those who are well informed and integrated in 41

49 the process will accept the decisions made by different authorities and undertake the right choices/decisions in cases of risks. Furthermore it proves the image of the organization. Case 3 highlighted that the fire service has a fine (positive) image towards the inhabitants compared to different other services and the government, where the image is significantly worse. The government will have to be a reliable partner concerning information exchange in the daily life, before they will be considered reliable at times of an incident. By quickly providing the correct information (concerns quality and quantity), the inhabitants will be provided with information like they want. In this way the government prevents a panic during a big incident or a critical event. In parallel to the benefits, the interviews highlight additional challenges (even if some of the case studies have no experience with it, the difficulties could be estimated, as seen in the table below). Reliability of the different sources Liability of the different sources Believability / Verifiability Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 high medium n/a high n/a high high high n/a high high medium medium n/a high n/a high high high n/a high high medium medium n/a high n/a high high high n/a high high Objectivity high medium n/a medium n/a medium high high n/a high high Timeliness low low n/a no n/a medium high high n/a high medium Reproducibility low low n/a medium n/a medium high high n/a high n/a Understandability Legal uncertainty low low n/a medium n/a low high high n/a high n/a low low n/a high n/a high high high n/a low n/a Case 7 Case 8 Case 9 Case 10 Case 11 Table 15: Difficulties of using Social Media (Source: own elaboration) Additionally to this case 3 stressed that some identified difficulties are among others the speed with which information is released, it s never able to quickly discover in which communities the most critical information is shared; you don t know who needs the information the most and messages cannot be quickly or sufficiently be checked on the truth. In fighting a crisis there is a crucial process of validation of the facts to prevent the ex post evaluation and the wrong conclusions. This is a complicated process at this fine moment. Some case studies and examples concerning dealing with Social Media can be found intelligence gfsa 42

50 5.2.9 Future Plans for Using Social Media in Emergencies (Questions: F1 F4) Those interviewees that replied to the question concerning future plans for using Social Media in emergencies estimate the impact as increase (cases 1, 2, 4, 5, 6, 7, 8). Furthermore, case 2 stated that in 5 10 years crowdsourcing and big data will be standard technologies in crisis management. Therefore it is necessary to focus on these items. In a crowd sourced and big data world, verification becomes the main skill of the PSAP operator and the information manager. Nevertheless case 1 voiced the concern that as long as the technical and personnel circumstances will stay as they are there will be no significant change. In cases 4 and 5 there are no examples of using Social Media in emergences by public administration. Probably some will observed as important issue only when this gap will be filled by citizens. For sure Social Media will be used mostly in field of situation monitoring and public warning. Case 6 dispatching center will promote the implementation of warnings, notifications etc. (see before). The attention should be moved also to gathering information from the public. Case 8 has no clear concept but will place the Social Media growth and usage on the own agenda shortly. Case 11 will consider the use of Instagram and Facebook, but there are no precise plans at the moment ( we know what Social Media is only for a few years ). Concerning new technologies, case 2 is focused on Coosto and case 6 aims at providing information to the general public using one app. This app should post a warning message to everybody on that area, regardless of the follower status. Case 6 and case 1 highlighted that Social Media will be present in the daily routines in dispatch centers, PSAPS, etc. and it is possible that the reporting of the fire departments will focus more toward Social Media and the older solutions (homepage etc.) will lose importance. All interviewed institutions / units agree in opinion that research projects like EmerGent are helpful to gain new insights. The units (as in case 3) are aware about the development around Social Media and know that they will have to work with it. Nevertheless, for government agencies this is a new phenomenon. At this moment, there are only experiments, for example Coosto, put together to look how and what you re able to do with Social Media during a crisis. 43

51 6 Summary 6.1 Social Media, Emergency Use and Role Patterns Social media is and was widely used during a variety of emergencies. After presenting a definition for social media (section 2.1) an overview of cases of social media use in literature (section 2.2) as well media patterns (section 2.3) are presented. Based on this role patterns to enable communication from citizen to citizen (C2C), in between citizen and authorities (A2C, C2A) and to support inter organizational collaboration (A2A) are derived. It is illustrated that affected citizen and volunteers often act autonomously with the help of Social Media resulting in potentials and challenges concerning the self organization of communities and their cooperation with authorities. Although there are some indications for preparatory activities through Social Media, most activity seems to occur during emergency response and recovery. While there is already a multitude of systems that analyze and visualize Social Media data, which might help citizens and authorities to scan and filter for information, only few focus on supporting relief activities of digital and real actors or improving the bidirectional communication between citizens and authorities. Hence, a range of studies outline design implications and requirements for these concerns that should be considered in the development of technological artifacts within the EmerGent project. The typology of role patterns (section 2.5) provides a comprehensive overview indicating that actors might focus on a multitude of different tasks and vary in how deeply they advance into one or multiple hemispheres. In terms of C2C, social software should enhance the communication, cooperation and collaboration between digital and real actors. A2C and C2A channels could create and extend synergetic effects in the cooperation between self help communities and units of authorities. Thus, the promotion of multi hemispheric actors could improve interfaces of interaction between citizen and authorities. 6.2 Data Quality Patterns After a definition of DQ (section 3.1) the section starts with patterns of DQ problems where four different categories of problems are identified (section 3.2). These problems cause four consequences: information overload, misjudgment, misinterpretation and misuse which have to be considered to achieve high DQ. The next sections lead to the main focus of the project: The use of Social Media Data wherein its risks and challenges are described (section 3.3 and 3.4). To understand the fundamental problems by low DQ in general, some examples from past emergencies are presented with focus on Social Media. Challenges and opportunities regarding Social Media data and ways to structure data the data with high potential are captured in the next section. Based on a literature review the problem is identified that structuring data from Social Media relies more on the context of the data and less on technical issues. There are already many technologies that can be used to model complex issues. At a first stage Social Media has to be categorized. Based on the literature a categorization can be done by self presentation / media richness. One result is that the most promising types of Social Media in emergencies are built up by the Social Network Sites (e.g. Facebook) and Content Communities (e.g. YouTube). Further, there are several issues regard 44

52 ing the information objects on the different types of Social Media. The content on Social Media is often noisy, messages are short and the social context has to be considered which of course aggravates the applicability of methods for Information Mining or assessment of DQ. All these concerns have to be taken in mind while trying to strengthen the tie between citizens and authorities via Social Media. 6.3 Related Research Projects Section 4 provides an overview of several research initiatives on utilizing Social Media in emergency management and related Social Media analysis tools. It summarizes studies and technical developments from European Framework projects (section 4.1) as well as other projects (section 4.2) giving an overview of Social Media research in emergency management. It is recognized that there is deep interest in utilizing Social Media in emergency management from various research fields. These research initiatives indicate that there is a positive interest from authorities for using Social Media in emergency management activities. Moreover, there is ongoing work in finding precise requirements in defining and creating Social Media monitoring tools. These initiatives are highlighting the importance of end users involvement in gathering the appropriate functional requirements in order to develop efficient solutions in the context of emergency management. 6.4 Interview Study The performed study (section 5) included a variety of respondents who helped collect the reported results. Participants were from different EU member states and their profile varied in terms of their role in emergency management, operational size, methods and tasks. Social media provides a direct and very fast channel for communicating with the public and spreading accurate and trusted information, especially during emergencies, when it is even more important for people to have access to trusted information. Social media provides the possibility for bi directional communication and it is currently used in the majority of the cases. However, the current practice shows higher use of broadcasting information to citizens, rather than responding to individual cases. The use of Social Media during the different stages of an EMC highlighted the importance of information gathering and validation. This is considered as one of the great challenges in the use of Social Media and in some cases it may prohibit or restrict further use. To overcome this barrier an alternative model of operation is explored by some authorities, where the purpose of using Social Media is distinguished between use for operational purpose versus informational purpose, or use during an emergency versus use during normal operation. The emerging difficulty in handling the information from Social Media, trigger the opportunity for supporting these processes with applicable tools. Authorities already experiment with the use of different tools, although the current study reports lack of a clear understanding of what tool to use or which tool is better in different situations. While implementation strategies or models for integration with current workflows are clearly interesting topics and some authorities currently explore or discuss different options, no clear or documented plans or goals were identified. Interviewees were not active in the research field of Social Media and emergency management. However, all shared an interest to follow 45

53 the advances of this theme and expect the importance of Social Media in emergencies to rise in the future. 46

54 7 References 7.1 Section 2 [Amer12] American Red Cross, ARC: More Americans Using Mobile Apps in Emergencies. URL release/more Americans Using Mobile Apps in Emergencies. abgerufen am [Ball13] Ballve, Marcelo: The World s Largest Social Networks. URL worlds largest social networks #ixzz30MjZz5TK [BeLG13] Bergstrand, Fredrik; Landgren, Jonas; Green, Viktor: Authorities don t tweet, employees do! In: Proceedings of the MobileHCI 2013, 2013 [Birk12] Birkbak, Andreas: Crystallizations in the Blizzard : Contrasting Informal Emergency Collaboration In Facebook Groups. In: Proceedings of the Nordic Conference on Human Computer Interaction (NordiCHI), 2012 ISBN , pp [BoEl07] Boyd, Danah M.; Ellison, Nicole B.: Social Network Sites: Definition, History, and Scholarship. In: Journal of Computer Mediated Communication vol. 13 (2007), Nr. 1, pp [ChCR12] Chen, Charlie; Carolina, North; Ractham, Peter: Lessons Learned from the Use of Social Media in Combating a Crisis: A Case Study of 2011 Thailand Flooding Disaster. In: Proceedings of the International Conference on Information Systems (ICIS). Orlando, USA, 2012, pp [CMPB14] Cobb, Camille; McCarthy, T; Perkins, Annuska; Bharadwaj, Ankitha: Designing for the Deluge: Understanding & Supporting the Distributed, Collaborative Work of Crisis Volunteers. In: Proceedings of the Conference on Computer Supported Cooperative Work (CSCW). Baltimore, 2014 ISBN [DeBK13] Denef, Sebastian; Bayerl, Petra Saskia; Kaptein, Nico: Social Media and the Police Tweeting Practices of British Police Forces during the August 2011 Riots. In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). Paris, France : ACM, 2013 [DeHP12] St. Denis, Lise Ann; Hughes, Amanda L; Palen, Leysia: Trial by Fire : The Deployment of Trusted Digital Volunteers in the 2011 Shadow Lake Fire. In: Proceedings of the 9th International ISCRAM Conference, 2012, pp [EbGH08] Ebersbach, Anja; Glaser, Markus; Heigl, Richard: Social Web. Konstanz : UVK, 2008 [EnRe14] Endsley, Tristan; Reep, James: The Source of the Story: Evaluating the Credibility of Crisis Information Sources. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). vol. 1, 2014, pp

55 [FAAB13] Fuchs, Georg; Andrienko, Natalia; Andrienko, Gennady; Bothe, Sebastian; Stange, Henrik: Tracing the German Centennial Flood in the Stream of Tweets: First Lessons Learned. In: SIGSPATIAL International Workshop on Crowdsourced and Volunteered Geographic Information. Orlando, USA, 2013 ISBN , pp [Gorp14] Gorp, Annemijn F Van: Integration of Volunteer and Technical Communities into the Humanitarian Aid Sector: Barriers to Collaboration. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM), 2014, pp [HDPA14] Hughes, Amanda L.; Denis, Lise A St; Palen, Leysia; Anderson, Kenneth M: Online Public Communications by Police & Fire Services during the 2012 Hurricane Sandy. In: Proceedings of the Conference on Human Factors in Computing Systems (CHI), 2014 [HeRu04] [HeZa10] [HuPa09] [Jenn12] [KaHa10] [KaRe14] [KFSY11] Helsloot, Ira; Ruitenberg, a.: Citizen Response to Disasters: a Survey of Literature and Some Practical Implications. In: Journal of Contingencies and Crisis Management vol. 12 (2004), Nr. 3, pp Heverin, Thomas; Zach, Lisl: Microblogging for Crisis Communication: Examination of Twitter Use in Response to a 2009 Violent Crisis in the Seattle Tacoma, Washington Area. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Seattle, USA, 2010 Hughes, Amanda L.; Palen, Leysia: Twitter Adoption and Use in Mass Convergence and Emergency Events. In: LANDGREN, J. ; JUL, S. (eds.): Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Gothenburg, Sweden, 2009 Jennex, Murray E: Social Media Truly Viable For Crisis Response? In: ROTHKRANTZ, L. ; RISTVEJ, J. ; FRANCO, Z. (eds.): Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Vancouver, Canada, 2012, pp. 1 5 Kaplan, Andreas M.; Haenlein, Michael: Users of the world, unite! The challenges and opportunities of Social Media. In: Business Horizons vol. 53 (2010), Nr. 1, pp Kaufhold, Marc André; Reuter, Christian: Vernetzte Selbsthilfe in Sozialen Medien beim Hochwasser 2013 in Deutschland. In: i com Zeitschrift für interaktive und kooperative Medien vol. 13 (2014), Nr. 1, pp Kavanaugh, Andrea; Fox, Edward A; Sheetz, Steven; Yang, Seungwon; Li, Lin Tzy; Whalen, Travis F.; Shoemaker, Donald; Natsev, Paul; et al.: Social Media Use by Government: From the Routine to the Critical. In: Proceedings of the International Digital Government Research Conference. Maryland, 2011 ISBN , pp

56 [LaSh11] Latonero, Mark; Shklovski, Irina: Emergency Management, Twitter, and Social Media Evangelism. In: International Journal of Information Systems for Crisis Response and Management (IJISCRAM) vol. 3 (2011), Nr. 4, pp [Lévy94] Lévy, Pierre: Collective Intelligence: Mankind s Emerging World in Cyberspace, 1994 [LiPS08] Liu, SB; Palen, Leysia; Sutton, Jeannette: In search of the bigger picture: The emergent role of on line photo sharing in times of disaster. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Washington D.C., USA, 2008 [NaVS12] Nagy, Ahmed; Valley, CMS; Stamberger, Jeannie: Crowd Sentiment Detection during Disasters and Crises. In: Proceedings of the 9th International ISCRAM Conference. Vancouver, Canada, 2012, pp. 1 9 [PaLi07] Palen, Leysia; Liu, Sophia B.: Citizen communications in crisis: anticipating a future of ICT supported public participation. In: Proceedings of the Conference on Human Factors in Computing Systems (CHI). San Jose, USA : ACM Press, 2007 [PBWH12] Perng, Sung Yueh; Büscher, Monika; Wood, Lisa; Halvorsrud, Ragnhild; Stiso, Michael; Ramirez, Leonardo; Al Akkad, Amro: Peripheral response: Microblogging during the 22/7/2011 Norway attacks. In: Proceedings of the 9th International ISCRAM Conference, 2012, pp [Pohl13] Pohl, Daniela: Social Media Analysis for Crisis Management: A Brief Survey. [PRLL13] Pipek, Volkmar; Reuter, Christian; Ley, Benedikt; Ludwig, Thomas; Wiedenhoefer, Torben: Sicherheitsarena Ein Ansatz zur Verbesserung des Krisenmanagements durch Kooperation und Vernetzung. In: Crisis Prevention Fachmagazin für Innere Sicherheit, Bevölkerungsschutz und Katastrophenhilfe vol. 3. Bonn, Germany (2013), Nr. 1 [QHZZ11] Qu, Yan; Huang, Chen; Zhang, Pengyi; Zhang, Jun: Microblogging after a Major Disaster in China: A Case Study of the 2010 Yushu Earthquake. In: Proceedings of the Conference on Computer Supported Cooperative Work (CSCW). Hangzhou, China : ACM, 2011 ISBN , pp [Quar84] Quarantelli, Enrico L: Emergent Citizen Groups in Disaster Preparedness and Recovery Activities, University of Delaware, 1984 [QuDy77] Quarantelli, Enrico L.; Dynes, Russell R.: Response to Social Crisis and Disaster. In: Annual Review of Sociology vol. 3 (1977), Nr. 1, pp [QuWW09] Qu, Yan; Wu, Philip Fei; Wang, Xiaoqing: Online Community Response to Major Disaster : A Study of Tianya Forum in the 2008 Sichuan Earthquake. In: Proceedings of the Hawaii International Conference on System Sciences (HICSS). Waikoloa, USA, 2009 [ReGK07] Reilly, W Scott Neal; Guarino, Sean L; Kellihan, Bret: Model Based Measurement of Situation Awareness. In: Proceedings of the 2007 Winter Simulation Conference, 2007 ISBN , pp

57 [ReHP13] Reuter, Christian; Heger, Oliver; Pipek, Volkmar: Combining Real and Virtual Volunteers through Social Media. In: Proceedings of the 10th International ISCRAM Conference, 2013, pp [ReMP11] Reuter, Christian; Marx, Alexandra; Pipek, Volkmar: Social Software as an Infrastructure for Crisis Management a Case Study about Current Practice and Potential Usage. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Lisbon, Portugal, 2011 [ReMP12] Reuter, Christian; Marx, Alexandra; Pipek, Volkmar: Crisis Management 2.0: Towards a Systematization of Social Software Use in Crisis Situations. In: International Journal of Information Systems for Crisis Response and Management (IJISCRAM) vol. 4 (2012), Nr. 1, pp [ReSc14] Reuter, Christian; Scholl, Simon: Technical Limitations for Designing Applications for Social Media. In: Workshop Proceedings der Tagung Mensch & Computer München, Germany : Oldenbourg Verlag, 2014, pp. 1 7 [SaWT13] [ShPS08] [Star13] [StPa10] San, Yee San; Wardell III, Clarence; Thorkildsen, Zoë: Social Media in the Emergency Management Field: 2012 Survey Results (2013), Nr. June Shklovski, Irina; Palen, Leysia; Sutton, Jeannette: Finding Community Through Information and Communication Technology During Disaster Events. In: Proceedings of the Conference on Computer Supported Cooperative Work (CSCW). San Diego, USA : ACM Press, 2008 Starbird, Kate: Delivering Patients to Sacré Coeur : Collective Intelligence in Digital Volunteer Communities. In: Proceedings of the 31th International Conference on Human Factors in Computing Systems, 2013 ISBN , pp Starbird, Kate; Palen, Leysia: Pass It On?: Retweeting in Mass Emergency. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Seattle, USA, 2010, pp [StPa11] Starbird, Kate; Palen, Leysia: Voluntweeters: Self Organizing by Digital Volunteers in Times of Crisis. In: Proceedings of the 29th International Conference on Human Factors in Computing Systems. Vancouver : ACM Press, 2011 ISBN , pp [StPa12] Starbird, Kate; Palen, Leysia: (How) Will the Revolution be Retweeted? Information Diffusion and the 2011 Egyptian Uprising. In: Proceedings of the Conference on Computer Supported Cooperative Work, 2012 ISBN , pp [StQu85] Stallings, Robert A; Quarantelli, Enrico L: Emergent Citizen Groups and Emergency Management. In: Public Administration Review vol. 45 (1985), Nr. Special Issue, pp [StSt10] Starbird, Kate; Stamberger, Jeannie: Tweak the Tweet : Leveraging Microblogging Proliferation with a Prescriptive Syntax to Support Citizen 50

58 Reporting. In: Proceedings of the 7th International ISCRAM Conference, 2010, pp. 1 5 [SuPS08] Sutton, Jeannette; Palen, Leysia; Shklovski, Irina: Backchannels on the Front Lines: Emergent Uses of Social Media in the 2007 Southern California Wildfires. In: FRIEDRICH, F. ; VAN DE WALLE, B. (eds.): Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Washington D.C., USA, [Sutt10] Sutton, Jeannette: Twittering Tennessee: Distributed networks and Collaboration Following a Technological Disaster. In: FRENCH, S. ; TOMASZEWSKI, B. ; ZOBEL, C. (eds.): Proceedings of the Information Systems for Crisis Response and Management (ISCRAM). Seattle, USA, 2010 [TiRa13] [VaOR13] [VHSP10] [WhPA14] [Wile14] [WMAR13] [WPKH09] [YCLL13] Tim, Yenni; Ractham, Peter: The Emergence of Social Media as Boundary Objects in Crisis Response: A Collective Action Perspective. In: Proceedings of the International Conference on Information Systems (ICIS). Milan, Italy, 2013, pp Valecha, Rohit; Oh, Onook; Rao, R: An Exploration of Collaboration over Time in Collective Crisis Response during the Haiti 2010 Earthquake. In: Proceedings of the International Conference on Information Systems (ICIS). Milan, Italy, 2013, pp Vieweg, Sarah; Hughes, Amanda L; Starbird, Kate; Palen, Leysia: Microblogging During Two Natural Hazards Events : What Twitter May Contribute to Situational Awareness. In: Proceedings of the 28th International Conference on Human Factors in Computing Systems, 2010, pp White, Joanne I; Palen, Leysia; Anderson, Kenneth M: Digital Mobilization in Disaster Response: The Work & Self Organization of On Line Pet Advocates in Response to Hurricane Sandy. In: Proceedings of the Conference on Computer Supported Cooperative Work (CSCW). Baltimore, USA : ACM, 2014 ISBN Wilensky, Hiroko: Twitter as a Navigator for Stranded Commuters during the Great East Japan Earthquake. In: Proceedings of the Information Systems for Crisis Response and Management (ISCRAM), 2014, pp Wulf, Volker; Misaki, Kaoru; Atam, Meryem; Randall, David; Rohde, Markus: On the Ground in Sidi Bouzid: Investigating Social Media Use during the Tunisian Revolution. In: Proceedings of the 2013 conference on Computer supported cooperative work (2013), pp ISBN White, Connie; Plotnick, Linda; Kushma, Jane; Hiltz, Starr Roxanne; Turoff, Murray: An online social network for emergency management. In: International Journal of Emergency Management (IJEM) vol. 6 (2009), Nr. 3/4, pp Yang, Seungwon; Chung, Haeyong; Lin, Xiao; Lee, Sunshin; Chen, Liangzhe: PhaseVis: What, When, Where, and Who in Visualizing the Four Phases of Emergency Management Through the Lens of Social Media. In: Proceedings of 51

59 [ZKTE14] the Information Systems for Crisis Response and Management (ISCRAM). Baden Baden, Germany, 2013, pp Zisgen, Julia; Kern, Julia; Thom, Dennis; Ertl, Thomas: #Hochwasser Visuelle Analyse von Social Media im Bevölkerungsschutz. In: i com Zeitschrift für interaktive und kooperative Medien vol. 13 (2014), Nr. 1, pp Section 3 [AgYi10] [AlMa99] [Baez09] [BAB07] [BoRo12] [Comm13] [CPD09] [CWT13] [EpHe04] [Eppl06] [ERGB12] Agarwal, N., & Yiliyasi, Y. (2010): Information Quality Challanges in Social Media. Paper presented at the International Conference on Information Quality (ICIQ) University of Arkansas at Little Rock (UALR). Janet E. Alexander, Marsha A. Tate (1999): Web Wisdom; how to Evaluate and Create Information Quality on the Webb (1st ed.). L. Erlbaum Assoc. Inc., Hillsdale, NJ, USA. Baeza Yates, R. (2009): User generated content: how good is it? Paper presented at the 3rd workshop on Information credibility on the web, Madrid, Spain. Bharosa, N., Appelman, J., & de Bruin, P. (2007): Integrating technology in crisis response using an information manager: First lessons learned from field exercises in the Port of Rotterdam. In Proceedings of the International Conference on Information Systems for Crisis Response and Management, Brussels, Belgium. Bontcheva, K., Rout, D. (2012): Making Sense of Social Media Streams through Semantics: a Survey, in 'Semantic Web journal'. COMMITTEE ON PUBLIC RESPONSE TO ALERTS AND WARNINGS USING SOCIAL MEDIA: CURRENT KNOWLEDGE AND RESEARCH GAPS (2013): Public Response to Alerts and Warnings Using Social Media: Report of a Workshop on Current Knowledge and Research. Chai, K, Potdar, V & Dillon, T. (2009): Content Quailty Assessment Related Frameworks for Social Media in Lecture Notes of Computer Science (LCNS) for Computational Science and Its Applications (ICCSA), pp Carter, S.; Weerkamp, W; Tsagkias, M. (2013): Microblog language identification: overcoming the limitations of short, unedited and idiomatic text. Lang. Resour. Eval. 47, 1 (March 2013), pp Eppler, M.; Helfert, M. (2004): A classification and analysis of data quality costs, Ninth International Conference on Information Quality, November 5 7, 2004, MIT. Eppler, M. (2003): Managing information quality: Increasing the value of information in knowledge intensive products and processes. Berlin, Germany: Springer Verlag. Emamjome, Fahame F., Rabaa'i, Ahmad A., Gable, Guy G., Bandara, W.asana (2013): Information quality in social media: a conceptual model. In Lee, Jae 52

60 [FiKi01] [GuKu12] [HCH10] [HuPa09] [Jens12] [JPH99] [KaRa12] [KaHa10] [KKKS+11] [Klei99] [KSW02] [LaSh11] Nam, Mao, Ji Ye, & Thong, James (Eds.) Proceedings of the Pacific Asia Conference on Information Systems (PACIS 2013), AIS Electronic Library (AISel), Jeju Island, Korea. Fisher C.W., Kingma B.R. (2001): Criticality of data quality as exemplified in two disasters, Information & Management, 39 (2), pp Gupta A., Kumaraguru P. (2012): Credibility ranking of tweets during high impact events, in Proceedings of the 1st Workshop on Privacy and Security in Online Social Media PSOSM 12. Huang, C., Chan, E., Hyder, A.A. (2010): Web 2.0 and internet social networking: A new tool for disaster management? Lessons from Taiwan. BMC Medical Informatics and Decision Making, 10(57). Hughes, A., Palen, L. (2009): Twitter adoption and use in mass convergence and emergency events. In Proceedings of the International Conference on Information Systems for Crisis Response and Management, Gothenburg, Sweden. Jensen, G.E. (2012): Key criteria for information quality in the use of online social media for emergency management in New Zealand. Master Thesis, Victoria University of Wellington Johnston, D., Paton, D. and Houghton, B. (1999): Volcanic hazard management: promoting integration and communication, in Ingleton, J. (Ed.), Natural Disaster Management, United Nations (IDNDR), Coventry. Kane, G., Ransbotham, S. (2012): Codification and Collaboration: Information Quality in Social Media. Thirty Third International Conference on Information Systems, Orlando. Kaplan Andreas M., Haenlein M. (2010): Users of the world, unite! The challenges and opportunities of Social Media, Business Horizons, Volume 53, Issue 1, pp Kuehn, A., Kaschewsky, M., Kappeler, A., Spichiger, A., Riedl, R. (2011): Interoperability and Information Brokers in Public Safety: An Approach toward Seamless Emergency Communications. Journal of Theoretical and Applied Electronic Commerce Research, 6(1), pp Klein, G. (1999): Sources of Power: How People Make Decisions. Cambridge, MA: MIT Press. Kahn, B. K.; Strong, D. M., Wang, R. Y. (2002): Information quality benchmarks: Product and service performance. Communications of the ACM, 45 (4), pp Latonero, M., Shklovski, I. (2011): Emergency Management, Twitter, and Social Media Evangelism. International Journal of Information Systems for Crisis Response and Management (IJISCRAM), 3(4), pp doi: /jiscrm

61 [LIM10] [LWMR13] [OAR13] [OKR10] [Piro09] [Redm96] [SLW97] [SPS08] [StPa11] [SLW97] [VREF08] [WLZ01] [WaSt96] [WPKH+09] Liu, S. B., Iacucci, A. A., Meier, P. (2010): Ushahidi in Haiti and Chile: Next generation crisis mapping. In Proceedings of the American Congress on Surveying and Mapping. Landrigan, L.C.; Wade, J.P.; Milewski, A.; Reagor, B. (2013): Lessons from the past: Inaccurate credibility assessments made during crisis situations, Technologies for Homeland Security (HST), 2013 IEEE International Conference on, vol., no., pp.754,759, pp Oh, O; Agrawal, M, Rao, H. R. (2013): Community intelligence and social media services: a rumor theoretic analysis of tweets during social crises. MIS Q. 37, 2 (June 2013), pp Oh, O.; Kwon, K. H., Rao, H. R. (2010): An Exploration of Social Media in Extreme Events: Rumor Theory and Twitter during the Haiti Earthquake 2010., in Rajiv Sabherwal & Mary Sumner, ed., 'ICIS', Association for Information Systems, p 231. Pirolli, P. (2009): Powers of 10: Modeling complex information seeking systems at multiple scales. IEEE Computer, 42(3): pp Redman, T. C. (1996): Data quality for the information age. Artech House. Strong Diane M., Lee Yang W., Wang Richard Y. (1997): Data quality in context, Communications of the ACM, v.40 n.5, pp Sutton, J., Palen, L., Shklovski, I. (2008): Back channels on the front lines: Emergent use of social media in the 2007 southern California fires. In Proceedings of the International Conference on Information Systems for Crisis Response and Management, Washington DC. Starbird, K., Palen, L. (2011): Voluntweeters: Self organizing by digital volunteers in times of crisis. In Proceedings of the ACM Conference on Computer Human Interaction, Vancouver, BC, Canada. Strong, D. M., Lee, Y. W., Wang, R. Y. (1997): 10 Potholes in the Road to Information Quality. Computer IEEE, 30, pp van de Ven, J., van Rijk, R., Essens, P., Frinking, E. (2008): Network centric operations in crisis management. In Proceedings of the International Conference on Information Systems for Crisis Response and Management, Washington, DC. Wang R.Y., Lee Y. W., Ziad M. (2001): Data Quality, Publisher: Springer Wang, R., Strong, D. (1996): Beyond accuracy: What data quality means to data consumers. Journal of Management Information Systems, 12(4), 5 3. White, C., Plotnick, L., Kushma, J., Hiltz, S., Turoff, M. (2009): An online social network for emergency management. In Proceedings of the International Conference on Information Systems for Crisis Response and Management, Gothenburg, Sweden. 54

62 [WSF95] Wang, R. Y., Storey, V. C., Firth, C. P. (1995): A framework for analysis of data quality research. IEEE Trans. Know. Data Eng. 7, 4, pp [W3C07] W3C, Wiki (2007): Disaster Management. [Online]. Available: [Accessed: 05/06/2014]. [ZhGa00] 7.3 Section 5 [WWW01] Zhu, X., Gauch, S. (2000): Incorporating quality metrics in centralized/distributed information retrieval on the World Wide Web. Proceedings of the 23rd annual international ACM SIGIR conference on Research and development in information retrieval, Athens, Greece pp European Emergency Number Association [online], available under: [accessed: 2014/06/06] [WWW02] Hootsuite [online], available under: [accessed: 2014/06/23] [WWW03] Storify [online], available under: [accessed: 2014/06/23] [WWW04] [WWW05] [WWW05] [WWW06] Social Mention [online], available under: [accessed: 2014/06/23] Coosto [online], available under: [accessed: 2014/06/23] TweetDeck [online], available under: [accessed: 2014/06/23] Twitter Alerts [online], available under: [accessed: 2014/06/23] 55

63 Annex A: Interview Questions Section A: Introduction to the organization A1. Specify the type of your organization. A2. Specify your PSAP model. A3. General Information about you A4. Geographical and regional information A5. What are the main risks in your region? A6. Specify the vulnerability of your region: A7. Estimate the average number of calls / emergency calls per anum for the following categories. A8. Estimate the average number of emergency operations per anum for the following categories. A9. Specify the number of your workforce (ft/pt) A10. On which administrative division is your organization operating in your state? Section B: Information about the PSAP and/or dispatching center you are part of, or responsible for. B1. Select the main tasks within your PSAP (Multiple choices are welcome) B2. Are there any further tasks your PSAP is responsible for? B3. Specify in which phases of the emergency management cycle your organization operates and describe the main tasks. B4. Select the communication channels that your PSAP supports to handle input messages (i.e. messages from citizens). B5. Select the communication channels that your PSAP supports to send out messages (i.e. messages to inform or support citizens) B6. Select which communication channels your PSAP uses to broadcast warning messages. A warning message should inhabitants make aware of a danger. B7. Please describe the special warning systems that your PSAP uses to send out warning messages. B8. Select the responsible roles for output messages. Section C: Technical Information about the PSAP and/or dispatching center you are part of, or responsible for. Contributed to D3.2 Guidelines for Social Media integration into existing authorities systems. Section D: The use of Social Media in emergencies 56

64 D1. Does your organization already use Social Media, e.g. to gather or spread information? D2. Select the Social Networks that are used by your organization. D3. Who is allowed to post/send out data? D4. Is information from Social Media provided by citizens used in authorities? D5. Which tools do you use to analyze/collect data from Social Media (e.g. Twitcident)? D6. Who is allowed to use gathered data from Social Media in your organization? D7. Does your organization use other web 2.0 technologies? D8. In which phases of an emergency do you use Social Media? Please select and describe the usage. D9. Please try to categorize the main purposes in using Social Media. D10. What is the main purpose in using Social Media? Please describe in detail. D11. How does your organization finance Social Media usage? D12. Please try to categorize the implementation strategy that would fit at most to your current implementation. D13. Describe your organizations implementation strategy on Social Media in detail. D14. Please describe how you or your organization has implemented Social Media into workflows or processes within your organization. Try to describe the main workflows and how the implementation is done. D15. Is Social Media an interesting topic for your organization? D16. Select the Social Networks which are most interesting for your organization in order to inform, encourage or engage citizens. D17. Please try to categorize the main purposes that you would like to focus in Social Media. D18. Describe the main purposes in the use of Social Media. D19. How would you finance the usage of Social Media? D20. Please try to categorize the implementation strategy that would fit at most. D21. Please try to describe a suitable implementation strategy on Social Media for your organization. D22. Please describe how you or your organization would implement Social Media into workflows or processes within your organization. Try to identify the main workflows and describe how the implementation is done. D23. Please describe why Social Media isn t an interesting topic for you. D24. Does your organization make research and development on Social Media? 57

65 D25. Please describe the research and/or development that your organization does on Social Media. D26. Is your organization following or would like to follow research and development on Social Media? D27. Please describe the research topics you're interested in. Section E: Challenges and benefits on Social Media in emergencies E1. What are the main benefits of Social Media use? E2. What are the main difficulties or challenges experienced with working with Social Media? E3. If you have some interesting references to cases studies, SWOT or risk analysis regarding Social Media, especially Social Media in emergencies, please paste them below. Section F: Future plans F1. How do you think the impact of Social Media in emergencies will change in the next years? F2. How will your organization deal with Social Media in the next years? F3. How will workflows and tools evolve? F4. What are the new technologies you are looking at? F5. Do you think research projects like EmerGent are helpful to gain new insights? 58