STAFF REPORT ACTION REQUIRED Social Sentiment Analysis Date: May 31, 2013 To: From: Wards: Reference Number: Government Management Committee Director, 311 Toronto Chief Information Officer All P:\2013\Internal Services\I&T\gm13005I&T (AFS # 17768) SUMMARY This report is in response to the request of the Government Management Committee for information on the potential implementation of social sentiment analysis (SSA) capability at the City of Toronto. Social sentiment analysis refers to the ability to gain insight into what people using social media think and want on a mass scale by gauging behavioural patterns and trends. The Deputy Chief Information Officer and the Director, 311 Toronto have begun to review and assess the potential of implementing social sentiment analysis capability at the City of Toronto. At this preliminary stage, we are utilizing the efforts of two research and advisory firms (Gartner Inc. and Forrester Research Inc.) for an environmental scan to better position the project to gain as much insight about SSA and how it is being used in the public and private sectors. As SSA is a new technology, this information will help determine if it is worthwhile to invest in a possible pilot study of SSA within 311 Toronto. RECOMMENDATIONS The Chief Information Officer and the Director, 311 Toronto recommend that: 1. The Government Management Committee receive this report for information. Staff report for action on Social Sentiment Analysis 1
Financial Impact There are no financial implications arising from this report at this stage The Deputy City Manager and Chief Financial Officer has reviewed this report and agrees with the financial impact information. DECISION HISTORY This report responds to a motion from Government Management Committee regarding social sentiment analysis. Item GM19.6, January 21, 2013, titled Social Sentiment Analysis, requested the Acting Chief Information Officer and the Director of 311 to report back on the potential implementation of social sentiment analysis capability at the City of Toronto with options and a strategy for implementation that includes an outline of the business objectives, resources, operational impact and training assessment, costbenefit analysis and an implementation framework. The minutes of the Committee meeting can be found here: http://app.toronto.ca/tmmis/viewagendaitemhistory.do?item=2013.gm19.6 ISSUE BACKGROUND Today, through social media channels such as Facebook and Twitter, people publicly share thoughts, knowledge, experiences and opinions on products, programs and services to an extent not previously possible. Businesses, educators, researchers and the media use social media analytics to better understand the voice of their customers. In the public sector setting, social sentiment analysis (SSA) the integration of classic text analytics and new techniques that track feelings and emotions allows governments to monitor, identify and analyze relevant unstructured data in the form of public conversations and discussions for emerging issues and trends. The City has an opportunity to better gauge public opinion and experience regarding key topics, such as open, transparent and accessible government and the provision of services to residents, visitors and businesses, through such monitoring and analysis. SSA (in accordance and compliance with the City's legislated privacy requirements) may provide an additional tool to assist City officials in making informed decisions. SSA technology can track, mine and parse data collected from tweets, public messages, blog posts, user forums and other social media sources to aggregate public reaction and opinion on services, products, brands, current events and so on. In addition to aggregating public opinion and highlighting emerging trends and issues, the City could crowdsource input from residents with SSA assistance. (Crowdsourcing, as applied here, is defined as the practice of mass collaboration to obtain information on services, share ideas and build content by soliciting contributions from a large group of people and especially from the online community.) Staff report for action on Social Sentiment Analysis 2
A simple example of this: 311 Toronto launches a new mobile application (or mobile app). SSA identifies that users are having difficulty downloading the app onto their Blackberry Curve smartphones. 311 learns of this through the use of SSA because, before launch, 311 had created a category in the SSA tool to search social media sites based on the key phrase, "new Toronto mobile app". As a result, 311 was able to use the analysed data to work with the mobile app developer to fix the Blackberry Curve bug prior to customers contacting 311 about the problem. This scenario already takes place on a smaller scale. 311 Toronto manually monitors tweets to its @311Toronto Twitter account and acts accordingly, based on the content, tone and number of messages coming in at any given time about a particular issue. This only works if the tweets are directed to the @311Toronto account or are tagged appropriately, whereas, by "listening" to social media, as in the above example, 311 would be able to act proactively. At a higher level, using this example, SSA technology could monitor where and when residents have concerns, the prevailing level of satisfaction, suggestions for improvement and other factors. The City could potentially then correlate this with information that currently exists within internal systems to inform future programs. This could be communicated back to the public in a proactive manner, resulting in reduced use of resources to manage complaints and an increase in trust and confidence in City government. COMMENTS This report provides an update on the Social Sentiment Analysis review that is being conducted by Deputy Chief Information Officer and the Director, 311 Toronto as per the request of the Government Management Committee. Corporate Meetings Social Sentiment Analysis is an emerging field. The scope of possibility for implementation of SSA technology ranges from the use of simple online "listening" tools to the adoption of much more complex and costly enterprise capabilities. Social listening is more about what the public is saying about city government than what they are saying to city government. To initiate the review and assessment of the current state of social sentiment analysis capability within the corporation, a working group of various City divisions and offices (Parks, Forestry & Recreation, Civic Engagement, Strategic Communications, I&T, 311 Toronto) met to identify their varied interests in utilizing SSA and to determine potential opportunities for conducting a pilot project. Examples include: Staff report for action on Social Sentiment Analysis 3
311 Toronto is the North American leader in municipal use of speech analytics (SA). 311 uses speech analytics to better understand the voice of their customers. Speech analytics utilizes the same functionality as social sentiment analytics (SA converts the audio recordings of each call to 311 into text for analysis). Using social sentiment analysis in conjunction with speech analytics, 311 would be able to gather information on key City issues such as public transit, public libraries, graffiti, potholes or other municipal information across the multiple channels (phone, e-mail, online, mobile app, etc) used by its customers. In addition to mining social media, the Parks, Forestry & Recreation Division plans to conduct text mining and sentiment analysis on comments captured in public meetings, deputations, open-ended text in surveys, news clippings and other electronic sources. A sentiment analysis tool will be an important part of the division's civic engagement and analysis capacity building program. The Social Development Finance & Administration's Research Unit is considering this type of technology to determine community salience and sentiment on broad citywide social issues. It will also be useful as an additional data source in public consultations, and as a social context to the new data to be released in 2013 through the on-line mapping application, Wellbeing Toronto. Although the possible applications for this solution may differ across City groups, the SSA technology foundation will allow for shared analytics capability across the corporation. Research and Environmental Scan The Deputy Chief Information Officer and the Director, 311 Toronto engaged with two research and advisory firms (Gartner Inc. and Forrester Research Inc.) for an environmental scan on the use of the SSA within public and private sectors with a primary focus on how other governments or municipalities are using SSA. Gartner Inc is currently conducting a readiness study on SSA and has committed to provide the results to the City by July 2013. Forrester Inc. has recently completed a maturity study of SSA and provided examples of other cities using SSA which include monitoring online reputation, soliciting and responding to feedback, determining outreach approaches and providing assistance during a natural disaster. There are a number of different technologies that are being used by other cities and there does not appear to be an identified leader within the field nor an identified roadmap for the best approach for utilizing this technology. As a result, and as SSA is an emerging technology, this review will be cognizant of whether or not the technology is mature enough to warrant investment in an SSA pilot. Staff report for action on Social Sentiment Analysis 4
Resources As part of the preliminary work for SSA, a business lead from 311 Toronto has been assigned to oversee a possible pilot project and to hire a Business Analyst. The business lead has expertise in speech analytics and worked to establish the speech analytics capabilities within 311. A Speech Specialist from Bell Canada will also be available to help us align these components (SSA and SA) together. The Bell Specialist is currently engaged on a separate speech project so there will be no additional cost for this work. In addition, a Business Analyst with expertise in social media was hired to conduct research, develop possible RFP documents and conduct the pilot study. Review Current Technical Capabilities 311 Toronto has engaged its current technical partners (KANA and Verint) to determine whether we can leverage the existing technology within 311 (i.e. speech analytics) for a potential pilot study so that there is no additional funding required and the start up time would be reduced. Action Taken Description 1. Corporate Meetings Met with a number of City divisions and offices including Parks, Forestry & Recreation, Civic Engagement, Strategic Communications, I&T, 311 Toronto, City Clerk's [CIMS], Social Development, Fin & Admin [Social Policy, Analysis & Research], CMO [Performance Management] to identify opportunities or existing projects for SSA 2. Research Completed detailed research on social sentiment analysis 3. Environmental Scan Gartner Inc Engaged the Gartner Group on opinions and strategies related to their experience with Social Sentiment Readiness Study anticipated to be completed in July 2013 Forrester Research Inc Engaged Forrester Research on opinions and strategies related to their experience with Social Sentiment Forrester Research provided a maturity study on SSA in May 2013 Staff report for action on Social Sentiment Analysis 5
Action Taken Description 4. Resources Assigned a business lead to oversee the pilot Engaged a Bell Canada Speech Specialist to assist in the alignment of 311 Toronto's Speech Analytics with Social Sentiment Analytics Hired a Business Analyst to conduct the research, prepare RFP documents (if required) and complete the pilot study 5. Current Technical Capabilities Met with 311 Toronto's current technical partners (KANA and Verint) to determine whether we can leverage the existing technology within 311 Next Steps The following is a list of the next steps for this SSA project: 1. Based on the findings from the readiness study being conducted by Gartner Inc in July 2013, determine if an SSA pilot study is feasible. 2. Conduct the pilot study (if appropriate). 3. Provide a follow-up report to the Government Management Committee in the fourth quarter of 2013. CONTACT Neil Evans Director, 311 Toronto nevans@toronto.ca 416-338-7789 Lan Nguyen Deputy Chief Information Officer lpnguyen@toronto.ca 416-397-9822 SIGNATURES Neil Evans Director, 311 Toronto Rob Meikle Chief Information Officer Staff report for action on Social Sentiment Analysis 6