Knowledge Management INF5100 Autumn 2006 Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 2 1
Application Domain: Rescue and Emergency Applications Participants from different organizations paramedics, police, fire, Rescue Site Leader, Team Leaders Dynamic environment movement and activity on site personnel arriving and leaving October 2006 3 Sparse Mobile Adhoc Networks minimal infrastructure, few nodes heterogeneity, limited resources (battery, bandwidth) a lot of movement; frequent disconnections; delay tolerance October 2006 4 2
Knowledge Management (KM) in Sparse MANETs Definition for KM: the tools, techniques and processes for the most effective management of an organization s intellectual assets (Davies et al 2003). Adapted to information sharing in Sparse MANETs: effective management of the intellectual assets (information resources) available for sharing in a Sparse MANET October 2006 5 Knowledge Management (KM) in Sparse MANETs Information sharing and content integration not solved sufficiently in middleware for SMANETs today. KM offer solutions, but these do not consider challenges posed by SMANETs Beneficial for dynamic environments (e.g. rescue operations) to combine middleware infrastructure provided by SMANET with KM solutions KM solutions may be valuable contribution to SMANETs - and vice versa October 2006 6 3
Problem Statement Network wide information sharing in rescue operations Avoid information overflow Cross organizational administration Information not static, frequent updates Only partial view of available information Three main tasks Establish who needs what information Enable vocabulary sharing & mapping Efficient metadata management October 2006 7 Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 8 4
What is Knowledge Management? Started in the 80s, Management Theory with the notion that all knowledge can be formalized the goal was to automatize production processes Multidisciplinary: political science, communication studies, IT, management sciences, Knowledge central seen as part of an organization's competence Central questions in KM: what is knowledge in the production process how can the knowledge flow be improved October 2006 9 What is Knowledge Management? the tools, techniques and processes for the most effective management of an organization s intellectual assets (Davies et al 2003) a dynamic, continuous organizational phenomenon of interdependent processes with varying scopes and changing characteristics. (Alavi/Leidner 2001) October 2006 10 5
Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 11 Knowledge - Complementary Definition (Gardner95): KNOWING what information is needed how information must be processed why which information is needed where information can be found to achieve a specific result when which information is needed October 2006 12 6
Perspectives on Knowledge Source: Alavi/Leidner 2001, p.111 October 2006 13 Hierarchical View of Knowledge Common in IT Data: raw numbers and facts - symbols not yet interpreted Information: interpreted data - data which has been assigned a meaning Always linked to specific situation, has only limited validity Knowledge: personalized information enables people to act and to deal intelligently with all the available information sources. (action component) Whole set of insights, experiences and procedures considered correct and true, guide people s thoughts, behavior and communication. Always applicable in several situations, valid over a relatively long period of time. October 2006 14 7
Types of Knowledge The most common taxonomy Explicit: facts, in documents, models, pictures articulated, codified, and communicated in symbolic form and/or natural language Tacit: implicit, a mental model, skills rooted in action, experience and involvement in a specific context cognitive elements: mental models: mental maps, beliefs, paradigms, view-points technical elements: concrete know-how, crafts, skills apply to specific context, e.g. knowledge of the best way to approach a customer. Individual: is created by and exists in the individual Social/Collective: is created by and inherent in the collective actions of a group October 2006 15 Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 16 8
Knowledge Management Processes Creating knowledge develop new - or replace existing - content within an organization s knowledge socialization, combination, externalization, internalization Storing/retrieving knowledge storage, organization, and retrieval of knowledge Transferring and Sharing knowledge communicating and sharing knowledge Applying knowledge integrate and make good use of knowledge in the organization October 2006 17 Knowledge Management Processes -RoleofIT Source: Alavi/Leidner 2001, p.125 October 2006 18 9
Knowledge Storage/Retrieval Organizational Memory Knowledge in various forms e.g., documentation, structured information in databases, knowledge stored in expert systems, organizational procedures and processes Semantic memory: general, explicit and articulated knowledge e.g., organizational archives of annual reports Episodic memory: context-specific and situated knowledge e.g. specific circumstances of organizational decisions and their outcomes, place and time October 2006 19 Knowledge Storage/Retrieval Role of IT Enhancement and expansion of semantic and episodic organizational memory Increase speed of access to organizational memory Effective tools: Query languages, multimedia databases, DBMSs Groupware: enable creation and sharing of intra-organizational memory October 2006 20 10
Knowledge Sharing (KS) and Transfer Sharing vs. Transfer: Transfer: focus, a clear objective, unidirectional Sharing: can be unintentionally, multiple directionally, without a specific objective may occur between and among individuals within and among teams among organizational units among organizations KM Systems for KS: repositories databases of knowledge (knowledge bases) networks facilitate communications among team members or groups of individuals October 2006 21 Knowledge Sharing (KS) and Transfer Knowledge about where the knowledge is often as important as the original knowledge itself Sharing this kind of metadata important E.g. corporate directories: who knows what in organization Knowledge transfer is driven by communication processes and information flows Forms of knowledge transfer: informal/formal, personal/impersonal Knowledge transfer to locations where it is needed and can be used is important October 2006 22 11
Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 23 Knowledge Management Systems (KMS) IT-based systems developed to support and enhance all KM processes Three common applications: the coding and sharing of best practices the creation of corporate knowledge directories the creation of knowledge networks Requirements must provide ontologies must provide search capabilities often provide filter capabilities (filters can be computerbased or human-based) provide opportunities for collaboration and use of expertise October 2006 24 12
KMS and Knowledge Bases Two main components of KMSs: knowledge bases and ontologies A knowledge base is a database Usually domain dependent Information may need to be abstracted, synthesized, or integrated with other information (e.g. in best practices databases) Ontologies provide shared vocabulary and facilitates reusability October 2006 25 Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 26 13
What is an Ontology The term ontology can mean different things glossaries & data dictionaries thesauri & taxonomies schemas & data models formal ontologies & inference Many definitions the most commonly used: An ontology is an explicit specification of a conceptualization. (Gruber) October 2006 27 What is an Ontology Basically a model of some part of the world (Universe of Discourse) Defines a common vocabulary for sharing information in a domain Specifies terms for classes/concepts and relations between these informal text or using formal language (e.g. predicate logic) October 2006 28 14
Ontology Modelling & Implementation Can be modelled using different knowledge modelling techniques and implemented in various kinds of languages Heavyweight ontologies: AI based languages (framebased, first order logic): e.g., Ontolingua, LOOM Ontology mark-up languages: RDF(S), DAML + OIL, OWL Only Lightweight ontologies : Techniques from software engineering & databases: UML, ER, SQL-scripts Not as expressive October 2006 29 Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 30 15
Types of Ontologies We will look at two categorizations These are based on the richness of the internal structure Lightweight ontologies Heavyweight ontologies (ontology proper) the subject of their conceptualization October 2006 31 Lightweight Ontologies Catalogs: controlled vocabulary a finite list of terms. Glossary: list of terms and meaning as natural language statements. Not machine processable. Thesaurus: a networked collection of controlled vocabulary terms synonym relationship. No explicit hierarchy. Informal is-a hierarchies: not strict subclass Top-level categories and specifications of these (e.g. Yahoo). October 2006 32 16
Heavyweight Ontologies Formal is-a strict subclass hierarchies, necessary for exploiting inheritance Formal instance relationships (formal is-a) includes domain instances Frames ontology includes classes with property information. All subclasses inherit properties. Value restrictions More expressive ontologies, can place restrictions on values that can fill a property. Expressing general logical constraints the most expressive, first order logic. October 2006 33 Lightweight vs. Heavyweight Ontologies Ontology Spectrum, (McGuinnes, 2002) October 2006 34 17
Types of Ontologies Based on the Subject of the Conceptualization Top-level ontologies aka Upper-level ontologies, general concepts, existing ontologies link root terms to these (e.g. Cyc, SUMO) Domain ontologies Reusable in a specific domain (KM, medical, law, engineering, chemistry etc. ) E.g., UMLS (medical) Application ontologies application dependent, often extend & specialize vocabulary of a domain October 2006 35 Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 36 18
Use of Ontologies in KM Knowledge representation Offer a way to cope with heterogeneous representations of resources Give shared and common understanding of a domain Can be communicated between people and application systems October 2006 37 Information Sharing and Integration Interoperability problem have to make the different systems and domains understand each other Structural heterogeneity data structures, schema solutions from domain of distributed databases Semantic heterogeneity meaning of content ontologies possible solution October 2006 38 19
Ontologies in Information Integration As solution to semantic heterogeneity problem: explicitly describe semantics of information sources language for translation 3 General approaches: (Wache et. al 2001) Single: global ontology with shared semantics Multiple: need mapping between (each pair of) ontologies (inter-ontology mapping) Hybrid: multiple ontologies are built on top of or linked to a shared vocabulary of basic terms (may function like a bridge or a translation) October 2006 39 Single, Multiple, and Hybrid Ontology Approaches single ontology approach global ontology Or Top-level ontology shared vocabulary local ontology local ontology local ontology multiple ontology approach local ontology local ontology local ontology hybrid ontology approach October 2006 40 20
Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 41 Ad-Hoc InfoWare Simplify application development for Sparse MANETS Configurable MW services scalable protocols and services Tradeoff between abstraction and awareness of location, resources, context,... between non-functional requirements, e.g. performance vs. security and availability Separation of mechanisms and policies Coordination of knowledge management and resource management Integration of information Information, data, meta-data, resources Context awareness Resource and QoS aware data placement Scenario domain: Rescue and emergency applications October 2006 42 21
Ad-Hoc InfoWare Architecture Overview Knowledge Manager Semantic Metadata & Ontology Framework Profile & Context Mgnt Query Mgnt XML/RDF parser Data Dict. Manager SDDD LDD Resource Manager Replic. Mgnt Proposal Unit Resource Monitor Adjac. Monitor Local Monitor Resource Avail. Watchdogs Watchdogs Manager Watchdogs Execution Envir. Distributed Event Notification Service State Mgnt Delivery Storage Mgnt Availability & Scaling Security and Privacy Manager Authentication Access Control Key Management Encryption October 2006 43 Knowledge Management (KM) in Ad-Hoc InfoWare Manage knowledge sharing and integration in a Sparse MANET Adds layer of knowledge Services that allow relating metadata descriptions to semantic context. Only give tools (not decide usage & content) Share information about where to find knowledge about what October 2006 44 22
Related to KM Elements Hierarchical view of knowledge Explicit knowledge Focused KM processes: Storage/Retrieval and Transfer ( or Knowledge Sharing) Not addressing learning aspect (knowledge creation) Use of ontologies Domain ontologies, e.g. medical, police, fire Upper level ontology/ shared vocabulary (similar to Hybrid approach) Ontology based update Metadata enriched with terms/concepts from ontologies Only ontology use (development etc not during rescue operation) October 2006 45 The Knowledge Manager Distributed Event Notification System Watchdogs Resource Management Knowledge Manager Data Dictionary Mgnt. LDD SDDD Semantic Metadata & Ontology Framework Profile & Context Mgnt Query Mgnt XML Parser AVAILABILITY RETRIEVAL UNDERSTANDING EXCHANGE INFORMATION OVERLOAD Security and Privacy Management SDDD = Semantic Linked Distributed Data Dictionary. LDD = Local Data Dictionary. October 2006 46 23
Three Types of Metadata Information structure and content description metadata Data Dictionary Management Content, formats, data types etc Semantic metadata Semantic Metadata and Ontology Framework Relations between concepts, e.g. is-a, haspart, hasresource, hasdevicetype Profile and context metadata Profile and Context Management User profile, device profile Context: location, time, situation October 2006 47 Profiles and Context Profiles What, who Device type, resources, groups etc (for device profile) User preferences, roles, personalia etc (for user profile). Fairly static information Context Where, when, why location, time, situation (e.g. rescue operation) Dynamic information (network nodes moving) Used in different meanings (the term context) time, location and situation for a device or user semantic or topical context October 2006 48 24
Three-layered Approach Conceptual Ontology layer Semantic/ topical Context (Instance) (Link) Implementation SDDD linking level Information layer LDD metadata SDDD = Semantic Linked Distributed Data Dictionary. LDD = Local Data Dictionary. October 2006 49 Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 50 25
Approach to Ontology Based Update Ontologies to represent rescue operation context model profiles for user, device and information Update priorities information types rescue operation roles Operational structure and organization October 2006 51 Issues of Dynamic Update Dynamicity and limited resources unstable availability Frequent updates increased communication needs consistency issues Need efficient metadata management to achieve ontology based update in this environment October 2006 52 26
Kinds of Dynamic Update - Overview (Vertical) Local update (Horizontal) Metadata Exchange (Horizontal) Ontology Based Between Entities Different level data dictionaries Metadata or Data Metadata Change or Append Both SDDDs Metadata Append SDDDs & KBs Both Both SDDD = Semantic Linked Distributed Data Dictionary. KB = Knowledge Base. October 2006 53 Outline Background Knowledge Management (KM) What is knowledge KM Processes Knowledge Management Systems and Knowledge Bases Ontologies What is an ontology Types of ontologies Use of ontologies in KM Ad-Hoc InfoWare (Example application) Ad-Hoc InfoWare and Approach Ontology Based Update Rescue Ontology Example October 2006 54 27
Example of Organization and Structure in Rescue Operations October 2006 55 Simple Model of Rescue Operation Roles October 2006 56 28
Upper Ontology for All Profiles October 2006 57 Information Profile and Example Information Priorities October 2006 58 29
User Profile October 2006 59 Example of DB Schema Information Profile: pr:informationprofile(pr:ipid, pr:item) pr:informationitem(pr:iid, pr:subject, pr:priority) pr:informationpriority(pr:iprid,...) UserProfile: pr:userprofile(pr:upid, pr:person, pr:role) pr:rescueoperationrole(pr:rorid, pr:roroletype, pr:reportsto, pr:responsibility, pr:ismemberof, pr:hasupdatepriority) pr:responsibility(pr:pid,...) pr:team(pr:tid,...) pr:person(pr:pid, pr:name,...) October 2006 60 30
Example of DB Content for User Profile October 2006 61 Rescue Scenario Timeline Populating the Knowledge Base Phase 1: initial population of knowledge base Phase 2: ontology individuals for current operation Phase 4: adjustments: changes and new arrivals October 2006 62 31
Handling Profile Ontologies in our Architecture Storage - who keeps what? Based on user role in rescue operation Each node keeps its own device profile and user profile Components Rescue ontology profiles Profile and Context Management Semantic Metadata and Ontology Framework Sharing and dynamic update Data Dictionary Manager Viewed as resources to be shared October 2006 63 Litterature M. Alavi and D. Leidner. Knowledge Management and Knowledge Management Systems: conceptual foundations and research issues; MISQuarterly Vol. 25 No.1, pp.107-136, March 2001. http://www.coba.usf.edu/departments/isds/faculty/abha tt/rm/alavi01-knowledgemanagement.pdf Deborah L. McGuinness. "Ontologies Come of Age". In Dieter Fensel, Jim Hendler, Henry Lieberman, and Wolfgang Wahlster, editors. Spinning the Semantic Web: Bringing the World Wide Web to Its Full Potential. MIT Press, 2002. http://www.ksl.stanford.edu/people/dlm/papers/ontolog ies-come-of-age-mit-press-(with-citation).htm October 2006 64 32