Enhancing Knowledge Management Systems with Cognitive Agents

1
Enhancing Knowledge Management Systems with
Cognitive Agents
Journée de recherche de lAIM, France, 26 March
2004
Thierry NABETH, Albert A. Angehrn, Claudia Roda
INSEAD CALT The Centre for Advanced Learning
Technologies, Fontainebleau, France
2
Situation

• Many of the approaches proposed until now for
  managing knowledge were too narrow (knowledge
  only document) and often mainly driven by
  technologies. As a consequence many of them have
  failed.
• Yet, the Management of Knowledge still
  represents a critical challenge in all sectors
  (many pressures to do more with less) and at all
  the levels (organisation, group, individual).
• New approaches have appeared, promising to
  change this situation (learning networks,
  ontology, knowledge ecology, personalization,
  etc. and agents).

3
Objective of this presentation

• Objective of this presentation
• To present on how cognitive agents can help in
  the design of the next generation knowledge
  management systems better able to support the
  Knowledge intensive organizations at the
  organizational, group and individual level.

4
Structure of this presentation

• Analyse the reasons for the relative failure of
  Knowledge management system.
• Indicate the direction for the next generation
  knowledge management systems.
• Present the concept of the cognitive agent.
• Indicate how cognitive agent can help in the
  design of this next-generation knowledge
  management systems.
• Next steps

5
Analysis of the failure

• Why knowledge management has not (yet?) fulfilled
  the expectations

6
The reasons of the failure

• The Reasons of the failure
• A two narrow technological vision (lets create
  a big database it is just a matter of tools).
• A too shallow and passive support of the
  knowledge processes (KM should include support
  for K-exchange, K-use, K-stimulation, cultural
  transformation, etc.).
• A user not enough in control (empowerment?).
• An under-estimation of the importance of the
  Human factors (resistance to change, social
  dynamic aspects, etc).

7
Some Myths to be challenged

• Knowledge is only in the document. The perfect
  Knowledge Management system is a big database
  system that will have captured all the knowledge
  of the organization.
• Universality. The more general, powerful and
  complete the solution, the better (lets provide
  the maximum of functionalities).
• Social interaction spontaneously emerges once
  you have provided the adequate communication
  infrastructure.
• People are self motivated and are eager to adopt
  new processes if this help the organization to
  become more efficient.

8
A vision for better Knowledge Management Systems

• What are the needs,
• What are the solutions

9
A vision for the future (the needs)

• The needs
• Encompassing all the processes (identification,
  capture, acquisition, exchange, use, stimulation,
  etc.), knowledge sources (including tacit
  knowledge), and the different categories of
  users.
• A deep and active support of the knowledge
  processes (high level active cognitive objects
  reflecting the mental models of the users).
• The empowered the user (the user is in control),
  and taking into account his/her specificity.
• Addressing the social factors, managing the
  change (cultural transformation).

10
A vision for the future (some directions)

• Next Knowledge Management Systems some
  directions
• (1) provide tailored (personalised) support to
  the users (taking into account the specificity /
  context)
• (2) better address the management of tacit
  knowledge, and in particular the social aspects
  of knowledge exchange.
• (3) provide to the users with high-level
  (cognitive) interface and actively engage
  actively them in the dynamic of knowledge
  processes (stimulation).

11
Executing the vision

• The tools of the vision
• personalize the interaction in order to maximize
  the value / impact of this interaction. Reduce
  information overload.
• provide mechanisms supporting deeply the social
  and human dimension.
• use active mechanisms proactively engaging the
  individual group into knowledge activities.

12
The agents

• Defining the concepts

13
The agents (defining the concept)

• What is an agent
• Perception, autonomy, social ability,
  proactiveness
• The reasons for using agents
• Designing complex active applications
  (distributed control, integrated with human
  organization)
• Agent technologies approaches
• Middleware but also models have been developed
• Application of agents in Knowledge Management
• Automating search, mediation mechanisms etc.

14
The agents (definition)

• Characteristic of agenthood (Wooldridge
  Jennings)
• Perception of the environment
• Autonomy (self-direction)
• Social ability (capability to interact with other
  entities)
• Proactiveness (initiative)
• Other properties (not mandatory)
• Consciousness, intelligence, adaptability, etc.

15
The cognitive agents

• Cognitive agent Agent some properties of
  consciousness
• Belief
• Desires
• Intentions
• Believability
• Maintain a high level state of the environment
• Explicit semantic
• Why to use cognitive agents
• They are able to support more deeply the human
  process
• The concept they manipulate are more similar to
  the human concept.

16
The cognitive agents (architectures)

• Pioneer The SOAR system (Laird, Newel
  Rosenbloom)
• Very complex
• Rule based (mainly).
• New architectures have appeared
• ConAg, Boid, ICARUS, etc.
• Further develop agent brain model (BDI, etc.)
• Evolutivity/adaptability/learning sometime
  built-in (ecology of behaviours)
• Semantic web oriented (ontology, semantic
  network, )

17
Cognitive agents for Knowledge Management

• Or how they can contribute to design KM systems
  more personalised, socially aware and
  cognitiveactive

18
The cognitive agents for KM activities

• Social agents
• The role of these agents is to support the social
  dimension. (group level).
• Social translucence agents, facilitators,
  aggregators, etc.
• Process agents
• Support the knowledge management processes of the
  organization. (organisational level)
• Automating the tasks, etc.
• Personal agents
• Support the knowledge worker. (individual level)
• Supporting the individual

19
The cognitive agents for KM activities
(illustrations)

• Social agents
• EdComNet project (supporting learning network in
  municipalities for the citizen).
• Social translucence agents, group forming agents,
  etc.
• Process agents
• KInCA (Agent for transforming behaviours
  attitudes towards the knowledge sharing
  organization)
• Pedagogical agents intervening in the activity.
• Personal agents
• Ontologging
• Ontology-based User modelling and agents for KM.

20
Next steps

• Where are we now?
• Where are we heading to?

21
When will these mechanisms be available?

• Some of them already present (social
  translucence, social collaboration filtering,
  interface agents, etc.).
• Some other one are currently attracting a lot of
  attention today (personalization /
  contextualization in particular in e-learning,
  personal agents, ).
• Other are still in the Labs (emotional agents,
  agent for decision making, etc.).

22
Annexes

• Myths to be challenged
• The technical mechanisms

23
Annexes 1

• Some myths to be Challenged

24
Some Myths to be challenged

• Knowledge is only in the document. The perfect
  Knowledge Management system is a big database
  system that will have captured all the knowledge
  of the organization.
• Universality. The more general, powerful and
  complete the solution, the better. (lets provide
  the maximum of functionalities to every user).
• Social interaction spontaneously emerges once
  you have provided the adequate communication
  infrastructure.
• People are self motivated and are eager to adopt
  new processes if this help the organization to
  become more efficient.

25
The reality (knowledge only in the documents?)

• A very important amount of Knowledge is not (and
  never will be) present in documents.
• An important role of KM Systems should be to
  provide mechanisms that support the circulation
  exploitation of the tacit knowledge. (the
  ultimate objective of KM is that K is used, not
  that it is stored!).
• Note Why tacit knowledge will remain important?
• Because making the knowledge explicit is an
  heavy operation (expensive), which can hamper the
  flexibility of the organization.
• Because Knowledge can sometime be difficult to
  formalize and risks exist of overcoming the
  formalization of important pieces of knowledge.
• Because people are lazy, and capturing knowledge
  is often boring.

26
The reality (the more the better?)

• People are getting overwhelmed by information
  overload (think of email for instance).
• KM systems should not try to provide to all
  users every functionality, and to deliver all the
  knowledge that is potentially useful, but rather
  to provide the individuals with what they really
  need.
• KM systems should develop a very deep
  understanding of the user (including his
  cognitive style and his working context) in order
  to be able to deliver him relevant (according to
  his profile and context) knowledge and support to
  his work.

27
The reality (communication tools ?social
interaction ?)

• Many of the first generation virtual community
  systems (computer supported knowledge networks in
  which the tacit knowledge flows) have died due to
  the belief that the availability of communication
  tools (bulletin boards, etc.) was a sufficient
  condition for social interaction.
• The process of creating, growing maintaining
  virtual community systems is complex and involve
  many human factors. KM systems should explicitly
  address and support the social dynamic aspects
  (creation, growth maintenance).

28
The reality (people are self motivated?)

• Many People are satisfied by the status-quo.
  They only change their practices when they have
  no other choice, or at least after they have well
  evaluated the risks and have some guaranty that
  they will received a minimum of support in this
  transition.
• KM systems should actively help and stimulate
  the users in engaging in a continuous knowledge
  management process and exchange.
• Also, people are different and are in particular
  driven by different motives. Systems should take
  this into account.
• Note the limitation of this self motivation
  is in particular visible in the difficulty of
  making people to share their knowledge.

29
Annexes 2

• Executing the vision of designing personalized,
  socially aware
• and cognitive active
• Knowledge Management systems

30
Executing the vision

• The tools of the vision
• personalize the interaction in order to maximize
  the value / impact of this interaction. Reduce
  information overload.
• provide mechanisms supporting deeply the social
  and human dimension.
• use active mechanisms proactively engaging the
  individual group into knowledge activities.
• Tools are not enough. Provide methodology.

31
Personalising the interaction

• Taking into account the context.
• Take into account the characteristic of the users
  (role, users current working activities,
  preferences, cognitive style, etc.).
• Personalising the interactions according to the
  organizational context (priorities, goal
  orientations).
• Note Privacy issues
• However, better knowing people also brings the
  possibility to better serve them

32
Personalising the interaction (2)

• Technologies Researches conducted in this
  direction
• Key technologies used Ontology (modelling the
  user, the work context, the organization) and AI
  (matching)
• Standards e-Learning standards, HR-XML, identity
  representation, etc.
• Example of works in this direction
• Ontology-based KM, (deep user modelling via
  ontology)
• next generation e-learning systems able to take
  into account the context

33
Supporting the social dimension

• Key concepts means.
• Supporting different modes of Communication
  (synchronous, asynchronous).
• Social translucence Making the social activity
  visible (social pressure, trust building,
  motivation, etc).
• Deep support for the social processes
  facilitation, recommender opinion, group
  formation.
• Managing the cultural transformation (i.e.
  attitude transformation for sharing knowledge),
  incentives etc.

34
Supporting the social dimension (2)

• Technologies research
• Communication tools (mail, bulletin board chats,
  etc.), Virtual community systems.
• Social translucence navigation tools. Provide
  real-time indicators of the social activity (who
  contributes, who read, what are the knowledge
  element the most accessed, social network
  visualization, etc.). Analysing digital trace.
• Advanced coordination tools (technical or not
  technical). Example moderation, facilitation,
  structure, etc.
• Change management (transforming people attitude).

35
Deep Proactive support for knowledge activities

• Technologies research
• High-level (cognitive) knowledge objects.
  Ontology are used to help to manipulate concepts
  familiars to the one used by the knowledge worker
  (people, projects, topics etc.)
• Serendipity in context search (navigation).
  (versus search engines)
• Personal artificial agents that develop a deep
  understanding of the user and intervene. Stimulus
  agents, KInCA project (change management)

36
Deep Proactive support for knowledge activities
(2)

• Technologies research
• Agents that exploit the social traces to
  intervene proactively and stimulate the social
  activities.
• Electronic circulation folders (BSCW)
• Interactive experiences (role playing
  multi-users virtual reality)
• Etc.