Analysing the Next Generation of DKM system: preliminary steps

1
Analysing the Next Generation of DKM system
preliminary steps
Michele Nori SRA Division - ITC-irst nori_at_itc.it
- http//sra.itc.it
2
The Goal

• Given a set of system requirements
• supporting the representation of individual
  knowledge (locality, autonomy),
• supporting coordination processes among
  distributed individuals
• Analyse what a user of DKM system need.
• Compare different technologies solution.
• Define a set of requirements and functionalities
  for the next generation of DKM application.

3
The Problem
Requirements
Software architecture
4
How to overcome the obstacle?

• Experience
• Success stories
• Best practices
• Design Pattern
• Analysis pattern
• Architectural pattern

5
The Goal

• Theoretical concepts gt Technological concepts
• Create a success story for DKM domain
• Help the evolution of existing DKM system
• Identify the next generation of DKM application

6
Case Study DKM

• We focus on distributed applications and P2P
  systems for KM
• Three case study
• Bank/Insurance company
• Healthcare domain
• Market-Place

7
The approach

• Tropos analysis
• Agent oriented methodology
• analysis pattern

8
AO Modeling with Tropos
Early Requirements
Late Requirement
Implementation
Detailed Design
Architectural Design
Sub-system Actors
9
Extracting the P2P Virtual Community Pattern
Napster
Gnutella
GD3
Peer i
GD2
Napster Server
provide services
servent i
servent i1
discovery
GD1
GD1
search services
GD2
Peer i1
GD3
GD1
download MP3 files
GD2
get availabe MP3 list
GD3
get search result
10
Extracting the P2P Virtual Community Pattern
The P2P Virtual Community Pattern
access resources
being member of
provide resources
share resources
dynamic
Individual
Virtual P2P Comm.
constraints
being active
equal
preferences
decentralized
autonomous
providing all services
11
P2P Infrastructure Design
share service i
dynamic
JXTA architectural design
Peer Group i
security
decentralized
equality
Peer Group i
decentralized
monitoring
comm.infr.
dynamic
discovery
publishing
security
membership protocol
access protocol
12
Designing of a DKM application

• Apply the Virtual Community Pattern
• actors want to keep a high level of autonomy in
  managing its own resources
• there is a subset of the actors that have
  pair-wise dependencies in such a way that any
  actor is dependent on many, if not most, other
  actors in accessing relevant resources
• Design the P2P layer
• associate a Peer to each individual
• associate a PeerGroup to a virtual p2p community
• Design the Application
• specify which application services have to be
  provided by Peers and PeerGroup in order to
  achieve both the individuals goals and the
  Communitys ones

13
The approach

• Features based analysis
• User-centered
• Focus on technology
• Bottom-up approach
• More general

14
Features Based Approach

• Analyse the state of the art of P2P and
  distributed systems
• Understand what a user (end user or organization)
  of DKM applications need
• Integrate the research solutions
• Handle the constraints and necessity coming from
  software development

15
Features Based Approach
16
Evaluating P2P systems

• Next presentation Andrey Golub

17
Requirement analysis

• Functional Requirements KEx users testing
• Usability
• Functionalities
• Organizational aspects Healthcare and
  Market-Place
• Intentional dimension of the organizational
  setting for delivering KM solutions
• Business Requirements

18
Research solution

• Integration algorithm/solution as part of
  framework layer
• Research topic
• Information push
• Process Learning / Procedural Knowledge
• Software interfaces or services (Jini)
• Services description and workflow
• Web services EDXML, WSDL
• Federation services (RDF, Jxta group-service)
• FIPA ontology and protocol
• multi-agent systems integration

19
Implementation

• Integration/extension the actual KEx system
• Workspace paradigm
• Mapping learning
• Semantic query propagation
• Workgroup facilities (instant messaging, chat)
• Network topology representation
• Design of a services layer for DKM application
• Entities K-Peer, K-Federation
• Services Discovery, Propagation
• Algorithms Ctx matching, trusting model
• New applications

20
DKM Service Layer

• Entity
• Organization
• Coordination

• Process
• Algorithm

Application
Communication Infrastructure

• Protocol
• Message Structure

21
Application layer

• Entity
• Peer, Context, Federation
• Organisational settings
• Membership policy (open, closed)
• Community
• Coordination
• Competitive (auction) or cooperative (coalition)
  interaction protocol

22
DKM Service layer

• Process
• Workflow
• Meaning of core features
• Discovery social or semantic proximity
• Data structures
• Algorithm
• Context matching
• Trusted based

23
Future Works

• Complete the technology evaluation doc
• Relate the Intentional Model analysis to
  organisation and coordination reusable design
  pattern (methodology)
• Define a set of high level services and
  functionalities for the DKM domain
• Map services to technology solutions
• Create architectural pattern useful for the
  development phase