Diagnostic Knowledge Representation

1
Diagnostic KnowledgeRepresentation

• Practical Considerations

2
Presentation Outline

• Objectives
• Authors
• Knowledge Engineering Lifecycles
• Knowledge Representations
• Art of Optimization
• Open Research Issues
• Group Discussion

3
Objectives

• Present Hewlett-Packard experience in application
  of assorted knowledge representations for
  troubleshooting
• Share Ideas within Community of Practice
• Your ideas to complement ours

4
Authors

• Patricia CornwellKnowledge Management Technology
  Strategy
• Jaap SuermondtDecision Support Technology/BBN
• George FormanDecision Support Technology/MBR,RBR,
  DG
• Aparna SeetharamanDecision Support
  Technology/NN, GP
• Evan KirshenbaumDecision Support Technology/GP-GA

5
Knowledge Engineering Lifecycles

• Knowledge Acquisition
• Knowledge Maintenance Support
• Knowledge Utilization in Knowledge-Based Systems
• Knowledge Use in Knowledge-Based Systems

6
Knowledge Acquisition

• Domain Analysis Modeling
• Domain Architecture
• Acquisition
• Codification
• Import
• Leverage
• Discovery

7
Knowledge Maintenance Support

• Maintenance
• Inventory Management
• Improvements
• Quality Assurance
• Functional Enhancement
• Repurposing
• Utilization Support
• Asset Access/Selection
• Knowledge Views
• Usage Consulting

8
Knowledge Utilization in Knowledge-Based System

• Solution Engineering
• Solution Architecting
• Choice of Reasoning Method(s)
• Choice of Control Flow
• Knowledge Utilization
• Solution Requirements
• Knowledge Identification
• Knowledge Selection
• Knowledge Tailoring
• Knowledge Integration into Solution Architecture

9
Knowledge Use in Knowledge-Based System

• Application Use
• Feedback
• Usage Support
• Peers
• On-line Assistants
• Documentation
• Usage Support Staff

10
Knowledge Representations

• Natural Language
• NL within Topics Hierarchy or Keyword Sort
• Decision Graphs
• Rule Bases
• Case Bases
• Models
• Bayesian Belief Networks
• Neural Networks

11
KRR across a Lifecycle

• Knowledge Acquisition
• Ease of authoring/importing/leveraging
• Adequacy for representing knowledge
• Knowledge Maintenance Support
• Ease of modification
• Ease of maintaining consistency completeness
• Ease of verification validation

• Knowledge Utilization (by solution creators)
• Ease of identification selection
• Ease of tailoring
• Ease of integration
• Applicability
• Knowledge Use (by end users)
• Relevance
• Usefulness
• Match to task
• Understandability
• Performance/Expectations

12
Non-Engineering Issues

• Business Goals Strategy
• Knowledge Management Strategy
• Knowledge Management Process
• Organization Design
• Resources
• People
• Infrastructure
• Knowledge Sources

13
Art of Optimization

• Pick Critical Few Criteria (lt 10)
• Prioritize Identify What Counts as Success
• Experiment
• Hypothesis
• Metrics
• Use
• Data Analysis (with stakeholders, colleagues)
• Conclusions
• Repeat

14
Open Research Issues

• Authoring Technology
• Hybrid Knowledge Representations
• Multi-modal Reasoning
• Knowledge Representation Transformation
• Content-Based Search
• Knowledge Base Constraint Maintenance

15
Group Discussion

• Share Ideas within Community of Practice
• Skillset for
• capturing diagnostic knowledge in AI knowledge
  representations
• designing appropriate reasoning engines
• maintaining knowledge bases
• utilizing knowledge bases reasoning engines
• Adequacy of KR for diagnosis
• Use by customer support engineer
• Use by IT engineer
• Use by help desk assistant
• Use by end user

16
(No Transcript)