Whither KBE An informal view

1
Whither KBE?(An informal view)

• September 2005

Ismail Deif OMG Meetings Atlanta
2
Why a Knowledge-Based System?

• Goals of Knowledge-Based Systems
• Getting the computer to do more
• Automating tasks
• Applying more complex reasoning
• Developing higher-level applications
• Using domain vocabularies and ontologies
• Applying higher-level concepts
• Dealing with what and why
• Focusing on results rather than algorithms
• Using richer representations
• Knowledge models that can grow and evolve over
  time
• Applying intelligence
• Adapting to new situations
• Evolution
• Constraint satisfaction and case-based reasoning
• Ease of application development

3
What is a Knowledge-Based System?

• Characteristics of Knowledge-Based Systems
• Explicit representation of knowledge
• Explicit reasoning about knowledge
• Knowledge repository and inference (reasoning)
  engines
• Separation of the knowledge repository from the
  engines
• Built-in inferencing algorithms
• Formal knowledge representation, usually through
  language

4
Traditional KBE

• A general modeling capability for engineering
  purposes
• Used for design space exploration
• Can Model Function
• Items existence based on function it performs
• Configuration states what is in the model
• Can Model Behaviour
• Expected outcomes and outputs of a component
• Interactions between components
• Can Model Form and Fit
• Geometric and assembly representations model form
  and fit
• Function and behaviour can affect form, and
  vice-versa
• Propagates design decisions forward and maintains
  design consistency
• In a broad sense, TMS and/or constraint
  satisfaction

5
Where is KBE?

• Today, KBE has been co-opted by CAD
• Geometry augmented with knowledge
• Equations, Relations and Constraints
• Traditional parametrics
• Geometry-centric approach
• Does not extend to PLM
• Difficult to extend to general engineering
  problems
• Difficult to remove/replace components of the
  design
• Parametrics and knowledge must be synchronized
• To traditional KBE, this is backwards
• KBE for PLM requires a different approach
• Knowledge-centric rather than geometry-centric
• Extends to PLM naturally
• Allows full expressiveness
• Geometry is a side-product of the
  design/engineering process

6
Whither KBE? Optimistic view

• KBE will move from the tool level to the PLM
  level
• This is a necessity
• This is the true scope of traditional KBE
• KBE will need to be extended to model such
  concepts as
• Requirements
• Processes
• Products
• Resources
• Manufacturing processes and capabilities
• Documentation
• Post-sales support
• Disassembly and disposal
• Full lifecycle concepts

7
Wither KBE? Pessimistic view

• KBEs original moment has passed
• It was never fully embraced
• Traditional KBE revolutionary, geometry-centric
  KBE evolutionary
• CAD companies resisted, then co-opted
• Industry is not ready for it
• Social pressures (in the short run, jobs may be
  eliminated)
• Knowledge languages are not easy enough
• Traditional approaches continue to dominate ( ex.
  Some people still use 2D today)
• Economic drivers are missing (mass customization
  not here yet)
• New generation of KBE will be developed, but
• What will it look like?
• How long will it take?

8
Proposal Knowledge-based PLM

• In Knowledge-based PLM, the system is concerned
  with applying explicit knowledge of Product,
  Process and Resource to the tasks of Design and
  Manufacturing
• And The system is also concerned with managing
  that explicit knowledge itself, through the
  following tasks
• Capturing and transforming the knowledge
• Storing and retrieving the knowledge
• Maintaining the knowledge
• Growing and evolving the knowledge
• Leveraging the knowledge as a living asset
• All of the above is done in addition to managing
  the actual Product, Process and Resource data and
  information

Knowledge Management
9
KBP (1)

• Direct Intellectual Property Modelling and
  Management
• Models
• Knowledge product models extend beyond geometric
  representation to fully support Functional,
  Logical and Physical aspects in integrated
  representation
• Knowledge process models manage the
  transformations of R-L-F-P
• Knowledge process model interactions between
  Enterprise processes and Design and Manufacturing
  processes
• Knowledge
• Applied to intelligent process scheduling,
  execution and monitoring
• Product, Process (enterprise and manufacturing),
  Resource knowledge models fully interplay
• Management
• Management of knowledge components
• Maintaining interrelationships among them

10
KBP (2)

• Automation
• Support for automation at knowledge model level
• Automation integrated with process execution and
  monitoring
• Support for multiple paradigms
• Procedural
• Opportunistic (black board)
• Scheduled
• Dynamic
• Complete audit trail capabilities

11
KBP (3)

• Flexible PLM System
• Easy to build applications
• Domain-focused ontologies
• Specialized high-level language(s)
• Easier to use than traditional programming
  environments
• Robust and complete, operating at a high
  reasoning level
• Fully extensible, able to accommodate the needs
  of most engineering and business users
• Open
• Easy to develop and apply best practices
• Working in user domain
• Apply knowledge management principles to best
  practices
• Easy to extend and customize
• By customers
• Can extend meta-models and develop ontologies
• Multiple levels of customization Business Rules
  to Knowledge models

12
KBP (4)

• True integration of Product, Process and Resource
• Unified model with multiple inference engines
• Cross-engine interaction
• True collaboration
• At the knowledge model level
• Users collaborate on concepts
• Software collaborates on execution
• Support for distributed modelling
• Extended Range of Knowledge Tools
• Cover more of the Knowledge tasks
• Discovery, elicitation, capture, search, re-use,
  etc.
• Extend to more functions of the enterprise

13
Target
Industry or Sector
Space for Knowledge-based PLM
Services
Utilities
Insurance
Government
Financial
Capture, Validation and Representation
Structuring, Formalizing and Modeling
Distribution, Sharing and Collaboration
Execution and Info Generation
Pharmaceutical and Biotech
Storage and Repositories
Elicitation and Acquisition
Search and Retrieval
Discovery and Creation
Management
Manufacturing
Re-use
Knowledge Function
RD
Design or Planning
Manufacturing or Production
Management and HR
KBP
IT and Infrastructure
Current
Finance and Resources
Future
Sales and Marketing
Service and Support
Organizational Function
Space for Knowledge-based PLM
14
Thank You