Transcript Document

Ontology Enabled Analysis and Interoperability
January 31, 2003
Elisa F. Kendall
CEO & Founder
[email protected]
(650) 960-2456
opportunity
The convergence of
 Internet technologies, XML
 dramatic advances in compute power, storage and network
availability and capacity
 a return to core competencies
 increasing globalization
have created an environment where collaboration within and across
organizations is no longer optional.
Examples:
 Mergers & acquisitions
 Collaborative design and development, concurrent engineering
 Just-in-time and outsourced manufacturing
 Service & operations
 Corporate business operations (HR and employee empowerment,
360º CRM)
 Bio and medical informatics
One of the greatest challenges facing CPC is interoperability.
Copyright ©2003 Sandpiper Software, Inc.
2
high tech manufacturing
Semiconductor
Manufacturers
Value Added
IP Vendors
(A,,C,D,Z,
uP
DSPs
Customer
Designs
(A,B,C,D,Z,
...
...
...
Standards,
Methods
New
Block
Common
Fabric
“Mix-and-match” of IP building blocks requires a sophisticated
collaborative engineering delivery system
Copyright ©2003 Sandpiper Software, Inc.
3
next generation collaboration
 Solutions need to reflect both extensive domain knowledge
and cross-domain, interdisciplinary interaction
 Interaction may include both
 dynamic and/or transitory relationships
 stable, long-term participation by individuals, organizations, and
resources
 Mandates automation of complex interactions among
interdependent activities
 Requires an understanding of the differences in business
rules and terminology and a mechanism for resolving
ambiguity and conflicts
Copyright ©2003 Sandpiper Software, Inc.
4
knowledge-based interoperability
Ontologies and ontology brokering technologies

can resolve ambiguity and conflicts among information
producers, consumers, and repositories

provide a rigorous, precise, and complete mechanism for
describing and maintaining shared IP

enable understanding of the distinctions in complex data,
business rules, and terminology

support communication that facilitates automation of
complex interactions among interdependent activities
Copyright ©2003 Sandpiper Software, Inc.
5
component-based methodology
 Need to move the bar forward on component-based vs. monolithic
ontology development to address scalability issues
 Layering and aligning ontology components facilitates scalability
 top-level, upper, fundamental meta knowledge definition layer
 foundation definition layer (e.g., industry standards)
 domain or “classic ontology” definition layer
 further refined (and distinct) interface, process, services, and role (user,
application, resource) definition layers
 UML-based approach supports component modeling strategy
 enables collaborative ontology development
 provides capabilities for maintenance, configuration management
 supports best practices in software engineering (documentation, test, reuse
and ontology sharing)
 begins to address scalability issues
Copyright ©2003 Sandpiper Software, Inc.
6
product roadmap
 Visual Ontology Modeler
–
–
–
–
–
Import/export of DAML+OIL, OWL, RDFS ontologies
Port to Rational XDE
Model analysis, merging, alignment capabilities
FOL expression editor
Import/export of KIF/CG based ontologies
 Knowledge Base Development Environment
– Automated generation/maintenance of object database for ontology
management, instance data management
– Interactive ontology browsing and analysis, merging, alignment,
verification motivated in part by Chimaera
– Instance data editor, with automated template construction from
ontologies
 Reference Ontologies and Library Development
– Semi-automated ontology development capabilities based on
keywords from text mining tools, database schemas, XML
schemas
– Additional ontology development based on ISO, industry standards
Copyright ©2003 Sandpiper Software, Inc.
7
product roadmap
 Context-driven Search
– Automatically generate complex queries for Verity, other search engines
from ontologies and instance data
– Provide context for text mining applications (e.g., inxight, Mohomine)
 Ontology Brokering Environment
– Concept resolution between ontologies, using techniques such as name
matching, synonym lookups, description matching, structural analysis,
statistical relevance and weighting algorithms that can be manipulated by
the user, and other algorithms, based in part on Anchor-PROMPT, work
done by USC/ISI, OBSERVER, and Chimaera
– Automated generation of transformations for use by other interoperability
components
– Inference capabilities for long term, more sophisticated concept matching
and resolution
– Query decomposition and planning based on InfoSleuth
– Results, explanation generation and justification through JTP, Vivomind
analogy engine
– Multi-agent framework for distributed ontology brokering – candidates
include Fujitsu’s java-based implementation of their April framework
Copyright ©2003 Sandpiper Software, Inc.
8
Medius® Visual Ontology Modeler
Rational Rose® 2002
Rose Enterprise Edition
Medius® UML
Ontology Library
Medius® Visual Ontology Modeler
KB Development Environment
Ontology
Knowledge Base
RDFS / DAML+OIL / OWL
Ontology Modeling Environment
 Top-level ontology components (e.g., IEEE SUO/SUMO)
 Meta-knowledge representation components (concepts from KIF, OKBC)
 Scientific notation data types and rules for conversion (SI Units, ISO Standards)
 Generates XML Schema, W3C Web Ontology Language (DAML+OIL, OWL)
 Domain ontologies based on customer requirements
Copyright ©2003 Sandpiper Software, Inc.
9
VOM capabilities
Features




Benefits
A distributed, multi-user development
environment in a rich, graphical
notation
A rich set of ontology authoring
wizards that create and maintain the
required UML model elements for the
user, saving time and substantially
reducing construction errors and
inconsistencies.
Automated export of XML schema,
RDFS / DAML+OIL
Integration with a commercial, scalable
object database that supports efficient
evaluation, comparison, validation, and
management of ontologies with OKBC
and CORBA access

Makes ontology modeling accessible to
subject matter (domain) experts and
others less skilled in knowledge
representation

Supports model configuration
management using industry standard
tools (e.g., ClearCase) for team
development of large ontologies
Automated model construction and
code generation saves time and
substantially reduces errors and
inconsistencies
Beta user productivity gain in DAML
generation and maint. was higher than
anticipated (e.g., ontology restructuring,
dependency mgmt., reuse, ontology use
across multiple KR paradigms)


Copyright ©2003 Sandpiper Software, Inc.
10
ontology broker capabilities
Deliver
–
Relevant results
–
From the user’s
perspective
–
–
In the user’s
vernacular
C++, XML, KIF
Enhance
M E D I U S®
Knowledge Brokering Suite
Intelligent
Agents
Resolution
Engine
Resolve
–
Workflow
Automation
–
Customizable
Agents subscription /
notification, data
fusion, vocabulary
enhancement
Ontologies
Support Services
solution planning, design, & implementation
Bridge and
Aggregate
–
Ambiguity and conflicts in
terms
–
Complex data and
rich metadata
–
Across applications and
organizations
–
Context and
Content
–
Cross-references,
complex relationships
Copyright ©2003 Sandpiper Software, Inc.
11
phase I broker environment
Medius
Ontology
Modeler
External
Ontologies
(DAML, OWL)
Ontology
Generation
Ontology
Analyzer
Ontology
Server
Information
Broker
Ontology
Storage
Reasoning
Engine
Query
and
Results
Knowledge
(Native
Format)
Bases
Phase 1 - Interactive Ontology Diagnosis and Reconciliation
Phase 1 Features
 Ontology Manager – Object-Oriented Database, CORBA Interface
 Ontology Analyzer – Ontology Rationalization and Term Resolution
 Ability to read generic DAML+OIL or XML from the Ontology Modeler
Copyright ©2003 Sandpiper Software, Inc.
12
phase II broker environment
Client
Domain
Manager
Agent-based
Communication
Information Broker
Ontology
Manager
Ontologies &
Maps
Service &
Interface Manager
Data Sources &
Problem Solving
Methods
Phase 2 – Static Ontology Alignment and Initial Service Layers
Phase





2 Features
Enhancements to the term resolution functionality
Generation of Static Ontology Alignment Maps
Domain Manager for resolution of incoming messages
Service / Interface Manager for interacting with Data Sources
Initial integration with Agent based communication
Copyright ©2003 Sandpiper Software, Inc.
13
phase III broker environment
Client
Agent
Communication
Process
Manager
Domain
Manager
Information Broker
Agent
Communication
Ontology
Manager
Agent
Communication
Ontologies &
Maps
Service
Manager
Interface
Manager
Data Sources &
Problem Solving
Methods
Phase 3 – Figure 6. Near Real-Time Ontology Reconciliation
Phase




3 Features
Further enhancements to the term resolution functionality
Dynamic term resolution in the Domain Manager
Process Manager for handling process driven interactions with PSL
Split Service and Interface Managers for improved interactions with Data
Sources
 Further usage of Agent based communication
Copyright ©2003 Sandpiper Software, Inc.
14
collaborative design
Component Design
Subsystem Design
M E D I U S
Knowledge Brokering Suite
Intelligent
Agents
– Prioritize & monitor
product portfolio
– Quicker decisionmaking due to availability
of relevant information
– Maximize reuse
– Leverage capabilities of
outsource partners, joint
ventures
Systems Integration
Engineering IT
Specialty Tools
Resolution
Engine
Ontologies
Design and Modeling
(CASE, MCAD, ECAD, etc.)
Productivity &
Communication
Support Services
solution planning, design, & implementation
Engineering Business
Applications
Foundation Infrastructure
Copyright ©2003 Sandpiper Software, Inc.
– Engineering
discipline-specific
– Process-related
– Company-specific
– Organizational
– Role / Function
– Language-specific
15
service & operations
M E D I U S
Knowledge Brokering Suite
– Device-independence
– Subscription / notification
– Goal-directed searching
– Timely, relevant, accurate
Hangar Maintenance
Intelligent
Agents
Line Maintenance
Resolution
Engine
Ontologies
Support Services
Component Shops
solution planning, design, & implementation
Planning & Scheduling
Maintenance Execution
– Aviation-specific
– FAA related
– Airline-specific
– Organizational
– Role / Function
– Language-specific
Engineering
Flight Operations
Copyright ©2003 Sandpiper Software, Inc.
16
unique capabilities

Addresses the need for people and applications to meaningfully share
information

Bridges incompatible resources, reducing integration development time
and limiting need for migration or intermediate storage

Delivers information dynamically as the user wants it

Aggregates information for content richness

Enhances vocabulary as context and usage changes

Platform, interface, and device independent – scaleable, flexible,
standards-based architecture

Leverages existing investments in UML

Focuses on content by creating multidimensional relationships and
interaction rules across critical business information

Enables development and deployment of intelligent agents capable of
subscription/notification, data fusion, application integration and other
complex interactions through an intelligent agent framework
Copyright ©2003 Sandpiper Software, Inc.
17