DAML PI Meeting Status Briefing

OntoAgents
Information Foodchain

OntoMat – the Annotation Tool
Authoring and Annotation Tool with support for DAML+OIL
Downloadable from http://annotation.semanticweb.org
Component-Framework:
DB-Reverse Plugin
Ontology-Focused-Crawler Plugin

TRIPLE – the Inference System
Inference Engine designed to handle multiple modeling languages with different semantics
Based on Horn logic with default negation
ASCII & RDF/DAML based syntax
Support for RDF and DAML+OIL (Namespaces, RDF Models)
Part of the RuleML initiative
Implementation based on XSB downloadable from http://triple.semanticweb.org

Ontology Articulation
and Transformation
InterDataWorking (structuring of data model transformations)
Layered approach, based on small reusable transformation components
Example transforms UML/XMI to RDF Schema
Similarity Flooding
determining correspondences in two ontologies
Pure algorithmic approach to ontology articulation
Works for graph-based datamodels
(DAML+OIL, XML Schema)
Interoperation among autonomous sources
Deal with stovepipes, within, outside of DoD
Example: DoD, various commercial transport services
Approach to interoperation among sources that employ     semantically heterogeneous ontologies
Tools to create articulation ontologies, based on dictionary etc. knowledge

OntoWebber – Ontology-driven
Portal Design Method & Tools
Tool-supported ontology-driven website design and creation methodology
Modeling and compilation approach
Heterogeneous-data integration & personalization
Separation of portal perspectives

OntoWebber –Ontology-driven
Portal Design Method & Tools
Facilitates ontologies for the domain as well as for website design (Navigation, Content, Presentation)
Shared infrastructure
Integrity constraints over website structure as well as the domain

Planned Activities
OntoMat
Start of SWOBIS, self-updating list of software tools for the semantic web. In collaboration with ISI (web scripter) (see http://tools.semanticweb.org)
Collaboration with Teknowledge. Integration of OntoMat with Teknowledge markup tools
Microsoft Word markup
Extension to Authoring Tool
Integration of information extraction components
Annotation Learning
Text re-recognition
PDF & Picture Annotation
OntoWebber
Larger case-study involving SemanticWeb.org
Fine Tuning of the representation ontologies and modeling tools

Planned Activities
Ontology Articulation
Integration with OntoWebber
Combination of Articulation and Similarity Flooding
Triple
development of an efficient inference core (pure Java)
Extend to mediation system
for integration with Web Services
Extended knowledge representation and query facilities for simplifying the tasks of specifying semantics
explicit negation, path expressions)
Development of different semantics modules and transformation facilities
UML to DAML+OIL

Metrics
OntoMat – the Annotation Tool
Cost of manual annotation (time required to annotate information resources)
Cost of semi-automatic annotation (trained on a given document corpus)
OntoWebber – Website development
Cost of conventional website development and maintenance (80%!)
Triple – Inference Engine
Cost of building and integrating query and inference facilities for specific data models
Ontology Articulation
Cost of manually determining the correspondences among ontologies
Evaluation involved user studies
Challenge:   So far no standard benchmarks in sight
Needed to assure scalability to relevant military problems
Help the scientific community with assessing research progress

Summary
Information Foodchain is the central notation
Metadata Creation (Annotation Editor)
Middleware (Inference Engine, Ontology Articulation)
Application (Portal Generation with OntoWebber)
Planned Activities include deployment and extension