Transcript Achieving Semantic Interoperability through Controlled Annotations

Achieving Semantic Interoperability
through Controlled Annotations
Michael Gertz
Department of Computer Science
University of California, Davis
[email protected]
Interoperability —The Facts
 There is a high degree of physical connectivity on the Web,
i.e., in terms of exchanging bits and bytes among
heterogeneous information sources.
 There is a lack of logical connectivity, i.e., the meaningful
exchange, integration, and querying of data.
 Metadata are supposed to help adding more semantics to
data, but are not frequently used.
 What are appropriate metadata schemes that help users
and creators of Digital Libraries to create and utilize
metadata?
The Idea
 Build domain specific ontologies (richer than standard
vocabularies); ontologies contain concepts, terms,
definitions, and semantic relationships among concepts.
 Allow users to associate concepts from ontology with
data found on Web pages and sites, at different levels
of granularity.
 Such associations are called annotations, providing
meaningful and well-defined metadata.
 Relationships modeled in ontology are inherited to
annotated Web data (semantic browsing)
An Example
 Collection of Monkey Brain Images
Cell type B
terms
properties
relationships
spatial
temporal
semantic
Data as on the Web
Ontology
The Architecture
Java
Client
Browser
HTTP
Annotation
server
Ontology
repository
Annotation
server
HTTP
Web
HTTP
Repository
The Major Questions
 Are users and creators of Digital Libraries willing to
develop ontologies and use them to annotate data?
 How difficult is it to develop cross-language ontologies?
 What are necessary and useful components of a
respective infrastructure? Does XML provide a basis?
 What types of annotations do users want to make?
 What are behavior models users employ to make and
use annotations?
 Do annotations lead to the so-called semantic Web?