Capturing and Exploiting Semantic Relationships for Information and 
Knowledge Management

Yolanda Gil
Varun Ratnakar
Timothy Chklovski

The EXPECT project
USC /Information Sciences Institute



TRELLIS is an interactive environment that allows users to add their observations, opinions, and conclusions as they analyze information by making semantic annotations to documents and other on-line resources. This is in essence a knowledge acquisition problem, where the user is adding new knowledge to the system based on their expertise as they analyze information. The TRELLIS project is described in detail here .

TRELLIS includes a vocabulary for semantic annotations of decisions and tradeoffs.  The initial version of this vocabulary is now available as a schema/ontology in XMLS, RDFS, and DAML+OIL. TRELLIS allows users to extend this vocabulary and its corresponding schemas.


People have used TRELLIS to annotate tradeoffs and decisions (e.g., travel), organizing materials (e.g., search results), analyzing disagreements and controversies on a topic (e.g., political debates), handling incomplete information (e.g., genealogy research), etc. TRELLIS generates these annotations in XML, RDF, and DAML+OIL according to the schemas and ontologies for the TRELLIS annotation vocabulary.  Users can extend the vocabulary using the TRELLIS interface.

The second version of TRELLIS was released in January 2002, and can be used from a Web browser. This version is still under development.
Documentation is available here .

  • Comparison of semantic markup languages : in order to analyze the features and tradeoffs of existing markup languages, we generated a table comparing XML Schema, RDF Schema, and DAML+OIL that is linked to examples of use of each feature.
  • Tutorial slides on semantic markup languages: the features of several markup languages are highlighted in tutorial slide presentations for XML , RDF , DAML .
  • A vocabulary to annotate decisions and tradeoffs (ongoing work)
  • Annotating the design rationale of knowledge bases, to create a new generation of "Resilient Hyper Knowledge Bases" (RHKBs).

Please send any questions or comments to Varun Ratnakar .