Friday, February 29, 2008

A semantic marathon!


Concepts in AGROVOC are all interrelated owing to the multidisciplinary nature of the subjects it addresses. Fixing them in rigid hierarchies is by no means totally convincing. Nevertheless, some kind of mapping concepts into a more semantic structure is required in the wake of new technologies.

Before embarking on the task, some hierarchies in AGROVOC are studied to feel the tips of the iceberg.
After this, hierarchies of each top term, i.e. terms without ant BT relation are studied alphabetically.

Task to be accomplished

The AGROVOC Thesaurus is currently structured as a traditional thesaurus and is stored in a relational database. It is the foundation for a new structure, more semantically organized, that FAO is planning to release, called the Concept Server, CS. Within the framework of the conversion to a concept server, all these traditional thesaurus relationships need to be refined. This revision primarily includes:

1) Revision and refinement of all BT/NT relationships and identification of real hierarchical relationships which can be represented as ‘suclass of” / “is_a” or as “is an instance of”. (28,625 BTs and 28,617 NTs)
2) Revision and refinement of all RT relationships. (28,219 RTs)
3) Revision of scientific terms and corresponding common names.
4) Revision of all terms which do not have broader terms and identify those that can to be assigned to one or more broader terms.
5) Revision and refinement of USE/UF/UF+ relationships. (19,158 Use, 17,386 UF and 1,775 UF+)

Friday, February 22, 2008

We started this blog to share our on going work on Agrovoc Revision and Refinement entrusted to ICRISAT by FAO. We seek your suggestions and support in restructuring Agrovoc so as to make it more amenable for semantic searching.