On the basis of the ideas presented in this paper, we plan to undertake the reengineering of AGROVOC into an ontology system that will also serve the functions of a traditional thesaurus. This involves creating a complete inventory of domain-relevant entity types and relationship types, which we will base on further analysis of AGROVOC and related vocabularies and on existing inventories, such as the UMLS Semantic Network. We tend towards using a frame representation of relationships, as presented in Slaughter and Soergel (2003).
Based on the encouraging results of our exploration of the applicability of the rules-as-you-go approach to AGROVOC, we plan to develop a system that streamlines the reengineering process. We expect that this system will drastically reduce the effort required and thus make the reengineering effort feasible.
The resulting knowledge base will enable
improved user interaction with the vocabulary for improved query formulation and more user learning about the domain;
intelligent behind-the-scenes support for query expansion;
intelligent support for human indexers and automated indexing/categorization systems;
support for artificial intelligence and semantic Web applications for agriculture, food processing, and food safety.
The approach will be incremental, starting with providing a proof of concept through pilot application that demonstrates that the expected benefits will in fact result. This will provide the basis for planning and efficient implementation of the large-scale effort to reengineer AGROVOC into a full-fledged ontology. We hope this effort can be carried out in a cooperative, distributed, and coordinated environment to promote active participation and ultimately use of a widely accepted Knowledge Organization System for the domain of food and agriculture.