V. DISCUSSION AND CONCLUSION
In this paper, a semantic information retrieval framework to improve the precision of search results by concentrating on the context of concepts is presented. Instead of keywords matching technique, the RDF triples is used. Document annotation is represented as an extension ontology and store them in a separate relational database. The triple searching and semantic matching is performed by the inference engine and results are passed to the ranker to sort them according to their relevancy to user'ร queries. In the current framework we focused on academic theses. Our near future is currently focusing on the aspect of document annotation. Current annotation is purely based on exact match by referring to the labels of each instances stored in the KB. We look into the possibility of doing document annotation by means inexact match or contextual term matching.