Concept-Based Retrieval And Interpretation For Large Data Sets

Abstract

Access to on-line information is growing at an exponential rate, fuelled by advances in computing and communications technologies. Current information retrieval methods are becoming ineffective due to their reliance on simple term-based methods, resulting in a massive number of matches, of which only a small proportion are relevant. We address this problem by developing new matching algorithms which understand the underlying meaning of documents in database repositories - by building semantic structures semi-automatically - and thus provide more relevant information to queries. This project will be of great benefit to a multitude of end-users in medicine, history, law and many other discipli..