Improving search engines with open Web-based SKOS vocabularies

Abstract

The volume of digital information is increasingly larger and even though organiza-tions are making more of this information available, without the proper tools users have great difficulties in retrieving documents about subjects of interest. Good infor-mation retrieval mechanisms are crucial for answering user information needs. Nowadays, search engines are unavoidable - they are an essential feature in docu-ment management systems. However, achieving good relevancy is a difficult problem particularly when dealing with specific technical domains where vocabulary mismatch problems can be prejudicial. Numerous research works found that exploiting the lexi-cal or semantic relations of terms in a collection attenuates this problem. In this dissertation, we aim to improve search results and user experience by inves-tigating the use of potentially connected Web vocabularies in information retrieval en-gines. In the context of open Web-based SKOS vocabularies we propose a query expan-sion framework implemented in a widely used IR system (Lucene/Solr), and evaluated using standard IR evaluation datasets. The components described in this thesis were applied in the development of a new search system that was integrated with a rapid applications development tool in the context of an internship at Quidgest S.A.