Contextual Search Using Ontology-Based User Profiles

Search engines, generally, return results without any regard for the concepts in which the user is interested. In this paper, we present our approach to personalizing search engines using ontology based contextual profiles. In contrast to long-term user profiles, we construct contextual user profiles that capture what the user is working on at the time they conduct a search. These profiles are used to personalize the search results to suit the information needs of the user at a particular instant of time. We present the results of experiments evaluating the effect of the original versus conceptual ranking and the use of multiple sources of information to build the contextual profile. We were able to achieve a 15% improvement over Google in the average rank of the result clicked by a user when contextual information extracted from open Word documents and Web pages was used to re-rank the results.