Ontological User Profiling for Precision-Oriented Information Retrieval

Description

While search engines are becoming the predominant means of discovering and accessing information on the Internet, there are limitations of these general purpose search tools. One of the problems of the existing search engines is that a given query will always lead to the same results returned no matter who the information seeker is. Consequently, information seekers often need to spend considerable amount of time to identify the really relevant documents from the result set. Another problem of the existing search engines is that too much emphasis has been put on using document popularity as the basis to estimate the relevance of Web documents. Unfortunately, the correlation between popularity and relevance could be very weak for newly created Web documents. The project aims at developing an ontology-based meta-search engine to enhance precision-oriented information retrieval. In particular, fuzzy ontology is proposed to capture a semantically rich representation of an information seeker's specific information needs. Thereby, the querying process can be customized for each individual's specific interests. A fuzzy ontology mining algorithm will be developed to automatically extract the representation of the information seeker's interests from a set of browsed documents. In addition, an ontology-based document re-ranking function will be developed to augment the popularity-based ranking function used by popular search engines. This research opens the door to the development of the next generation of meta-search engine to alleviate the problem of information overload.