PERSONALIZED WEB SEARCH USING KEY PHRASE-BASED USER PROFILES

In the context of the personalized web, the structure of user profiles remains a challenge due to its effect on information retrieval efficiency. In creating a user profile, how to extract collections of metadata of documents is a significant point. Although the efficiency of the structures such as keyword-based has proven, there is not research on key phrase based user profiles in the process of personalization. Previous methods to create user profiles emphasize keyword extraction while ignoring the existence of key phrases in the documents. This article proposes the use of keyphrase-based user profiles in personalized web search. We investigate the state of the art keyphrase extraction algorithms by considering different models of supervised and unsupervised methods. This can overcome the problems of missing keyphrases in user profiles and increase the accuracy of information retrieval. Finally, we evaluate keyphrase-based user profiles using a re-ranking algorithm to complete the process of personalization using different datasets. Personalized models based on supervised keyphrase extraction approaches obtained more accuracy by 7% than unsupervised approaches.