Personalized information retrieval in digital ecosystems
Dengya Zhu, Heinz Dreher · 2008
Search results personalization is considered a promising approach to boost the quality of text retrieval. In this paper, a personalized information retrieval paradigm is proposed which not only implicitly creates user profile by learning users’ search history, search preferences, and desktop information by kNN algorithm; but also intends to deal with the problem of search concepts drift through adjusting the weight of category which represents users’ search preference. By comparing the cosine similarities between vectors represent personal valued search concepts in user profiles, and vectors represent search concepts in the retrieved search results, the search results will be tailed to better match users’ information needs.