Improving web search personalization using Luhn-Inspired vector re-weighting
Hanze Liu · Memorial University Research Repository (Memorial University) · 2012
Web search personalization has been studied as a way to tailor Web search results to individual users based on their interests and preferences. Commonly, document and personalization profile features are stored in vector space models using measures such as term frequency (TF) and term frequency-inverse document frequency (TF*IDF). Inspired by Luhn's model of term importance, a novel approach is proposed in this thesis to identify and re-weight significant terms in the vector-based personalization models. Evaluations with a set of ambiguous queries illustrate that the order of the search results using this approach is superior to the TF approach and comparable to the TF*IDF approach. However, it is based only on the information stored in the personalization profiles, rather than requiring access to the distribution of each term across the document collection. As such, it can be applied more broadly when only limited information regarding the collection being searched is available.