PLSA-Based Personalized Information Retrieval with Network Regularization

Introduction

With the enormous volume of Web pages containing massive information on the Internet, the technology on the information retrieval resolves the problem that people how to find profitable information to some extent. Some search engines such as Baidu and Google have already made a great progress on the information retrieval. However, there also exist some problems. At first, these search engines ignore different requests of users instead of taking different strategy to diverse users. Moreover, the majority of queries submitted to search engines are short and ambiguous, which leads to a limited query expression. Even for the same query, different users might express different preferences over the retrieved documents. Thus, the search engines do not make clear meaning of user’s query intention (Jaime et al., 2010). The two reasons above lead to the decrease of the recall ratio and the precision ratio. To resolve the problem that user’s query intention is not understood by engines effectively, the scholars put forward to implementing personalized informational retrieval according to the characteristics of textual information. For example, some scholars design information retrieval model based on the query expansion (Claudio & Giovanni, 2012; Liu et al., 2010; Li et al., 2004), some leverage on user reviews and carry out personalized information retrieval (Chen & Chen, 2015; Chen et al., 2015) and some extract user interests and comprehend user intention by query history (Matthijs & Radlinski, 2011; Bennett et al., 2012;Zhang et al., 2013; Rami et al., 2013).