WIRF: A Web2.0 Based Information Retrieval Framework
Zhigui Jiang, Hui Zhang, Zeng, Lina, Rui Liu · ASME Press eBooks · 2010
With the rise of Web2.0, information retrieval is infused with novel energy and chance to develop. Traditional information retrieval methods like VSM model, PageRank link analysis model and so on did not take the Web2.0 features into consideration, and research on Web2.0 oriented information retrieval in recent years mainly concentrate on Tag Based information retrieval. Besides the tag information, this paper also mines the features of Web2.0 that can be brought into information retrieval like users' rating, and users' preference in profile. This paper proposes a Web2.0 based information retrieval framework named WIRF: In the first place, this paper proposes a topic semantic space based on LDA model, with which the semantic vector of user, query and document is calculated, and proposes TagRank and PersonalTagRank respectively. In the second place, with the cluster result based on the semantic vector of document, we smooth the information retrieval language model and propose a clusterbased language model—TLM(Topic Language Model). In the third place, this paper proposes a social information retrieval model named SocialRank which uses RankingSVM to integrate TagRank feature, TLMRank feature, traditional TFIDF feature and user rating feature. In the last place, this paper proposes a personalized information retrieval model named PersonalRank using the features of user rating and users' preference in profile by integrating the result of SocialRank.