Title: A weighted-graph-based approach for diversifying search results
Authors: Guangzhi Qu, Hui Wu
Addresses: Computer Science and Engineering Department, Oakland University, Rochester, MI, 48309, USA. ' Computer Science and Engineering Department, Oakland University, Rochester, MI, 48309, USA
Abstract: Despite the explosive growth of internet usage, satisfying users| experience in web search while understanding the dynamic nature of search engine ranking remains a challenge. The main reason is that user queries and web documents may belong to different categories given a taxonomy of information. In this study, we attempt to detangle the ambiguity in web search by using a systematic approach. A weighted graph model is utilised to represent the complicated documents network from web search results. In this model, document topics are explicitly defined from the weighted document graph with an unsupervised learning method. On weighing each topic, both the size of the topic and the intra-topic document importance distribution are considered. In order to achieve web search results diversity, the reranking method proposed in this work leverages on the relevance of the documents to the query and related topics. Evaluation results on synthetic data and realistic web pages show the efficacy of a proposed system.
Keywords: search results; diversification; relevance; weighted graph; topics; clustering; web intelligence; web search engines; internet search; user queries; web documents; unsupervised learning; result ranking; document relevance; relevant documents; information retrieval.
DOI: 10.1504/IJKWI.2011.038626
International Journal of Knowledge and Web Intelligence, 2011 Vol.2 No.1, pp.15 - 35
Published online: 07 Mar 2015 *
Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article