A3CRank: An adaptive ranking method based on connectivity, content and click-through data

Abstract

Due to the proliferation and abundance of information on the web, ranking algorithms play an important role in web search. Currently, there are some ranking algorithms based on content and connectivity such as PageRank and BM25. Unfortunately, these algorithms have low precision and are not always satisfying for users. In this paper, we propose an adaptive method based on the content, connectivity and click-through data triple, called A3CRank. The aggregation idea of meta search engines has been used to aggregate ranking algorithms such as PageRank, BM25, TF-IDF. We have used reinforcement learning to incorporate user behavior and find a measure of user satisfaction for each ranking algorithm. Furthermore, OWA, an aggregation operator is used for merging the results of the various ranking algorithms. A3CRank adapts itself with user needs and makes use of user clicks to aggregate the results of ranking algorithms. A3Crank is designed to overcome some of the shortcomings of existing ranking algorithms by combining them together and producing an overall better ranking criterion. Experimental results indicate that A3CRank outperforms all other single ranking algorithms in P@n and NDCG measures. We have used 130 queries on University of California at Berkeley's web to train and evaluate our method.

An ant-colony based approach for real-time implicit collaborative information seeking

“…Normally, evaluating an IR system requires experimental sets containing queries, documents, and relevance judgments; however, building such collections requires a significant amount of work (in other words, data on queries and judgments). Thus, in many recent studies ( Gayo-Avello & Brenes, 2009;Joachims, 2003;Jung, Herlocker, & Webster, 2007;Liu, Fu, Zhang, Ma, & Ru, 2007;Zareh Bidoki et al, 2010 ), click-through data were employed to evaluate search engines' performance. The concept is simple: employ clicks as relevance judgments, assuming that a user evaluates a result as relevant if it is chosen among the search results related to a query.…”

“…The majority of studies in the implicit area are based on collaborative querying techniques that upgrade information systems with data on past query preferences related to other users. As recently demonstrated ( Yue, Han, He, & Jiang, 2014 ), such studies primarily tested implicit collaborative information-seeking systems using simulated query formulation instead of employing user analysis involving human participants. In our research, we employed a classic approach by using two existing datasets to simulate queries to evaluate our system in a real setting.…”

“…Ranking results of informational and transactional queries is another matter: given the high frequency of Web pages' updates and the ever-increasing need to obtain answers in real time, the World Wide Web hyperlinks' configuration is no longer the only effective relevance measure that users assign to Web pages. Thus, devising new relevance indicators -to be placed alongside the existing ones (in other words, those based on Link Analysis Ranking) -with the goal of further improving the ranking by considering other relevance measures valued by users is necessary ( Zareh Bidoki, Ghodsnia, Yazdani, & Oroumchian, 2010 ).…”

A study of relevance propagation in large topic ontologies

“…For instance, Chibane and Doan (2007) use a traditional model of information retrieval based on content and link similarity to propagate relevance through hyperlinks. In a similar way, Bidoki, Ghodsnia, Yazdani, and Oroumchian (2010) propose a content-and link-based relevance propagation model, which is iteratively enriched by information from the user's behavior. Another scheme to compute the topical authoritativeness of a web page is presented by Dai, Davison, and Wang (2010).…”