Exploring folksonomies for adaptive query expansion

20th Conference on User Modeling, Adaptation, and Personalization

Fabio Gasparetti

Others  :  http://ceur-ws.org/Vol-872/sasweb2012_invitedtalk_3.pdf
PID  :  46106

来源: CEUR

【 摘 要 】

Web is growing at an astonishing rate, making it necessaryto devise effective methods for helping users find what they are looking for. Adaptive query expansion (QE) allows users to better define their search domain by supplementing the original query with additional terms related to their preferences and information needs [3]. While Automatic QE has a long history in the information retrieval com- munity, there are a very few attempts to make this technique adaptive to user current needs (e.g., [6, 1, 7]). Most of the Automatic QE approaches are based on the following assumption: documents classified higher by an initial search contain many useful terms that can help discriminate relevant documents from irrelevant ones. Despite the large number of studies, a crucial issue is that the expansion terms identified through traditional methodologies from the pseudo-relevant documents may not be all useful [2]. One of the failure reasons of the query expansion has been identified in the lack of relevant documents in the local collection. Consequently, some works advance the use of an external resource for query expansion in order to improve the effectiveness of query expan- sion, such as thesaurus (e.g., WordNet) [4], Wikipedia [8], and search engine query logs [5]. Recently, several authors have focused on social annotations in various tasks, largely motivated by their increasing avail- ability through many Web-based applications. The proposed approach1 is an extension of the traditional QE techniques, which rely on the computation of two-dimensional co-occurrence matri- ces. Our system makes use of three-dimensional co-occurrence matrices, where the added dimension is represented by semantic classes (i.e., cat- egories comprising all the terms that share a semantic property) related to the folksonomy extracted from social bookmarking services such as delicious, Digg, and StumbleUpon. The whole procedure of adaptation is completely transparent to the user, as it takes place in an implicit way based on his profile. The user profile is created and dynamically updated using the information related to visited pages and corresponding search queries. The system analyzes the