Mining query log graphs towards a query folksonomy

The human interaction through the web generates both implicit and explicit knowledge. An example of an implicit contribution is searching, as people contribute with their knowledge by clicking on retrieved documents. When this information is available, an important and interesting challenge is to extract relations from query logs, and, in particular, semantic relations between queries and their terms. In this paper we present and discuss results on query contextualization through the association of tags to queries, i.e., query folksonomies. Note that tags may not even occur within the query. Our results rely on the analysis of large query log induced graphs, namely click induced graphs. Results obtained with real data show that the inferred query folksonomy provide interesting insights both on semantic relations among queries and on web users intent.