Constraint-Based Query Clustering.

Summary The analysis of query logs over a search engine provides valuable information on the user needs. Query clustering allows for the identification of similar queries and forms the basis for recommendations and query personalization. In this study, we propose constraint-based query clustering to enhance the query grouping process: Prior knowledge about the similarity of given queries is used to guide the formation of clusters, thus resulting in more homogeneous and reliable groups. We apply the constraint-based clustering algorithm C-DBSCAN [10] on query logs and show that constraints improve cluster quality. [ABSTRACT FROM AUTHOR]