Evaluating Clustering Techniques for Collaborative Querying
Chandrani Sinha Ray, Dion Hoe‐Lian Goh, Schubert Shou-Boon Foo, Nyein Chan, Soe Win · 2006
Collaborative querying is a technique which makes use of past users' search experiences in order to help the current user formulate an appropriate query. In this technique, related queries are extracted from query logs and clustered. Queries from these clusters that are related to the user's query are then recommended. This work compares a range of similarity measures and features (e.g. titles, URLs, snippets) used for clustering, and also proposes an extended K-means clustering algorithm. Experimental results reveal that the best clusters are obtained by using a combination of these sources rather than using only query terms or only result URLs alone.