Refining web search engine results using incremental clustering

In this article, we present a new solution to improve the Web search performance. Our algorithm is based on a new clustering algorithm that classifies the results of a query from a search engine into subgroups and assigns each group a short series of keywords together with some statistics data. Then, the user may look into the group with the keywords that he/she finds interesting. Compared with the approaches available in the literature, our algorithm does not require the number of groups as the prior knowledge; it starts from a single prototype group and adaptively expands the prototype set based on a self‐spawning splitting scheme until all the groups are finally identified. © 2004 Wiley Periodicals, Inc.