PlusDBG: Web community extraction scheme improving both precision and pseudo-recall

Abstract

This paper proposes PlusDBG to improve both precision and pseudo-recall by extending the conventional Web community extraction scheme. Precision is defined as the percentage of relevant Web pages extracted as members of Web communities and pseudo-recall is defined as the sum of the number of relevant Web pages extracted as members of Web communities. The proposed scheme adopts the new distance parameter defined by the relevance between a Web page and a Web community, and extracts the Web community with higher precision and pseudo-recall. Moreover, we have implemented and evaluated the proposed scheme. Our results confirm that the proposed scheme is able to extract about 3.2-fold larger numbers of members of Web communities than the conventional scheme, while maintaining equivalent precision. © Springer-Verlag Berlin Heidelberg 2005.