METHOD AND DEVICE FOR FORMING RETRIEVAL INDEX BY COMMUNITY EXTRACTION

摘要

PROBLEM TO BE SOLVED: To form a retrieval index appropriate to general Web pages as well as blogs.;SOLUTION: The method comprises steps of: grouping queries having a strong correlation from a query log for clustering; labeling term groups of clusters; forming a route set based on the term of each cluster; forming a base set based on each node contained in the route set; performing community extraction based on a unique vector corresponding to a maximum unique value and a unique vector corresponding to a unique value other than the maximum unique value from the base set by HITS algorithm; extracting phrase from the authority and hub of the community extraction resu forming a directory based on the cluster, label, term and phrase; and forming a retrieval index based on the directory.;COPYRIGHT: (C)2008,JPO&INPIT