摘要 |
PROBLEM TO BE SOLVED: To contribute to improvement in retrieval efficiency of a user by providing a recommended query of high precision for the user. SOLUTION: An extraction part 3 sets position information of a query linked with a clicked URL in sequence and calculates the importance of the URL when extracting a retrieval log of each query linked with the URL from a storage part 2. A graph generation part 4 generates a click graph based upon the extracted retrieval log. A feature quantity extraction part 5 calculates a feature quantity between queries of the click graph. A graph generation part 6 generates a skip graph based upon the extracted retrieval log. A feature quantity extraction part 7 calculates a feature quantity between queries of the skip graph by using the distance between queries based upon the position information and the importance of the URL. A feature quantity composition part 8 calculates a feature quantity between queries by putting together the feature quantities calculated by the parts 5 and 7. A recommended query is presented to the user according to the feature quantity. COPYRIGHT: (C)2012,JPO&INPIT |