Characterizing browsing strategies in the World-Wide Web
Proceedings of the Third International World-Wide Web conference on Technology, tools and applications
Query Expansion by Mining User Logs
IEEE Transactions on Knowledge and Data Engineering
Implementation of the SMART Information Retrieval System
Implementation of the SMART Information Retrieval System
A novel log-based relevance feedback technique in content-based image retrieval
Proceedings of the 12th annual ACM international conference on Multimedia
Query chains: learning to rank from implicit feedback
Proceedings of the eleventh ACM SIGKDD international conference on Knowledge discovery in data mining
Defining a session on Web search engines: Research Articles
Journal of the American Society for Information Science and Technology
Beyond the session timeout: automatic hierarchical segmentation of search topics in query logs
Proceedings of the 17th ACM conference on Information and knowledge management
Automatically adapting the context of an intranet query
FDIA'08 Proceedings of the 2nd BCS IRSG conference on Future Directions in Information Access
Hi-index | 0.00 |
We make use of search logs provided by the Belga News Agency to recommend images downloaded by previous users to new users. Each search session in the logs consists of a session ID number, the ID of the images which were downloaded at the conclusion of that session, and the various search terms which were input leading up to the selection and downloading of those images. In our approach, we match the queries of future users against the search terms in each session of the logs, and return the images selected in the best matching search sessions. In this way images considered relevant by previous users are recommended to future users with similar queries. An evaluation using P@50 for ten common queries produced encouraging results. This work describes a variation on the traditional Information Retrieval paradigm, where instead of text documents or images being indexed according to their content, they are indexed according to the search terms previous users have used in finding them.