A Search Engine based on Query Logs and Search Log Analysis at the University of Sunderland

This work describes a variation on the traditional Information Retrieval paradigm, where instead of text documents being indexed according to their content, they are indexed according to the search terms previous users have used in finding them. We determine the effectiveness of this approach by indexing a sample of query logs from the European Library, and describe its usefulness for multilingual searching. In our analysis of the search logs, we determine the language of the past queries automatically, and annotate the search logs accordingly. From this information, we derive matrices to show that a) users tend to persist with the same query language throughout a query session, and b) submit queries in the same language as the interface they have selected, except in a large number of cases where the English interface is used to submit Latin queries. ACM

[1]  Vittorio Loreto,et al.  Language trees and zipping. , 2002, Physical review letters.

[2]  Michael R. Lyu,et al.  A novel log-based relevance feedback technique in content-based image retrieval , 2004, MULTIMEDIA '04.

[3]  Wei-Ying Ma,et al.  Query Expansion by Mining User Logs , 2003, IEEE Trans. Knowl. Data Eng..