Query Language and Query Intent in Cross-Lingual Web Search

This paper examines query intent (or domain) classification as a proxy for user information need and its use in improving the performance of cross-lingual information retrieval (CLIR). Through several human relevance studies, information needs that cross language and culture barriers are identified. Therefore, we propose to use query intent classifiers as a filter to pick queries that may benefit from CLIR. The analysis presented in this paper are based on experiments conducted on three populations with a high percentage of bilingual internet users: the Arabs, the French Canadians, and Spanish speakers in the US. While these users differ significantly in their usage of search engines, the effect of CLIR on the quality of the search experience for individual intents was largely similar. This leads us to expect a wider applicability of our findings on a global scale.