Temporal Semantics Extraction for Improving Web Search

Current Web-search engines can be improved through techniques that consider a temporal dimension in both query formulation and information extraction processes. An accurate recognition of temporal expressions in data sources is required, as well as extracting semantic embedded meta-information concerning time in a standard format that allows reasoning without ambiguity. Temporal Web searching entails an interesting advance on the way to the Semantic Web. We propose a temporal expressions recognition and normalization (TERN) system for contents in Spanish, which has been integrated into a Web-search engine prototype. The contribution of this system lies in its capability of taking into account the semantics of time in the user queries as well as in the search collections, independently of its representation. Evaluation figures show that the inclusion of temporal information management aptitudes to Web searching means considerable improvements in its performance.