Paper: A Weighting Scheme for Open Information Extraction

ACL ID N12-2011
Title A Weighting Scheme for Open Information Extraction
Venue Annual Conference of the North American Chapter of the Association for Computational Linguistics
Session Student Session
Year 2012
Authors

We study1 the problem of extracting all pos- sible relations among named entities from un- structured text, a task known as Open Infor- mation Extraction (Open IE). A state-of-the- art Open IE system consists of natural lan- guage processing tools to identify entities and extract sentences that relate such entities, fol- lowed by using text clustering to identify the relations among co-occurring entity pairs. In particular, we study how the current weighting scheme used for Open IE affects the clustering results and propose a term weighting scheme that significantly improves on the state-of-the- art in the task of relation extraction both when used in conjunction with the standard tf ? idf scheme, and also when used as a pruning fil- ter.