Semantic Rule Filtering for Web-Scale Relation Extraction

摘要: Web-scale relation extraction is a means for building and extending large repositories of formalized knowledge. This type automated knowledge requires decent level precision, which hard to achieve with automatically acquired rule sets learned from unlabeled data by distant or minimal supervision. paper shows how precision can be considerably improved employing wide-coverage, general-purpose lexical semantic network, i.e., BabelNet, effective filtering. We apply Word Sense Disambiguation the content words extracted rules. As result set relation-specific relevant concepts obtained, each these then used represent structured semantics corresponding relation. The resulting subgraphs BabelNet are as filters estimating adequacy For seven relations tested here, filter consistently yields higher at any relative recall value in high-recall range.