Knowledge fusion framework based on Web page texts

摘要

With the proliferation ofWeb page texts, it is important to fuse these texts to useful documents that users need. However, there is still no complete and unified theoretical model for studying the research issues including redundancy, localization, and fuzziness existing in the process of fusing Web page texts. This paper proposes a fusion framework calledWeb Pages Knowledge Fusion Framework (WPKFF) to synthesize the knowledge of Web page texts. First, sentences in Web page texts are extracted and transformed into triple semantic net as knowledge representation. Then a semantic description of attribute fusion rules, description information fusion rules and attribute-value and description information fusion rules are defined in WPKFF. These rules are used to fuse the attributes of same domain concepts in triple semantic net. The features of attributes include description (string) and value data (number). The results of the experiments indicate that the fusion framework is a feasible model in terms of precision and recall.