Representation and verification of attribute knowledge

Abstract

With the increasing growth and popularization of the Internet, knowledge extraction from the web is an important issue in the fields of web mining, ontology engineering and intelligent information processing. The availability of real big corpora and the development of technologies of internet network and machine learning make it feasible to acquire massive knowledge from the web. In addition, many web-based encyclopedias such as Wikipedia and Baidu Baike include much structured knowledge. However, knowledge qualities including the incorrectness, inconsistency, and incompleteness become a serious obstacle for the wide practical applications of those extracted and structured knowledge. In this paper, we build a taxonomy of relations between attributes of concepts, and propose a taxonomy of attribute relations driven approach to evaluating the knowledge about attribute values of attributes of entities. We also address an application of our approach to building and verifying attribute knowledge of entities in different domains. © 2013 Springer-Verlag Berlin Heidelberg.