Elementary

• DOI: 10.4018/jswis.2012070103
• Publication Date: 2013-01-14T21:57:39Z
• Authors: Feng Niu; Ce Zhang; Christopher Ré; Jude Shavlik
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology

Researchers have approached knowledge-base construction (KBC) with a wide range of data resources and techniques. The authors present Elementary, prototype KBC system that is able to combine diverse different techniques via machine learning statistical inference construct knowledge bases. Using they implemented solution the TAC-KBP challenge quality comparable state art, as well an end-to-end online demonstration automatically continuously enriches Wikipedia structured by reading millions webpages on daily basis. describe several challenges their solutions in designing, implementing, deploying Elementary. In particular, first conceptual framework architecture Elementary integrate principled manner. They then discuss how address scalability enable Web-scale deployment. empirically show this decomposition-based approach achieves higher performance than prior approaches. To validate effectiveness Elementary’s KBC, experimentally its ability incorporate signals has positive impacts quality.