Research in Computing Science, Vol. 70, pp. 183-194, 2013.
Abstract: An infobox of a Wikipedia article generally contains key facts in the article and is organized as attribute-value pairs. Infoboxes not only allow readers to rapidly gather the most important information about some aspects of the articles in which they appear, but also provide a source for many knowledge bases derived from Wikipedia. However, not all the values of infobox attributes are updated frequently and accurately. In this paper, we propose a method to automatically detect outdated attribute values in Wikipedia infoboxes by using facts extracted from the general Web. Our method uses the pattern-based fact extraction approach. The patterns for fact extraction are automatically learned using a number of available seeds in related Wikipedia infoboxes. We have tested and evaluated our system on a set of 100 well-established companies in the NASDAQ-100 index on their employee numbers, presented by the num_employees attribute value in their Wikipedia article infoboxes. The achieved accuracy is 77% and our test result also reveals that 82% of the companies do not have their latest numbers of employees in their Wikipedia article infoboxes.
Keywords: Information Extraction, Wikipedia Update, Pattern Learning
PDF: Automatic Detection of Outdated Information in Wikipedia Infoboxes
PDF: Automatic Detection of Outdated Information in Wikipedia Infoboxes