Wikitology: Wikipedia as an ontology
Thursday, August 28, 2008, 14:00pm - Thursday, August 28, 2008, 15:30pm
University Center Ballroom
Wikipedia has become an important source of online knowledge for people that is kept up to date and available in many languages. We describe an approach to extracting information from Wikipedia and related sources to construct an ontology and associated knowledge base. The core idea is to use Wikipedia's articles and associated pages as a topic ontology. The benefits of the approach are that the ontology terms are developed through a social process, maintained and kept current by the Wikipedia community, represent a consensus view, and have meaning that can be understood by reading the associated pages. We have demonstrated the use of Wikitology to improve the performance of an information retrieval system and as a source of evidence in a intra-document entity co-reference task. Processing Wikipedia to construct the knowledge base and maximizing the precision and recall of queries against it are computationally expensive. We will describe our current and planned use of parallel computing techniques to achieve good performance on these tasks.