Topic Indexing with strongWikipediastrong - aaai.org.pdfVIP

Topic Indexing with Wikipedia Olena Medelyan, Ian H. Witten and David Milne Computer Science Department University of Waikato New Zealand {olena, ihw, dnk2}@cs.waikato.ac.nz Abstract dia applicable to nearly any domain. However, its vast size Wikipedia can be utilized as a controlled vocabulary for creates new challenges. identifying the main topics in a document, with article titles This paper shows how Wikipedia can be utilized effec- serving as index terms and redirect titles as their synonyms. tively for topical indexing. The scheme is evaluated on a Wikipedia contains over 4M such titles covering the termi- set of 20 computer science articles, indexed by 15 teams of nology of nearly any document collection. This permits con- computer science students working independently, two per trolled indexing in the absence of manually created vocabu- team. The automatic approach reaches the average per- laries. We combine state-of-the-art strategies for automatic controlled indexing with Wikipedia’s unique property—a formance of these teams, and needs very little training. richly hyperlinked encyclopedia. We evaluate the scheme by comparing automatically assigned topics with those cho- 2. Related work sen manually by human ind