Horizontal Search Method for Wikipedia Category Grouping

Category hierarchies, which show the basic relationship between concepts, are utilized as fundamental clues for semantic information processing in diverse research fields. These research works have employed Wikipedia due to its high coverage of real-world concepts and data reliability. Wikipedia also constructs a category hierarchy, and defines various categories according to the common characteristics of a concept. However, some limitations have been uncovered in the use of a vertical search (especially top-down) to form a set of domain categories. In order to overcome these limitations, this paper proposes a horizontal search method, and uses Wikipedia components to measure the similarity between categories. In an experimental evaluation, we confirm that our method shows a wide coverage and high precision for similar (domain) category grouping.