Mining soft-matching rules from textual data.pdfVIP

Proceedings of the Seventeenth International Joint Conference on Artificial Intelligence(IJCAI-01), pages 979-984, Seattle, WA, August, 2001 Mining Soft-Matching Rules from Textual Data Un Yong Nahm and Raymond J. Mooney Department of Computer Sciences, University of Texas, Austin, TX 78712-1188fpebronia,mooneyg@ Abstract Text mining concerns the discovery of knowledge from unstructured textual data. One important task is the discovery of rules that relate specific words and phrases. Although existing methods for this task learn traditional logical rules, soft-matching methods that utilize word-frequency information generally work better for textual data. This paper presents a rule induction system, TEXTRISE, that allows for partial matching of text-valued features by combining rule-based and instance-based learn- ing. We present initial experiments applying TEX- TRISE to corpora of book descriptions and patent documents retrieved from the web and compare its results to those of traditional rule and instance based methods. 1 Introduction Text mining, discovering knowledge from unstructured natural-language text, is an important data mining problem attracting increasing attention [Hearst, 1999; Feldman, 1999; Mladenic?, 2000]. Existing methods for mining rules from text use a hard, logical criteria for matching rules [Feldman and Hirsh, 1996; Ahonen-Myka et al., 1999]. However, for most text processing problems, a form of soft matching that utilizes word-frequency information typically gives superior results [Salton, 1989; Cohen, 1998; Yang, 1999]. Therefore, the induction of soft-matching rules from text is an important, under-studied problem. We present a method, TEXTRISE, for learning soft- matching rules from text using a modification of the RISE algorithm [Domingos, 1996], a hybrid of rule-based and instance-based (nearest-neighbor) learning methods. Such a hybrid is good match for text mining since rule-induction provides simple, interpretable rules, while near