作者: Krishnaprasad Thirunarayan , Trivikram Immaneni , Mastan Vali Shaik
DOI: 10.1007/978-3-540-73351-5_11
关键词:
摘要: This work deals with determination of meaningful and terse cluster labels for News document clusters. We analyze a number alternatives selecting headlines and/or sentences in (obtained as result an entity-event-duration query), formalize approach to extracting short phrase from well-supported headlines/sentences the that can serve label. Our technique maps sentence into set significant stems approximate its semantics, comparison. Eventually label is extracted selected headline/sentence contiguous sequence words, resuscitating word sequencing information lost formalization semantic equivalence.