¡¡Chinese Journal of Computers   Full Text
  TitleResearch on Maximum Entropy Model for Keyword Indexing
  AuthorsLI Su-Jian1) WANG Hou-Feng1) YU Shi-Wen1) XIN Cheng-Sheng2)
  Address1)(Institute of Computational Linguistics, Peking University, Beijing 100871) 2)(The Information Center of PEOPLE¡¯S DAILY, Beijing 100733)
  Year2004
  IssueNo.9(1192¡ª1197)
  Abstract &
  Background
`
keywords are very useful to document management, text clustering/classification, information retrieval and so on. Thus, authors propose to use Maximum Entropy £¨ME£© model to conduct automatic keyword indexing. ME model is a mature mathematical model and has been applied to solve many problems. However, it¡¯s flexible how to use the model. Aiming at the indexing task and the resources available, firstly, the useful features for ME model is introduced. Secondly, three tests are given with their experimental results. Lastly, the application of ME model in automatic keyword indexing are analyzed and discussed. Keywords keyword indexing; keyword extraction; maximum entropy model; information extraction