计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (10): 115-121.DOI: 10.3778/j.issn.1002-8331.1901-0166

• 模式识别与人工智能 • 上一篇    下一篇

基于TAKE的中文关键短语提取算法研究

刘晨晖,张德生,胡钢   

  1. 西安理工大学 理学院,西安 710054
  • 出版日期:2020-05-15 发布日期:2020-05-13

Research on Chinese Key Phrase Extraction Algorithm Based on TAKE

LIU Chenhui, ZHANG Desheng, HU Gang   

  1. Faculty of Sciences, Xi’an University of Technology, Xi’an 710054, China
  • Online:2020-05-15 Published:2020-05-13

摘要:

针对传统的中文关键短语提取算法所提取关键短语准确率低、歧义性强和涵盖信息量少等问题,在英文关键短语提取算法TAKE(Totally Automated Keyword Extraction)的启发下,加入基于多领域特异性的新词识别技术,并改进了原有算法的文本分词、词语过滤和特征计算方法,提出了一种改进的TAKE算法,并应用于中文文本关键短语挖掘中。与多种传统关键短语提取算法的对比实验结果表明,该算法提取的精确率、召回率和[F]值指标的量化结果相比于传统算法有比较明显的提升。

关键词: 文本挖掘, 分词, 词语过滤, 特征计算, 关键短语提取

Abstract:

Aiming at the problems that traditional Chinese key phrase extraction algorithm has low accuracy, strong ambiguity and little information, inspired by the English key phrase extraction algorithm TAKE(Totally Automated Keyword Extraction), this paper adds a new word recognition technique based on multi-domain specificity and improves the text segmentation, word filtering and feature calculation methods of the original algorithm. An improved TAKE algorithm is proposed and applied to Chinese text key phrase mining. By compared with traditional key phrase extraction algorithms, it is shown that the proposed algorithm has a significant improvement in the quantization results of precision, recall and [F] values.

Key words: text mining, word segmentation, word filtering, feature calculation, key phrase extraction