计算机应用 ›› 2012, Vol. 32 ›› Issue (07): 2038-2040.DOI: 10.3724/SP.J.1087.2012.02038

• 典型应用 • 上一篇    下一篇

面向文本情感分析的商品评价信息检测

庞海杰   

  1. 青岛滨海学院 理科基础学院,山东 青岛266555
  • 收稿日期:2011-12-20 修回日期:2012-02-05 发布日期:2012-07-05 出版日期:2012-07-01
  • 通讯作者: 庞海杰
  • 作者简介:庞海杰(1976-),男,山东沂水人,硕士,主要研究方向:话题检测与跟踪、自然语言处理。

Text sentiment analysis-oriented commodity review detection

PANG Hai-jie   

  1. College of Science Foundation Institute, Qingdao Binhai University, Qingdao Shandong 266555, China
  • Received:2011-12-20 Revised:2012-02-05 Online:2012-07-05 Published:2012-07-01
  • Contact: PANG Hai-jie

摘要: 为及时有效地获取商品评价信息,提出了基于评价对象识别的商品评价信息检测方法。首先在中文分词的基础上,依据词性标注结果抽取商品评价信息中的候选评价对象;然后基于规则过滤和共现扩展的方法得到精准评价对象;最后实现了基于评价对象识别的商品评价信息检测方法。实验结果表明,与基本模型相比,提出的商品评价信息检测方法的F-Measure提高了34.8%,证明了充分挖掘商品评价信息中的评价对象可以非常有效地改善检测方法的性能。

关键词: 情感分析, 话题检测, 评价对象, 共现, 商品评价

Abstract: To help sellers timely and effectively get the commodity reviews, the author proposed a commodity review detection method based on the key word extraction. Based on the Chinese word segmentation, this method firstly extracted the candidate keywords based on part of speech tagging, and then filtered out noises based on the rules. The paper extracted the dynamic co-occurrence of the commodity reviews, and extended the keywords based on these dynamic co-occurrences. Finally the commodity review detection method based on the keyword was realized. The experimental results show that, compared to the basic model, the proposed method can increase the F-Measure by 34.8%, which proves that efficient key word extraction is very useful to the commodity review detection.

Key words: sentiment analysis, topic detection, opinion target, co-occurrence, product review

中图分类号: