Computer Engineering and Applications ›› 2019, Vol. 55 ›› Issue (20): 95-100.DOI: 10.3778/j.issn.1002-8331.1806-0372

Previous Articles     Next Articles

Research on Sentiment Classification of Title and TextRank Extracting Key Sentences

ZHENG Cheng, QIAN Gailin, ZHANG Jinping   

  1. 1.Key Laboratory of Intelligent Computing & Signal Processing(Anhui University), Ministry of Education, Hefei 230601, China
    2.School of Computer Science and Technology, Anhui University, Hefei 230601, China
  • Online:2019-10-15 Published:2019-10-14



  1. 1.计算智能与信号处理教育部重点实验室(安徽大学),合肥 230601
    2.安徽大学 计算机科学与技术学院,合肥 230601

Abstract: Considering that different sentences have different degrees of importance in judging the sentiment tendency of a document, thus, distinguishing a key sentence and a specific sentence will help improve the performance of the sentiment classification. Meanwhile, taking into account Title and contextual information, a sentiment analysis method SKTT based on Title and weighted TextRank is proposed to achieve effective sentiment analysis. Firstly, the contribution of Title is calculated according to the sentiment weight of the document Title, considering the impact of punctuation and semantic rules on sentiment orientation. Then, according to the idea of weighted TextRank algorithm, an sentiment sentence directed graph is constructed in the document body to extract key sentences. Finally, the sentiment tendency of all key sentences is calculated to achieve sentiment classification. Experimental results across four domains show that the performance of the SKTT method is significantly better than the Baseline method and has high efficiency.

Key words: Title, TextRank algorithm, key sentence, sentiment classification, semantic rules

摘要: 考虑到不同句子对判断文档情感倾向的重要程度不同,因而区分文档的关键句和细节句将有助于提高情感分类的性能。同时,考虑到Title和上下文信息,提出了一种基于Title和加权TextRank抽取关键句的情感分析方法SKTT,实现了高效的情感分析。根据文档Title的情感权重计算Title贡献度,考虑到标点和语义规则对情感倾向的影响;根据加权TextRank算法思想,在文档正文中构建了一个情感句有向图来提取关键句;计算所有关键句的情感倾向进行情感分类。在4个领域上进行实验,实验结果表明,该SKTT方法性能明显优于Baseline,具有高效性。

关键词: Title, TextRank算法, 关键句, 情感分类, 语义规则