基于主题的关键词提取对微博情感倾向的研究

Research on sentiment tendency of weibo based on topic keyword extraction

在已有基于关键句的微博情感分析中,存在提取的关键句质量不高的问题,本文提出了一种与主题相关的关键词提取算法。首先,将得到的关键词作为特征函数之一来进行关键句的提取;针对7种影响情感倾向的词性搭配,对其计算规则进行了改进,而后计算关键句情感值;最后,将句子的情感值计算扩展到微博的情感值计算,对其情感倾向进行分析。实验表明,本文方法与同类算法相比准确率提高了3.23%,同时,与主题相关的关键词在提取的关键句中占了很大比重,较好地解决了关键句质量不高的问题。

In the existing research of the emotional analysis of the micro-blog based on key sentence,there is a problem that the quality of the extracted key sentence is not high.In this paper,a topic-related keyword extraction algorithm is proposed.Firstly,the key words are extracted as one of the characteristic functions to extract the key phrases.For the seven parts of speech collocations that affect the emotional tendencies,the calculation rules are improved,and then the key emotion values are calculated.Finally,the emotional value calculation of sentences is extended to the emotional value calculation of micro-blog,and the emotional tendency is analyzed.Experiments show that the accuracy of this method is improved by 3.23% compared with that of other algorithms.At the same time,the key words related to the topic account for a large proportion of the key sentences extracted,which solves the problem that the quality of key sentences is not high.