网络商品评论细粒度情感分析系统关键技术研究

发布时间：2018-03-24 19:53

本文选题：网络评论　切入点：数据获取　出处：《大连海事大学》2017年硕士论文

【摘要】：网络商品评论情感分析主要分为粗粒度的情感分析与细粒度的情感分析,粗粒度的情感极性分析重在用于判断文本整体的情感倾向,主要的方法特征项选择、特征项降维和特征项权重计算。粗粒度的情感分析并没有从属性角度出发更详细地在解决消费者面临的评论数据超载和识别过程复杂的问题,而细粒度的情感极性分析重在从评论中找出这些商品属性,并判断消费者对于各个属性的情感倾向。为使情感分析的效用更加精准的作用于某一属性之上,并展示给大众,本文利用网络爬虫技术获取网络评论语料,然后将语料中的属性词与情感词抽取出来进行细粒度情感倾向分析与计算,并设计相应的系统将实验分析与结果展示出来。本文主要研究工作为,首先利用Shark-PageRank主题爬取算法生成待爬取网页队列方法、Bloom Filter网页去重方法和Jsoup网页数据获取方法三者相结合的网络爬虫技术进行数据获取,将抓取到的网络评论作为语料库。然后结合中文网络评论语言表达方式的特点,使用一种基于语义分析的属性词与情感词词对联合识别抽取的方法,并将该抽取方法与面向属性词的方法相结合提出了一种基于语义分析网络评论细粒度情感分析方法。最后,在上述工作基础上,结合Web前端技术,设计了一个网络评论细粒度情感分析系统,此系统提供网络评论数据获取,网络评论细粒度情感分析,并将评论文本的细粒度情感分析结果展示的功能。经过对本论文的钻研分析,可以将用户评论的情感分析的精确性更深一步提升,并使其通过情感分析在商品属性上进行修整、制定消费决策等两方面发挥更大的指导作用。
[Abstract]:The emotional analysis of online commodity reviews is mainly divided into coarse-grained emotional analysis and fine-grained emotional analysis. Coarse-grained affective polarity analysis is mainly used to judge the emotional tendency of the text as a whole. Coarse-grained emotional analysis does not address in more detail the complexity of comment data overload and identification process faced by consumers from the perspective of attributes. The fine-grained affective polarity analysis focuses on finding out these commodity attributes from the comments and judging the consumer's emotional propensity for each attribute, in order to make the utility of emotional analysis more accurate on one attribute and show it to the public. In this paper, the web crawler technology is used to obtain the network comment corpus, and then the attribute words and emotion words from the corpus are extracted for fine-grained emotional tendency analysis and calculation. And design the corresponding system to display the experimental analysis and results. Firstly, the Shark-PageRank topic crawling algorithm is used to generate the web crawler technology which combines the bloom Filter web page removal method and the Jsoup web page data acquisition method to obtain the data. The collected comments are used as a corpus, and then a method based on semantic analysis is used to identify and extract attribute words and affective words according to the characteristics of the Chinese network comment language. A fine-grained emotional analysis method based on semantic analysis for network comments is proposed by combining the extraction method with the attribute-oriented method. Finally, based on the above work, combined with the Web front-end technology, A fine-grained emotional analysis system for network comments is designed. The system provides data acquisition for network reviews and fine-grained emotional analysis for network reviews. Through the research and analysis of this paper, we can further enhance the accuracy of the emotional analysis of user comments, and make it through emotional analysis to modify the commodity attributes. Making consumer decisions and other two aspects to play a greater guiding role.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.1

【参考文献】