基于在线网站评论的中文文本挖掘

发布时间：2018-05-05 14:22

本文选题：文本挖掘 + 网络评论　；参考：《华中科技大学》2016年硕士论文

【摘要】：随着电子商务的快速发展和不断普及,人们的消费习惯发生了深刻的变化,越来越多的消费和交易行为在网上进行。而这种频繁的网络活动产生了大量的网络评论,因而对于消费者与消费者、消费者与商家来说这是一个重要的渠道,能够快捷方便地交换信息,以此来获得所需要的有价值的信息。对于潜在消费者来说,通过商务网站的网络评论,可以全面详细的了解所关注的产品或服务,并根据其他评论用户的评论信息做出是否购买的决定;对于商家来说,从消费者的评论信息中,可以了解到消费者的消费习惯、兴趣特征和消费意向,从而改善产品和服务,提升企业竞争力。文章运用的文本挖掘方法,包括情感倾向分析和LDA主题模型,对评论文本进行深入分析。以我国的在线旅游网站的评论文本为例,以R统计软件为工具,利用统计分析及文本挖掘方法对评论文本数据进行实证分析。首先基于情感词典,计算评论中语句所包含的情感词的平均值,将评论文本划分为正面评论和负面评论,然后运用LDA主题模型,分别提取这两类评论的主题。通过情感倾向分析和主题提取,提高评论的可靠性和有用性,提升消费者对评论的信任度;使企业能够及时有效追踪用户需求,为互联网企业改善产品和服务以及精细化营销策略提供参考和建议。
[Abstract]:With the rapid development and popularization of electronic commerce, people's consumption habits have undergone profound changes. This frequent network activity has produced a large number of online reviews, so it is an important channel for consumers and consumers, consumers and businesses to exchange information quickly and conveniently. To get the valuable information you need. For potential consumers, online reviews on business sites can provide a comprehensive and detailed understanding of the products or services they are interested in, and make decisions about whether to buy them based on comments from other users; for businesses, From the consumer's comment information, we can know the consumer's consumption habit, interest characteristic and consumption intention, so as to improve the products and services and enhance the competitiveness of enterprises. The text mining methods, including affective tendency analysis and LDA topic model, are used to analyze the comment text deeply. Taking the comment text of our country's online tourism website as an example, using the R statistical software as the tool, the paper makes an empirical analysis of the comment text data by using the statistical analysis and text mining methods. Firstly, based on the emotion dictionary, the average of the emotional words contained in the sentence is calculated, and the comment text is divided into positive comment and negative comment, and then the topic of the two kinds of comments is extracted by using the LDA topic model. Through emotional tendency analysis and topic extraction, improve the reliability and usefulness of comments, enhance consumers' trust in comments, and enable enterprises to track user needs in a timely and effective manner. For Internet enterprises to improve products and services and elaborate marketing strategies to provide reference and advice.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.1;F724.6;F592.6

【参考文献】