面向中文在线评论意见的挖掘算法研究及应用

发布时间：2018-08-22 20:07

【摘要】：网络购物行业的发展日益成熟,越来越多的消费者在购物网站发布评论信息。产品评论反映了消费者对产品的态度和意见,很有实用价值。一方面,产品评论可以影响其他消费者的购买意向;另一方面,产品评论反馈了产品各方面信息,方便商家提高产品和服务质量。但是,想要短时间内从大量的产品评论文本数据中得到有意义的信息,是非常困难的。研究中文评论意见挖掘方法,对于提高文本信息提取效率有重要意义。本文主要研究内容如下:以在线中文评论为研究对象,首先采用人工的方式,将主观性评论文本从产品评论中分离出来。然后利用自然语言处理技术,对中文在线评论意见文本进行预处理。针对已有的意见挖据工作对中文产品评论信息挖掘的低查全率和低查准率问题,文中提出了一种改进的中文在线评论意见挖掘算法。该方法根据自然语言表达方式,将评论文本分为四类句式结构。然后基于副词抽取各类评论文本中的产品意见词和特征词。实验结果表明,该方法能够有效提高中文产品评论意见挖掘的查全率和查准率。基于《同义词词林扩展版》将抽取出的产品特征和意见词中的同义词合并,再根据支持度阈值剪枝,得到最终的特征词和意见词。实验表明,该方法提高了同义词合并的准确性。本文改进的中文产品评论意见挖掘方法,可充分利用自然语言表达特点,实现对中文产品评论特征和意见词自动抽取。
[Abstract]:With the development of online shopping industry, more and more consumers publish comments on shopping websites. Product reviews reflect consumer attitudes and opinions on products and are of practical value. On the one hand, product review can affect the purchase intention of other consumers; on the other hand, product review feedback the information of all aspects of the product, which is convenient for merchants to improve the quality of products and services. However, it is very difficult to get meaningful information from a large amount of product review text data in a short time. It is important to study the mining method of Chinese comments for improving the efficiency of text information extraction. The main contents of this paper are as follows: firstly, the subjective comment text is separated from the product review by artificial method. Then the text of Chinese online comments is preprocessed by natural language processing technology. In order to solve the problem of low recall and low precision of Chinese product comment information mining, an improved Chinese online comment mining algorithm is proposed in this paper. According to the natural language expression, the commentary text is divided into four types of sentence structure. Then, product comment words and feature words are extracted from all kinds of comments based on adverbs. The experimental results show that this method can effectively improve the recall and precision of Chinese product comment mining. Based on the extended version of synonym forest, the extracted product features are combined with the synonyms in the opinion words, then pruned according to the support threshold, the final feature words and opinion words are obtained. Experiments show that this method improves the accuracy of synonym merging. The improved mining method of comments on Chinese products can make full use of the characteristics of natural language expression and realize the automatic extraction of comment features and words of Chinese products.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.1

【参考文献】