基于情感分析和证据理论的多属性在线评论决策方法

Multi-Attribute Online Review Decision Making Method Based on Sentiment Analysis and Evidence Theory

近年来,大量的有关产品或服务的在线评论信息在众多网站涌现,如何基于在线评论进行决策分析是一个值得关注的研究问题。提出了一种基于情感分析和证据理论的多属性在线评论决策方法,在该方法中,首先针对决策者关注的备选方案及属性采集相应的在线评论;然后针对采集的多属性在线评论进行情感分析并得到在线评论针对各属性在不同情感强度等级上出现的频次;进一步,依据证据理论,将在线评论在不同情感强度等级上出现的频次转化为针对不同情感强度等级的信任度,并依据证据合成规则将备选方案各属性的信任度集结为综合信任度;在此基础上,通过计算所有备选方案的效用值,进而得到方案的排序结果用于决策者购买决策;最后,以基于汽车之家网站的在线评论信息进行汽车购买决策为例说明了该方法的可行性和有效性。

In recent years, a large amount of online review information on products or services emerges on many websites. Therefore, it is worthwhile to make decision analysis based on online. A method for multi-attribute online review decision was proposed based on the sentiment analysis and evidence theory. First, online reviews with respect to each attribute of alternatives concerned by decision maker were collected. Then, the different emotional intensity level frequency of online reviews collected with respect to each attribute was obtained using the sentiment analysis. After that, according to the evidence theory, the different emotional intensity level frequency of online reviews was transformed into a degree of belief which was assigned to the different emotional intensity levels. According to the evidence synthesis rules, the comprehensive degree of belief was obtained by integrating the degree of belief with respect to each attribute of alternatives, based on which, the overall utility value of each alternative was calculated, and the ranking result of all alternatives were determined. Finally, a car purchase example based on the online review information from https://www.autohome.com.cn was used to illustrate the feasibility and validity of the proposed method.