摘要
随着互联网的广泛应用,在Blog、BBS、Wiki等网络站点中出现了大量的针对商品或服务的客户评论,这些客户评论中所包含的丰富信息,对企业管理具有重要的价值.通过数据挖掘算法对客户针对某一产品的大量评论进行分析,可以挖掘出这些产品的主要特征,并有望进一步发现客户对这些特征的意见和态度.在英文世界中已经有学者开始对这一研究进行探索,然而由于语言结构等方面的差异,英文的研究成果尚无法直接应用于中文客户评论的挖掘中.本研究针对中文的特点,提出了面向中文的客户评论挖掘方法.该方法基于改进关联规则算法实现了针对中文产品评论的产品特征信息挖掘.本研究采用通过互联网获得的针对手机、数码相机、书籍等5种产品的评论语料,对该方法进行了数据实验,实验结果初步验证了该方法有效性.
Nowadays, more and more customers read online reviews on products before making the decision of purchase. It is also a common practice for merchants and manufacturers to get useful feedback from reviews written by their customers on products and associated services. Therefore, mining features of products from online reviews has emerged to be an important research topic. However, most present studies focused mainly on English reviews. As China becomes a potential e-commerce market in the world, Chinese have already been to become one of the most important group of customers. The numbers of Chinese online reviews in- creased greatly in recent years. It makes the study of features retrieved from Chinese reviews imperatively im- portant. The techniques used for English reviews can hardly be applied directly to Chinese reviews due to the differences in characteristics between these two languages. Association rule based methods for products＇ fea- ture retrieving to English reviews were modified and improved to fit Chinese reviews in this study. Experiments demonstrate the validity of this new method.
出处
《管理科学学报》
CSSCI
北大核心
2009年第2期142-152,共11页
Journal of Management Sciences in China
基金
国家自然科学基金资助项目(70771032,70501009)
香港理工大学研究基金资助项目(G-YX93)
关键词
用户评论
产品特征
关联规则
数据挖掘
customer reviews
product features
association rule
data mining
