挖掘中文网络客户评论的产品特征及情感倾向  

Mining product features and sentiment orientation from Chinese customer reviews

作　　者：李实[1] 叶强[2] 李一军[2] 罗嗣卿[1] 

机构地区：[1]东北林业大学信息与计算机工程学院,哈尔滨150040 [2]哈尔滨工业大学经济与管理学院,哈尔滨150001

出　　处：《计算机应用研究》2010年第8期3016-3019,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(70971033)

摘　　要：为探索中文客户评论中的产品特征及相关情感倾向的挖掘,以帮助生产商和服务商改进产品、改善服务,提高竞争力,提出采用基于Apriori算法的非监督型产品特征挖掘算法,结合监督型情感分析技术,实现对于评论中产品特征及其情感倾向的综合信息挖掘;并根据用户的关注权重将产品特征和情感倾向进行排列。采用几种从互联网下载的真实产品评论语料,对该方法进行了数据实验,实验结果初步验证了该方法的有效性。At mining product features and polarity of customer’s opinion from Chinese online customer reviews,in order to help manufactures and service providers improve their products and services,and gain competitive advantages. This paper proposed an approach based on Apriori algorithm which unsupervised feature extraction algorithm and supervised sentiment analysis technology to realize the summery information of customer reviews. And then ranked the work features and opinions based on customer attention strength. Several actual Internet products reviews were tested by an experimental model. Final results demonstrate the validity of the method.

关 键 词：用户评论 产品特征 数据挖掘 情感分析 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]