摘要
提出了一种依据商品在线评论的基于多粒度情感强度分析和随机逼近理想点排序法的商品排序方法.使用爬虫软件和ICTCLAS对消费者关注的备选商品的在线评论进行获取和预处理.依据预处理后的评论,通过提出多粒度情感强度分析算法确定每条评论针对商品属性的情感强度值.通过对得到的情感强度值进行统计分析,得到备选商品针对商品属性的多粒度情感强度分布形式的属性值.最后,依据得到多粒度情感强度分布形式的属性值,采用随机逼近理想点排序法确定备选商品的排序.基于中关村在线中的数码相机在线评论,给出了提出方法应用的实例分析.
How to automatically analyze the huge amounts of online reviews and rank products is a new important research topic. This paper proposes a method based on multi-granularity sentiment strength analysis and stochastic technique to order preferences for products through online reviews according to the closeness to an ideal solution(TOPSIS). In this method, online reviews of alternative products are first crawled by web crawler software and processed by ICTCLAS software. Then, according to the processed online reviews, an algorithm is given to calculate the sentiment strengths of online reviews concerning product features. Furthermore, according to the results of sentiment strength analysis, the feature values in the form of distribution concerning multi-granularity sentiment strengths can be obtained by statistical analysis. According to the obtained feature values, the ranking of alternative products can be determined by stochastic TOPSIS method. Finally, based on the online reviews on digital camera from the Zhongguancun online, a case analysis is given to illustrate the proposed method.
作者
毕建武
刘洋
樊治平
Bi Jianwu;Liu Yang;Fan Zhiping(School of Business Administration,Northeastern University,Shenyang 110167,China)
出处
《系统工程学报》
CSCD
北大核心
2018年第3期422-432,共11页
Journal of Systems Engineering
基金
国家自然科学基金资助项目(71771043
71571039
71271049
71371002)
中央高校基本科研业务经费资助项目(N170605001)
关键词
商品排序
在线评论
多粒度情感强度
累积分布
随机逼近理想点排序法
goods ranking
online reviews
multi-granularity sentiment strength
cumulative distribution
stochastic technique for order preference by similarity to an ideal solution