摘要
用户在线评论是影响消费者汽车产品选择决策的重要信息来源。已有的基于用户在线评论的产品排序研究中,较少考虑用户评价标准差异对排序结果的影响。基于文本型、数值型和极性评论数据,本文提出一种考虑个体评价标准差异的汽车产品排序方法:首先分析文本型评论中的用户情感,构建情感词典,得到初步情感得分;然后利用数值型和极性评论相关值为参数的可变语言尺度函数对情感得分进行调整,以平衡个体评价标准差异的影响;最后用概率语言集来表达调整后的情感得分及其概率密度,根据DS-PLWA算子进行融合得到最终评价结果。基于汽车之家论坛上6种汽车产品的用户在线评论数据,利用本文推荐排序方法进行实例分析,结果表明考虑个体评价标准差异会影响汽车产品排序结果。
Online reviews are regarded as a significant information origin of users’ selectionof automobile products.Little consideration has been taken on individual standard differenceby most research on automobile products ranking.To make evaluation results more aligned with individual standard difference, text,digital and polarity comments have been taken for automobile products ranking.First,the automobile sentiment dictionary has been constructed for the initial text reviews’ scores;Second,digital and polarity comments have been applied as critical factors in changeable linguistic scale function,which is used for scores adjustment to reduce the influence of individual standards difference;Finally,probabilistic linguistic sets have been built for score distribution and merged by?DS-PLWA operator.Illustrated by the results,considering individual standard difference or not can make the recommendation ranking different,when six kinds of automobile products from Automobile Home are compared.
作者
李宛哲
李英
LI Wan-zhe;LI Ying(Business School,East China University of Science and Technology,Shanghai 200237,China)
出处
《系统工程》
北大核心
2021年第3期143-152,共10页
Systems Engineering
基金
国家自然科学基金面上项目(72074076)。
关键词
在线评论
个体评价标准差异
产品排序
概率语言集
语言尺度
Online Reviews
Individual Standard Difference
Products Recommendation
Probabilistic Linguistic Terms Sets
Linguistic Scale
