摘要
[目的/意义]基于多维度测量电商商品热度,并分析不同属性组合的商品热度差异,为消费行为和心理探究、企业产品设计研发和商家经营策略制定提供重要参考。[方法/过程]从电商商品用户评论数据出发,基于传统的威尔逊置信区间商品好评率排名算法,考虑商品评论的时间因素,并额外加入评论数据的情感倾向因素,运用熵权法获取各指标的权重,构建出一种新的电商商品热度测量模型,并基于多份不同商品数据对模型进行应用研究。[结果/结论]与传统的热度模型相比,所构建的多维度模型加入了时间因素和情感因素后,测量结果解释性更强,更加贴合实际情况;在不同商品的应用效果上,均能有效地区分出各商品热度间的实际差距。
[Purpose/significance]Measuring the e-commerce commodity popularity from multi-dimension and analyzing the commodity popularity differences of different attribute combinations,can provide important references for consumer behavior and psycho⁃logical exploration,enterprise product design and development and business strategy formulation.[Method/process]Starting from the product review data,combined with Wilson interval product favorable rating ranking algorithm,the paper selects review time,sentiment value as important indicators,uses the entropy weight method to obtain the weight of each indicator to build a new popularity evaluation model,and applies the model based on multiple different commodity data.[Result/conclusion]Compared with the traditional popularity model,after adding time factor and emotional factor,the measurements of new multi-dimensional popularity model are more explanatory and more relevant to the actual situation;In the application effect of different commodities,it can effectively distinguish the actual gap be⁃tween the popularity of each commodity.
作者
游云贝
王毅
刘静
You Yunbei;Wang Yi;Liu Jing(School of Data Sciences,Zhejiang University of Finance&Economics,Hangzhou Zhejiang 311482)
出处
《情报探索》
2023年第6期85-91,共7页
Information Research
关键词
电商商品
文本挖掘
情感分析
热度测量
e-commerce product
text mining
sentiment analysis
popularity measurement
