作　　者： (）;

机构地区： 五邑大学经济管理学院

出　　处： 《江苏商论》 2020年第4期22-24,共3页

摘　　要： 用户的购物兴趣点具有时效性的特点,即随着时间的推移,产品的更新加快和替代品的出现,用户感兴趣的产品会发生改变。挖掘用户的兴趣并进行有效的产品推荐已经是电商平台研究的重点。本文致力于为电商平台提供一种能够获取用户的兴趣随着时间变化的模型,以自然语言处理为工具,以微博上对产品进行评论的数据为样本。通过词频统计提取样本中的关键词,与事先建立起来的情感词库进行相似度分析,得到用户评论产品时的感情色彩。根据感情色彩来决定是否添加或从原有的模型中淘汰,从而获得兴趣演化模型,为电商平台进行产品推荐提供指导性的建议。最后为了提高模型的精确性,以用户在Web端的动作行为为指令,根据隐式反馈行为的特性,对动作行为赋加权重值,优化兴趣演化模型。 The shopping interest points of users have the characteristics of timeliness,that is,as time goes by,product updates accelerate and the emergence of alternatives,the products that users are interested in will change.Digging users'interests and making effective product recommendations have been the focus of e-commerce platform research.This article is dedicated to providing an e-commerce platform with a model that can capture the user's interest over time,using natural language processing as a tool,and using Weibo's data for product reviews as examples.The keywords in the sample are extracted through word frequency statistics,and the similarity analysis is performed with the emotional lexicon established in advance to obtain the emotional color of the user when reviewing the product.Based on the emotional color,whether to add or eliminate from the original model is determined.In this way,an interest evolution model is obtained,which provides guiding suggestions for product recommendation on e-commerce platforms.Finally,in order to improve the accuracy of the model,the user's action behavior on the web side is used as an instruction.According to the characteristics of implicit feedback behavior,weighted action values are assigned to optimize the interest evolution model.

关 键 词： 时效性 词频统计 自然语言处理 相似度 兴趣演化

领　　域： []