• 首页
  • 期刊简介
  • 编委会
  • 投稿指南
  • 收录情况
  • 杂志订阅
  • 联系我们
引用本文:石玉鑫,杨泽青,赵志滨,姚 兰.一种面向商品评价对象挖掘的领域词典构建法[J].软件工程,2019,22(1):1-7.【点击复制】
【打印本页】   【下载PDF全文】   【查看/发表评论】  【下载PDF阅读器】  
←前一篇|后一篇→ 过刊浏览
分享到: 微信 更多
一种面向商品评价对象挖掘的领域词典构建法
石玉鑫,杨泽青,赵志滨,姚 兰
(东北大学计算机科学与工程学院,辽宁 沈阳 110819)
摘 要: 通过挖掘商品评论中的评价对象,可以得知用户更关心商品哪些方面的属性,从而帮助企业改进商品, 帮助用户选择商品。因此,商品评价对象的挖掘具有重要的意义。本文提出了一种用于商品评价对象挖掘的领域词典构 建方法:首先基于LDA模型,提出了一种领域基础词典的构建方法;然后,分别提出了基于词汇之间的PMI值和基于依 存句法分析的领域词典扩充方法。本文基于京东商城的洗衣液产品真实评论数据集,使用构建的词典分别进行了一级标 签评价对象挖掘和二级标签评价对象挖掘的实验。实验结果表明,本文提出的方法在进行评价对象挖掘时具有良好的性 能;相比一级标签评价对象,扩充后的词典对二级标签评价对象挖掘的效果有更好的提升。
关键词: 领域词典;对象挖掘;商品评论;LDA;PMI
中图分类号: TP391    文献标识码: A
基金项目: 本文受the National Key R&D Program of China under grant(2018YFB1004700)资助.
A Method on Domain Dictionary Construction for Object Mining on Commodity Comments
SHI Yuxin,YANG Zeqing,ZHAO Zhibin,YAO Lan
( School of Computer Science and Engineering, Northeastern University, Shenyang 110819, China)
Abstract: Enterprises hope to be aided by object mining on comments of their products,which reveals the clients' concerns,to improve their manufacturing.This object mining also makes sense to subsequent consumers while they are making their choice.Therefore,it is significant to mine objects of a comment.This paper proposes a method on domain dictionary construction for object mining on comments of commodity:Firstly,a method based on the LDA model,a basic domain dictionary is proposed;then,the domain dictionary expansion methods based on the PMI value of words and dependency parsing are proposed respectively.Data applied for experiments in this paper is from detergent sale data of JD.COM.The dictionaries are applied on this data set for the first-level and second-level label object mining.The experimental results prove the proposed method’s great potential in object mining.Compared with the first-level label object mining,the extensive dictionary has improved the second-level label object mining.
Keywords: domain dictionary;object mining;commodity comment;LDA;PMI


版权所有:软件工程杂志社
地址:辽宁省沈阳市浑南区新秀街2号 邮政编码:110179
电话:0411-84767887 传真:0411-84835089 Email:semagazine@neusoft.edu.cn
备案号:辽ICP备17007376号-1
技术支持:北京勤云科技发展有限公司

用微信扫一扫

用微信扫一扫