分离视角下产品迭代设计的决策信息挖掘方法  

Decision information mining method of product iteration design from the perspective of separation

作　　者：黄晟青 张杰[1] 李明 顾全 HUANG Sheng-qing;ZHANG Jie;LI Ming;GU Quan(School of Art Design and Media,East China University of Science and Technology,Shanghai 200237)

机构地区：[1]华东理工大学艺术设计与传媒学院,上海200237

出　　处：《机械设计》2021年第5期138-144,共7页Journal of Machine Design

基　　金：国家社会科学基金艺术学资助项目(20BH154);上海市社会科学基金资助项目(2019ZJX002)。

摘　　要：为将在线评论中用户体验信息准确应用于产品迭代设计中,从分离视角综合运用数据挖掘技术构建了一个用户需求挖掘模型。通过归纳面向用户分离的语法结构与候选词集,提取消费-使用角色分离的评论,应用概率主题模型(Latent Dirichlet Allocation,LDA)对分类语料库进行主题提取,解析并获得产品设计迭代决策的关键信息。通过对4组不同品牌的老年人智能腕表在线评论进行实证分析,快速获取了用户反馈中4组产品各自不同的竞争优势及目标产品4项缺陷,通过对比分析,对目标产品迭代方向进行定位。该框架下的LDA挖掘方法克服了传统词频统计方法噪声较大的缺陷,提高了在线评论中挖掘用户体验信息的准确性,为企业产品开发人员提供参考,缩短产品的开发周期。In order to apply user experience information generated from online comments precisely in product iterative design system,data mining technology was used to establish a user requirement mining model from separation perspective. The consumption-use role separation comments were extracted through the summary of user separation oriented. Latent Dirichlet Allocation was used to extract the subject of sorting language database. The key information of product design iteration decision was obtained.Four elderly smart watches from different brands were performed empirical analysis of online comments. The different competitive advantages and four defects of target 4 groups of products were obtained raplidly. The iteration directions of target products were located through the comparison. The LDA mining method under the framework overcame the noise in traditional word frequency statistics problem,improved the mining accuracy from online comments. This could provide refereces for product developer and shorten product development period.

关 键 词：设计迭代 在线用户评论 用户角色分离 决策信息 主题模型 

分 类 号：TH166[机械工程—机械制造及自动化] TP391.1[自动化与计算机技术—计算机应用技术]