面向产品设计迭代的缺陷信息挖掘方法研究

Defect Information Mining Method for Product Design and Improvement

中文关键词: 产品设计创新  用户贡献内容  产品缺陷挖掘  主题模型  关联算法

英文关键词: product design innovation  user-generated content  product problem mining  topic model  Apriori analysis

中文摘要:

      目的 基于主题模型与关联算法，研究中文环境下服务于产品设计迭代的缺陷挖掘方法。方法 以网络产品社区作为研究对象，利用数据挖掘技术抓取社区用户的贡献内容，归纳面向产品缺陷分析的语法结构与候选词集，使用主题模型分析语料库中所包含的主题（产品属性）数量以及各个主题下的关键词分布，利用关联算法与紧凑规则找出每个主题下的强关联规则，解析后获得产品缺陷信息。结果 通过对小米4型智能手机用户贡献内容的实证分析，识别出了该产品用户声量最高的十四个关键问题。 结论 两组对比实验的结果表明，所提出的方法能较好地识别、定位用户贡献内容中所包含的产品缺陷/不足信息，拥有较高的准确率与召回率。该方法能够为企业开展产品设计创新活动提供必要的支持。

英文摘要:

      This work aims to explore the defect mining method for product design and improvement in the Chinese context based on topic model and Apriori analysis. Firstly, the online product community was selected as the research object and the user-generated contents were collected with the assist of data mining technique. Then, a set of trivial lexi-cal-Part-of-Speech patterns were proposed to prepare candidate corpus, and a topic model was adopted to find the optimal number of topics and get the words distributions in each topic. Finally, combined Apriori analysis and compactness rules, the expected strong rules in each topic were found out. Product defect information was obtained after analysis. Through the empirical analysis of the content contributed by Mi 4 smartphone users, the 14 key issues with the highest user volume of the product were identified. The results of the comparison experiment reveal that the proposed method can extract the product problems effectively. It has high accuracy and recall rate, can provide support for enterprises to initiate innovation activities.