Abstract: To obtain users'feedback on phones'defects from social media,based on the improvement of current social media analytics framework,a creative framework of phones defect discovery from social media was put forward by taking characteristics of various platform factors into consideration.With analysis and comparison,the support vector machine classification method was used to identify the emotion of reviews from the social media,and defects corpus were created.On this basis,the defects of phones were analyzed by themed model,and different defects clustering were formed.The relevant experiments were operated,and the result showed that the proposed method had better accuracy and recall,which could identify the phones'defects from Chinese social media effectively.

Key words: social media analytics, defect discovery, text mining, phones

摘要： 为了从社交媒体中获取用户反映手机产品缺陷的数据信息,在对现有社交媒体分析框架进行改进的基础上,结合不同的平台特征,构建了基于社交媒体分析的手机产品缺陷识别框架;在综合分析比较后采用信息增益作为特征评估函数,利用支持向量机分类方法对社交媒体数据进行情感识别,构建手机产品缺陷语料集;基于主题模型算法对手机产品缺陷进行聚类分析,形成不同的缺陷内容聚类。相关实验研究结果表明,该方法能够有效地识别出中文社交媒体中的手机缺陷,具有较高的准确率和召回率。

关键词: 社交媒体分析, 缺陷识别, 文本挖掘, 手机