Abstract: Product review mining is the process of finding information from product reviews on the Web through natural language process technology.It is a rising field that is the sub field of unstructured data mining from plain text.The information mined from product reviews can help manufacturers to improve their product，and help user to buy product with more rationality.A survey of product review mining is discussed.Firstly，the framework of product review mining is analyzed.Then，the tasks of product review mining that include subjective sentence identify，product feature extracting，user attitude extracting，polarity classifying and mining result show are also described in detail，and finally the future reseach directions about product review mining are pointed out.

Key words: textual pattern, subjective feature, word polarity

摘要： 产品评论挖掘是以Web上用户发表的产品评论为挖掘对象，采用自然语言处理技术，从大量的文本数据中发现关于产品的功能和性能的评价信息的过程。产品评论挖掘是一个新兴的研究领域，是对自然语言描述的无结构数据进行数据挖掘的典型代表。产品评论中挖掘得到的信息不仅可以帮助生产厂商改进产品，还可以帮助用户合理的购买产品。对产品评论挖掘进行了全面深入地讨论，介绍了产品评论挖掘系统的通用框架，然后对产品特征提取、主观句定位、用户态度提取、态度极性判定、挖掘结果显示这5个子任务进行了详细地阐述，最后介绍了产品评论挖掘的最新方向。

关键词: 用户评论, 主观特征, 词语极性