产品评论挖掘可视化实验平台的开发

针对目前研究人员已经提出多种中文评论挖掘方法,缺乏统一的评论实验数据集的现状,首先从知名网站上随机抽取手机评论,经过垃圾去除、手工标注,最终构造出手机领域的评论挖掘实验数据集.基于实验数据集构造出手机领域的情感词库,并利用模式匹配方法建立了产品特征粒度树,开发出一个可视化平台,研究人员可以直接用其检验挖掘方法的效果,也可以对不同的挖掘方法进行客观比较.

A visualization platform development for product review mining

Up to now,researchers have proposed a variety of mining methods for Chinese reviews.However,there are not any unified review experimental data sets now.For this situation,the paper extracts reviews about mobiles from four famous websites randomly.After spam reviews removing and artificial marking,an experimental data set in the field of mobiles for product reviews mining is constructed.Later,we build an emotional lexicon based on the experimental data set and a feature-granularity tree by the pattern matching method.Then a visualization platform is developed.Researchers can not only use it to perform their experiments directly,but do some objective comparisons of different mining methods.