一种基于启发式规则的半监督垃圾评论分类方法

发布时间：2018-11-20 21:02

【摘要】：互联网业已深入每个人的生活,团购平台、在线商店、在线消费等形式的电子商务平台已成为人们时下最流行的消费方式。几乎所有的电商平台都允许和鼓励用户在消费之后对产品或者服务进行评论,而且用户评论对潜在消费者和商家都具有极高的价值。这使得广告、虚假评论等形式的垃圾评论被人为地夹杂在用户评论中,以期达到虚假宣传、推广产品或者诋毁其他商家信誉的目的。垃圾评论检测和分析便是在这样一种应用背景下,研究如何有效地排除垃圾评论干扰,发挥有效评论价值的方法。针对COAE2015设定的垃圾评论识别任务,利用其提供的语料资源,设计了一种基于启发式规则的半监督垃圾评论分类方法。实验结果证明,提出的方法可以有效地识别垃圾评论,同时能够保持对有效评论的识别精度。
[Abstract]:The Internet has become the most popular consumption mode in the form of e-commerce platform, group purchase platform, online store, online consumption and so on. Almost all e-commerce platforms allow and encourage users to comment on products or services after consumption, and user reviews are highly valuable to potential consumers and merchants. This makes spam comments in the form of advertisements, false comments and other forms are artificially mixed in user reviews, in order to achieve the purpose of false publicity, promotion of products or defamation of the credibility of other businesses. Garbage comment detection and analysis is a method to study how to effectively eliminate the interference of garbage comment and give full play to its value under such an application background. Aiming at the garbage comment recognition task set by COAE2015, a semi-supervised classification method based on heuristic rules is designed by using the corpus resources provided by it. The experimental results show that the proposed method can recognize garbage comments effectively and maintain the recognition accuracy of the effective comments at the same time.
【作者单位】：山西大学计算机与信息技术学院;计算智能与中文信息处理教育部重点实验室;
【基金】：国家自然科学基金资助项目(61573231,61632011,61672331) 山西省科技基础条件平台计划资助项目(2015091001-0102)
【分类号】：TP391.1

【相似文献】