摘要
虚假评论检测包括有监督的、无监督的、基于分类的、基于排序的等,针对这些已有的算法,为提高整体虚假评论检测性能,结合这些算法,本文提出一种基于无监督集成学习的虚假评论检测方法,该方法利用投票表决原理来检测虚假评论,通过对不同算法生成的虚假评论排序列表进行集成,以获得一个新的整体的排序列表。实验结果表明,该模型的整体排序列表性能以大概率超越大部分被集成的排序列表性能。
Fake review detection includes supervised,unsupervised,classification-based,ranking-based detections,etc.For these proposed algorithms,in order to improve the overall fake review detection performance,combined with these algorithms,this paper proposes a fake review detection method based on unsupervised ensemble learning.This method uses the voting principle to detect fake reviews,and integrates the fake review ranking lists generated by different algorithms to obtain a new overall ranking list.The experimental results show that the overall ranking list performance of the model exceeds most of the integrated ranking lists with high probability.
作者
李慧
王琢
LI Hui;WANG Zhuo(Shenyang Ligong University,Shenyang 110159,China)
出处
《沈阳理工大学学报》
CAS
2021年第6期31-35,共5页
Journal of Shenyang Ligong University
关键词
虚假评论检测
无监督
集成学习
投票
fake review detection
unsupervised
ensemble learning
vote
