基于卷积神经网络的食品安全领域谣言检测方法  

RUMOR DETECTION METHOD IN FOOD SAFETY FIELD BASED ON CONVOLUTIONAL NEURAL NETWORK

作　　者：周丽娜 谭励[1] 曹娟[2] 马子豪 吕芯悦 Zhou Li’na;Tan Li;Cao Juan;Ma Zihao;Lü Xinyue(School of Computer and Information Engineering,Beijing Technology and Business University,Beijing 100048,China;Institute of Computing Technology Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]北京工商大学计算机与信息工程学院,北京100048 [2]中国科学院计算技术研究所,北京100190

出　　处：《计算机应用与软件》2022年第3期45-50,115,共7页Computer Applications and Software

基　　金：国家自然科学基金项目(61702020);国家重点研发计划项目(2016YFD0401104);北京市自然科学基金项目(4172013);北京市自然科学基金-海淀原始创新联合基金项目(L182007)。

摘　　要：近年来,食品安全已经成为政府和社会最关心的问题之一。与此同时,大量食品安全领域的谣言也在社交媒体平台上疯狂传播。现有的研究大多针对通用领域的谣言进行检测,忽视了不同领域谣言之间存在的差异性。针对食品安全这一领域谣言的特点,提出一种基于卷积神经网络的食品安全领域谣言检测方法。用领域实体抽取策略得到的实体特征与文本卷积神经网络提取的深度文本特征做拼接融合,再进行谣言判断。实验结果表明,在食品安全领域谣言数据集上,采用该方法检测效果明显优于通用的谣言检测方法,在准确率、召回率和F1值等评价指标上都表现最优。Food safety has become one of the most concerned issues for government and society in recent years.Meanwhile,numerous rumors of food safety also bloom and spread wildly on the open social media platforms.Most of the existing researches focus on rumors in the general field,so it ignores the differences between rumors in different fields.Aiming at the characteristics of rumors in food safety,this paper proposed a rumor detection method in food safety field based on convolutional neural network.The entity characteristics obtained by the domain entity extraction strategy were merged with the deep text features extracted by the text convolutional neural network,then,judged rumors.The experimental results show that the method in this paper is significantly better than the general rumor detection methods on food safety rumor data set.It has the best performance in precision,recall and F1 score.

关 键 词：深度学习 领域实体特征 卷积神经网络 特征融合 食品安全领域谣言检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]