基于多示例学习的食品健康领域长文谣言检测  

Long rumor detection in the field of food health based on multiple instance learning

作　　者：谭励[1] 王舸 周丽娜 曹娟[2] TAN Li;WANG Ge;ZHOU Li-na;CAO Juan(School of Computer Science and 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：《计算机工程与设计》2022年第11期3101-3107,共7页Computer Engineering and Design

基　　金：北京市自然科学基金项目-海淀原始创新联合基金项目(L182007);国家自然科学基金项目(61702020)。

摘　　要：为解决长文本的谣言检测问题,以食品健康领域的长文本谣言为例,提出一种基于多示例学习的长文本谣言检测方法。将带有明显段落结构的长文本作为包,以文章段落作为示例,结合文本卷积神经网络与注意力机制建立MI-TCNNSA模型。实验结果表明,该模型在准确率、召回率与F1得分等通用指标取得优异成绩,高于传统通用方法。通过多示例学习改进后的短文本谣言检测方法与原方法对比,验证了多示例学习在长文本谣言检测的有效性,也为该问题提供一种思路。To solve rumors detection in long texts,long text rumors in the food health field were taken as an example and a long text rumors detection method based on multi-instance learning was proposed.The long text with apparent paragraph structure was brought as a package,the article paragraph was taken as an example,and the text convolutional neural network and the attention mechanism were combined to establish the MI-TCNN-SA.Experimental results show that the proposed model achieves excellent results in general accuracy,recall,and F1 score,which are higher than traditional available methods.The improved short text rumor detection method after multi-instance learning is compared with the original method,which verifies the effectiveness of multi-instance learning in the direction of long text rumor detection,and provides an idea for this direction.

关 键 词：食品健康领域 谣言检测 多示例学习 卷积神经网络 深度学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]