融合权重惩罚BiGRU模型的网络敏感信息发现及实证研究  

Network Sensitive Information Discovery and Empirical Research Based on Punishing BiGRU Model by Fusion Weight

作　　者：吴树芳[1] 杨强 朱杰[2] Wu Shufang;Yang Qiang;Zhu Jie(School of Management,Hebei University,Baoding 071000;School of Mathematics and Information Science,Hebei University,Baoding 071000)

机构地区：[1]河北大学管理学院,保定071000 [2]河北大学数学与信息科学学院,保定071000

出　　处：《图书情报工作》2024年第13期144-153,共10页Library and Information Service

基　　金：全国教育科学“十三五”规划一般项目“高等教育促进区域创新驱动发展战略实现路径优化研究”(项目编号:BIA200203)研究成果之一。

摘　　要：[目的/意义]网络敏感信息发现对于净化网络空间和维护社会稳定具有重要意义。针对当前网络敏感信息发现研究忽略长距离上下文语义,从而导致发现性能欠佳的问题,提出敏感词融合权重惩罚BiGRU模型的网络敏感信息发现方法。[方法/过程]首先,得到敏感词的统计权重、类别权重和情感权重,并将三者融合得到敏感词的融合权重;其次,利用融合权重构建敏感词加权损失函数,以惩罚BiGRU模型对包含敏感词文本的错误发现;最后,基于惩罚后的BiGRU模型实现对网络敏感信息的发现。[结果/结论]在新浪微博真实数据集上的实证结果显示,与已有方法相比,提出的方法在精确率、召回率和F1值上均有一定提高。[Purpose/Significance]The discovery of network sensitive information is of great significance for purifying cyberspace and maintaining social stability.The current research on network sensitive information discovery ignores the long-distance contextual semantics,which leads to poor discovery performance.This paper proposes a network sensitive information discovery method based on punishing BiGRU model by fusion weight of sensitive terms.[Method/Process]Firstly,statistical weight,category weight and sentiment weight of sensitive terms are obtained,and the three are fused to obtain the fusion weight of sensitive terms.Secondly,the weighted loss function of sensitive terms is constructed by using the fusion weight for punishing misidentification of the text containing sensitive terms on BiGRU model.Finally,the discovery of network sensitive information is realized based on the punished BiGRU model.[Result/Conclusion]Empirical results on a real dataset from Sina Weibo indicate that compared to the existing methods,the proposed method has a certain improvement in Precision,Recall and F1 value.

关 键 词：长距离上下文语义 敏感词权重 敏感信息发现 BiGRU 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]