结合交叉注意力的双通道恶意评论识别方法OA北大核心CSTPCD
Two-channel Malicious Comment Recognition Method Combined with Cross-attention Mechanism
恶意评论识别本质上是一个文本分类的问题.相较传统的文本分类,恶意评论往往伴随着表达方式更微妙且随意的特点,使得传统文本分类网络识别精度不高、识别效果不好,无法达到需求.为解决上述问题,本文提出一种结合交叉注意力机制的双通道文本分类网络(Two-channel text classification network combined with cross-at-tention mechanism,CA2TC).该模型同时使用图卷积神经网络(Grap…查看全部>>
The detection of malicious comments is essentially a text classification problem.Compared to typical text classification,malicious comments are often accompanied by more subtle and unpredictable expressions,which results in low identification accura-cy,poor recognition effect,and inability to satisfy demands.To tackle the aforementioned issues,this paper proposes a two-channel text classification network combined with cross-attention mechanism(CA2TC),which e…查看全部>>
张琳钰;卢益清
北京信息科技大学 信息管理学院,北京 100192北京信息科技大学 信息管理学院,北京 100192
计算机与自动化
恶意评论识别双通道图卷积神经网络双向长短期记忆网络交叉注意力机制
malicious comment recognitiontwo-channelGCNBiLSTMcross-attention mechanism
《山西大学学报（自然科学版）》 2024 (4)
751-760,10
国家自然科学基金(U1936111)
评论