Detecting fake news by enhanced text representation with multi-EDU-structure awareness

作者：

Highlights：

• Using EDU expressing content with an intermediate granularity to detect fake news.

• A sequence-based EDU representation is gotten by TextCNN.

• A graph-based EDU representation is obtained by EDU dependency graph and RGAT.

• Text representation is gotten by GRU with Global Attention mechanism.

• Extensive experiments on four cross-source datasets demonstrate the superiority.

摘要

•Using EDU expressing content with an intermediate granularity to detect fake news.•A sequence-based EDU representation is gotten by TextCNN.•A graph-based EDU representation is obtained by EDU dependency graph and RGAT.•Text representation is gotten by GRU with Global Attention mechanism.•Extensive experiments on four cross-source datasets demonstrate the superiority.

论文关键词：Fake news detection,EDU,Sequential structure,Dependency graph structure,TextCNN,RGAT

论文评审过程：Received 19 August 2021, Revised 5 May 2022, Accepted 4 June 2022, Available online 14 June 2022, Version of Record 22 June 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117781