What is it about?
The work proposes the MVAE-FakeNews, a Multimodal Variational Autoencoder that represents fake news texts. The method explores two modalities: (i) embeddings generated by the BERT that capture syntactic and semantic characteristics of texts; and (ii) fake news topic information, since fake news is more frequent in well-defined topics such as politics, pandemics, science, religion, etc. Once represented, the fake news is detected through One-Class Learning (OCL). In OCL, the user defines a class of interest and provides a set of examples for the algorithm to train. Thus, when a new example is presented to the OCL algorithm, the example can be assigned to the class of interest or not. Therefore, it is possible to minimize the user's label effort, as he/she won't label real news.
Featured Image
Photo by Sammy Williams on Unsplash
Why is it important?
In the One-Class Learning scenario, MVAE-FakeNews performed better in ten out of twelve scenarios than nine other methods (eight unimodal and one multimodal). Thus, the proposed method proved to be promising to represent the texts in the OCL scenario to detect fake news.
Read the Original
This page is a summary of: Learning Textual Representations from Multiple Modalities to Detect Fake News Through One-Class Learning, September 2021, ACM (Association for Computing Machinery),
DOI: 10.1145/3470482.3479634.
You can read the full text:
Contributors
The following have contributed to this page







