Home // eKNOW 2021, The Thirteenth International Conference on Information, Process, and Knowledge Management // View article

Detecting Fake News Through Emotion Analysis

Authors:
Andrew Mackey
Susan Gauch
Kevin Labille

Keywords: fake news classification, misinformation, emotion analysis, natural language processing

Abstract:
Automating the detection of fake news is a challenging problem for the research community due to the various degrees of falsified information and ways in which it can be classified. In this work, we present a BERT-based machine learning model that captures linguistic and emotional features of a document to improve the task of classifying misinformation. The different types of psychological emotions are presented along with the methods used to capture the emotions of words. We investigate how different emotional features can augment existing data to facilitate the detection of fake news and improve upon existing baseline results. Our work demonstrates the ability for emotional features, when combined with other word-embedding models, such as BERT, to improve the performance benchmarks of fake news detection tasks.

Pages: 65 to 71

Copyright: Copyright (c) IARIA, 2021

Publication date: July 18, 2021

Published in: conference

ISSN: 2308-4375

ISBN: 978-1-61208-874-7

Location: Nice, France

Dates: from July 18, 2021 to July 22, 2021