The advent of the internet has changed the way people access and share information, making it easier for malicious individuals to spread biased, unreliable or false news. Recent technological advances, including artificial intelligence (AI) models that can generate realistic texts, audio recordings and images, are further contributing to this wave of misinformation.
In recent years, telling real and fake news apart has become increasingly difficult, which creates the perfect breeding ground for ignorance, confusion and polarization. Developing effective tools to rapidly identify and take down online fake news from popular websites and search engines is thus of crucial importance.
Researchers at National Yang Ming Chiao Tung University, Chung Hua University and National Ilan University recently developed a new multi-modal model that could help to quickly detect online fake news. This model, introduced in a paper published in Science Progress, can identify fake news by processing both textual and visual data, as opposed to a single type of data.
“The existing literature primarily focuses on analyzing individual features in fake news, overlooking multimodal feature fusion recognition,” Szu-Yin Lin, Yen-Chiu Chen and their colleagues wrote in their paper.
“Compared to single-modal approaches, multimodal fusion allows for a more comprehensive and enriched capture of information from different data modalities (such as text and images), thereby improving the performance and effectiveness of the model. This study proposes a model using multimodal fusion to identify fake news, aiming to curb misinformation.”
To improve fake news detection, Lin, Chen and their colleagues set out to develop an alternative model that would simultaneously analyze both the textual and visual features of online news. The model they developed starts by cleaning data, to then extract these features from the clean data.
The researchers’ model integrates textual and visual information using various fusion strategies, including early fusion, joint fusion and late fusion techniques. In initial tests, this multi-modal approach was found to perform remarkably well, detecting fake news better than well-established single-modality techniques, including BERT.
The team’s multi-modal model was tested on the Gossopcop and Fakeddit datasets, both of which are often used to train models for fake news detection. On these same two datasets, single-modality models were previously found to detect fake news with unsatisfactory accuracies of 72% and 65%, respectively.
“The proposed framework processes textual and visual information through data cleaning and feature extraction before classification,” wrote Lin, Chen and their colleagues. “Fake news classification is accomplished through a model achieving accuracy of 85% and 90% in the Gossipcop and Fakeddit datasets, with F1-scores of 90% and 88%, showcasing its performance.
“The study presents outcomes across different training periods, demonstrating the effectiveness of multimodal fusion in combining text and image recognition for combating fake news.”
The promising findings gathered by Lin, Chen and his colleagues highlight the potential of multi-modal fusion models for fake news detection. They could thus encourage other teams to develop similar models that rely on multiple modalities.
In the future, the new model could also be tested on more datasets and real-world data. Eventually, it could contribute to worldwide efforts aimed at tackling and reducing online misinformation.
More information:
Szu-Yin Lin et al, Text–image multimodal fusion model for enhanced fake news detection, Science Progress (2024). DOI: 10.1177/00368504241292685
© 2024 Science X Network
Citation:
Alternative model can identify fake news by processing both textual and visual data (2024, November 4)
retrieved 4 November 2024
from https://techxplore.com/news/2024-11-alternative-fake-news-textual-visual.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
