ReINTEL Challenge 2020: A Comparative Study of Hybrid Deep Neural Network for Reliable Intelligence Identification on Vietnamese SNSs

TL;DR

This paper presents a hybrid deep neural network model that combines metadata and content analysis to improve the reliability assessment of social media posts, achieving high accuracy on Vietnamese SNS data.

Contribution

It introduces a novel multi-input model leveraging both post content and metadata, with advanced finetuning and training strategies for misinformation detection.

Findings

Achieved 0.9462 ROC-score on VLSP test set

Demonstrated effectiveness of combining metadata and content

Enhanced model performance with finetuning techniques

Abstract

The overwhelming abundance of data has created a misinformation crisis. Unverified sensationalism that is designed to grab the readers' short attention span, when crafted with malice, has caused irreparable damage to our society's structure. As a result, determining the reliability of an article has become a crucial task. After various ablation studies, we propose a multi-input model that can effectively leverage both tabular metadata and post content for the task. Applying state-of-the-art finetuning techniques for the pretrained component and training strategies for our complete model, we have achieved a 0.9462 ROC-score on the VLSP private test set.