TriVerBERT-LLM: An Ensemble Multimodal Approach for Credibility Assessment of YouTube Video Transcripts via Logical Fallacy Detection and Claim Verification

DOI

10.2991/978-94-6463-866-0_7How to use a DOI?

Keywords

TriVerBERT-LLM; misinformation detection; logical fallacies; RoBERTa; SciBERT; Large Language Models; SpaCy

Abstract

Science misinformation on platforms like YouTube poses significant challenges to public understanding, necessitating reliable evaluation tools. This project presents TriVerBERT-LLM, a multimodal framework designed to assess the credibility of YouTube video transcripts through logical fallacy detection, scientific context evaluation, and claim verification. A fine-tuned Stacked model (RoBERTa and SciBERT) classifier, trained on the CoCoLoFa dataset, identifies logical fallacies and assigns confidence scores (Sm). SciBERT is employed to evaluate the scientific context, ensuring relevance and contextual accuracy. By spaCy, the claim extractions are then corroborated with a Large Language Model (LLM) to align with credible sources. The transcript’s final score (Sf) credibility is computed as a weighted sum of the confidence in logical fallacy, ratios of the fallacies, and ratios of verified claims. The higher the score, the more it is considered credible, with fewer logical fallacies and more confirmed claims. An approach toward holistic scoring methodology consists of linguistic analysis, scientific testing, and fact-checking verification within a strong framework for analyzing video content. The TriVerBERT-LLM has endless open doors for implementation as an application or browser extension that can help students measure the worth of the information, they find in online media to ensure transparency and accuracy in educational and scientific conversations.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Vidhushavarshini Sureshkumar
AU  - Aaron Don Kattasserry
AU  - S. Nivediitha
AU  - Suryakrishna Sukumar
PY  - 2025
DA  - 2025/10/31
TI  - TriVerBERT-LLM: An Ensemble Multimodal Approach for Credibility Assessment of YouTube Video Transcripts via Logical Fallacy Detection and Claim Verification
BT  - Proceedings of the International Conference on Intelligent Systems and Digital Transformation (ICISD 2025)
PB  - Atlantis Press
SP  - 55
EP  - 66
SN  - 2589-4919
UR  - https://doi.org/10.2991/978-94-6463-866-0_7
DO  - 10.2991/978-94-6463-866-0_7
ID  - Sureshkumar2025
ER  -