Work

Automated false claims detection using deep neural networks

Year: 2018
Type: dissertation
Abstract: Recently false claims and misinformation have become rampant in the web, affecting\nelection outcomes, stock markets, and various other societal issues. Consequently,\nfact-checking and claim verification websites such as snopes.com are becoming\nincreasingly popular and are also being integrated into news search engines such as\nGoogle news. However, these websites require expert analysis which is slow and\nnot scalable. Many recent papers have proposed machine learning methods using\nhandpicked linguistic and source-based cues to automate the claim verification\nprocess. In this thesis, we propose deep neural models which avoid tedious feature\nengineering and strong assumptions and yet detect false claims with high accuracy.\nTo achieve this, we propose a hybrid model which combines textual content of the\nnews articles as well as the reactions they receive in social media forums such as\nReddit. Using large-scale manually curated data from fact-checking websites such\nas snopes.com, politifact.com and emergent.info we perform extensive experiments\nto show that our models outperform the state-of-the-art CRF-based models.
Author: Erlend Rekve
Language: English
Open Access status: green
OpenAlex ID: W2892547540
Cites: 24
Cited by:
Related to: 10