Date

2022-6-29

Author

Temiz, Orkun

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

286
views

149
downloads

Cite This

Social media has become a prevalent platform for consuming and sharing information online. The vast amounts of information, shared easily and rapidly by social media, have increased the demand for fact-checking. Misinformation threatens not only the reputation of individuals and organizations but also society. When the COVID-19 pandemic broke out, the concerns around misinformation, which threatens public health and society, have significantly increased. In this thesis, a new zero-shot fact extraction and verification pipeline for user posts related to COVID-19 against the medical articles is proposed. The pipeline comprises preprocessing of user posts, claim extraction, document retrieval, evidence selection, and verdict assignment components. The proposed pipeline not only labels the claim but also presents the related evidence set extracted from the pipeline regarding the claim, which gives interpretable results for the society about the claim. Also, it does not need to see previously labeled posts unlike numerous supervised studies in the literature instead; it uses the zero-shot capabilities of existing models. The proposed pipeline obtains on-par and stable performance compared with the state-of-art supervised techniques for classifying raw user posts (CoAID) and rumors collected from social media (COVID-19 Rumors Dataset). On the other hand, it achieves superior performance in detecting new emerging misinformation topics.

Subject Keywords

Data Mining, Fact Checking and Verification System, Natural Language Processing, Information Retrieval, Social Media

URI

https://hdl.handle.net/11511/98156

Collections

Graduate School of Informatics, Thesis