Submissions:2019/Cure My FEVER : Building, Breaking and Fixing Models for Fact-Checking
This submission has been accepted for WikiConference North America 2019.
Title:
- Cure My FEVER : Building, Breaking and Fixing Models for Fact-Checking
Theme:
- Reliability of Information
+ Tech & Tools
Type of session:
- Presentation
Abstract:
Many tasks such as question answering and reading comprehension rely on information extracted from unreliable sources. These systems would thus benefit from knowing whether a statement from an unreliable source is correct. We present experiments on the Fact Extraction and VERification (FEVER) dataset, which involves selecting sentences from Wikipedia and predicting whether a claim is supported by those sentences, refuted, or there is not enough information. We describe results across three phases: the development of a fact-checking system, the creation of adversarial examples, and the improvement of the system in handling adversarial examples.
Academic Peer Review option:
- No
Author name:
E-mail address:
- tariq
cs.columbia.edu
Wikimedia username:
Affiliated organization(s):
- Columbia University
Estimated time:
- 30 min
Preferred room size:
Special requests:
Have you presented on this topic previously? If yes, where/when?:
If your submission is not accepted, would you be open to presenting your topic in another part of the program? (e.g. lightning talk or unconference session)