Accenture at CheckThat! 2020: If you say so: Post-hoc fact-checking of claims using transformer-based models

We introduce the strategies used by the Accenture Team for the CLEF2020! Lab, Task 1, on English and Arabic. This shared task evaluated a claim in social media text should be professionally fact checked. To journalist, a statement presented as fact, which would be of interest to a audience, requires professional fact-checking before dissemination. We BERT and RoBERTa models to identify claims in social media text a fact-checker should review, and rank these in priority order for fact-checker. For the English challenge, we fine-tuned a RoBERTa model and an extra mean pooling layer and a dropout layer to enhance to unseen text. For the Arabic task, we fine-tuned-language BERT models and demonstrate the use of back-translation to the minority class and balance the dataset. The work presented here was 1st place in the English track, and 1st, 2nd, 3rd, and 4th place in the track.