Fact Extraction and VERification (FEVER) is a recently introduced task that consists of the following subtasks (i) document retrieval, (ii) sentence retrieval, and (iii) claim verification. In this work, we focus on the subtask of sentence retrieval. Specifically, we propose an evidence-aware transformer-based model that outperforms all other models in terms of FEVER score by using a subset of training instances. In addition, we conduct a large experimental study to get a better understanding of the problem, while we summarize our findings by presenting future research challenges.

In this paper, we present our system designed to address the W-NUT 2020 shared task for COVID-19 Event Extraction from Twitter. To mitigate the noisy nature of the Twitter stream, our system makes use of the COVID-Twitter-BERT (CT-BERT), which is a language model pre-trained on a large corpus of COVID-19 related Twitter messages. Our system is trained on the COVID-19 Twitter Event Corpus and is able to identify relevant text spans that answer pre-defined questions (i.e., slot types) for five COVID-19 related events (i.e., TESTED POSITIVE, TESTED NEGATIVE, CAN-NOT-TEST, DEATH and CURE & PREVENTION). We have experimented with different architectures; our best performing model relies on a multilabel classifier on top of the CT-BERT model that jointly trains all the slot types for a single event. Our experimental results indicate that our Multilabel-CT-BERT system outperforms the baseline methods by 7 percentage points in terms of micro average F1 score. Our model ranked as 4th in the shared task leaderboard.

Deep-learning-based models have been successfully applied to the problem of detecting fake news on social media. While the correlations among news articles have been shown to be effective cues for online news analysis, existing deep-learning-based methods often ignore this information and only consider each news article individually. To overcome this limitation, we develop a graph-theoretic method that inherits the power of deep learning while at the same time utilizing the correlations among the articles. We formulate fake news detection as an inference problem in a Markov random field (MRF) which can be solved by the iterative mean-field algorithm. We then unfold the mean-field algorithm into hidden layers that are composed of common neural network operations. By integrating these hidden layers on top of a deep network, which produces the MRF potentials, we obtain our deep MRF model for fake news detection. Experimental results on well-known datasets show that the proposed model improves upon various state-of-the-art models.