Testing the Generalization of Neural Language Models for COVID-19 Misinformation Detection

Abstract

A drastic rise in potentially life-threatening misinformation has been a by-product of the COVID-19 pandemic. Computational support to identify false information within the massive body of data on the topic is crucial to prevent harm. Researchers proposed many methods for flagging online misinformation related to COVID-19. However, these methods predominantly target specific content types (e.g., news) or platforms (e.g., Twitter). The methods’ capabilities to generalize were largely unclear so far. We evaluate fifteen Transformer-based models on five COVID-19 misinformation datasets that include social media posts, news articles, and scientific papers to fill this gap. We show tokenizers and models tailored to COVID-19 data do not provide a significant advantage over general-purpose ones. Our study provides a realistic assessment of models for detecting COVID-19 misinformation. We expect that evaluating a broad spectrum of datasets and models will benefit future research in developing misinformation detection systems.

J. P. Wahle and N. Ashok—Equal contribution.

A Appendix

1.1 A.1 Dataset Details

COVID-19 Open Research Dataset (CORD-19) [33] is the largest open source dataset about COVID-19 and coronavirus-related research (e.g. SARS, MERS). CORD-19 is composed of more than 280K scholarly articles from PubMed,¹³ bioRxiv,¹⁴ medRxiv,¹⁵ and other resources maintained by the WHO.¹⁶ We use this dataset to extend the general pre-training from selected neural language models (cf. Sect. 3) into the COVID-specific vocabulary and features.

Covid-19 heAlthcare mIsinformation Dataset (CoAID) [7] focuses on healthcare misinformation, including fake news on websites, user engagement, and social media. CoAID is composed of 5 216 news articles, 296 752 related user engagements, and 958 posts about COVID-19, which are broadly categorized under the labels true and false.

Twitter Stance Dataset (COVID-CQ) [19] is a dataset of user-generated Twitter content in the context of COVID-19. More than 14K tweets were processed and annotated regarding the use of Chloroquine and Hydroxychloroquine as a valid treatment or prevention against the coronavirus. COVID-CQ is composed of 14 374 tweets from 11 552 unique users labeled as neutral, against, or favor.

ReCOVery [36] explores the low credibility of information on COVID-19 (e.g., bleach can prevent COVID-19) by allowing a multimodal investigation of news and their spread on social media. The dataset is composed of 2 029 news articles on the coronavirus and 140 820 related tweets labeled as reliable or unreliable.

CMU-MisCov19 [17] is a Twitter dataset created by collecting posts from unknowingly misinformed users, users who actively spread misinformation, and users who disseminate facts or call out misinformation. CMU-MisCov19 is composed of 4 573 annotated tweets divided into 17 classes (e.g., conspiracy, fake cure, news, sarcasm). The high number of classes and their imbalanced distribution make CMU-MisCov19 a challenging dataset.

COVID19FN¹⁷ is composed of approximately 2 800 news articles extracted mainly from Poynter¹⁸ categorized as either real or fake.

1.2 A.2 Model Details

General-Purpose Baselines. BERT [8] mainly captures general language characteristics using a bidirectional Masked Language Model (MLM) and Next Sentence Prediction (NSP) tasks. RoBERTa [16] improves BERT with additional data, compute budgets, and hyperparameter optimizations. RoBERTa also drops the NSP as it contributes little to the model representation. BART [15] optimizes an auto-regressive forward-product and auto-encoding MLM objective simultaneously. DeBERTa [11] improves the attention mechanism using a disentanglement of content and position.

Intermediate Pre-Trained. SciBERT [2] optimizes the MLM for 1.14M randomly selected papers from Semantic Scholar¹⁹. BioClinicalBERT [1] specializes on 2M notes in the MIMIC-III database [13], a collection of disidentified clinical data. BERTweet [21] optimizes BERT on 850M tweets each containing between 10 and 64 tokens.

COVID-19 Intermediate Pre-Trained. COVID-Twitter-BERT [20] (CT-BERT) uses a corpus of 160M tweets for domain-specific pre-training and evaluates the resulting model’s capabilities in sentiment analysis, such as for tweets about vaccines. BioClinicalBERT [1] fine-tunes BioBERT [14] into clinical narratives in the hope to incorporate linguistic characteristics from both the clinical and biomedical domains.

Cui et al. [7] propose CoAID and investigate the misinformation detection task by comparing traditional machine learning (e.g., logistic regression, random forest) and deep learning techniques (e.g., GRU). In a similar layout, Zhou et al. [36] compare traditional statistical learners, such as SVM and neural networks (e.g., CNN), to classify news as credible or not. In both studies, the results show deep learning architectures as the most prominent options.

1.3 A.3 Evaluation Details

Pre-Training. We use the data from the abstracts of the CORD-19 dataset for pre-training. For pre-processing the CORD-19 abstract data, we consider only alphanumerical characters. We use a sequence length of 128 tokens, which reduces training time while being competitive to longer sequence lengths when fine-tuning [24]. We mask words randomly with a probability of .15, a common configuration for Transformers [8, 11], and perform the MLM with the following remaining parameters: a batch size of 16 for all the base models, and eight for the large models, the Adam Optimizer ($\alpha = 2e-5$, $\beta _1 = .9$, $\beta _2 = .999$, $\epsilon = 1e-8$), and a maximum of five epochs. All experiments were performed on a single NVIDIA GeForce GTX 1080 Ti GPU with 11 GB of memory.

Fine-Tuning. The classification model applies a randomly initialized fully-connected layer to the aggregate representation of the underlying Transformer (e.g., [CLS] for BERT) with dropout ($p=.1$) to learn the annotated target classes with cross-entropy loss for five epochs and with a sequence length of 200 tokens. We use the same configuration of the optimizer as in pre-training.