Not-NUTs at W-NUT 2020 Task 2: A BERT-based System in Identifying Informative COVID-19 English Tweets

As of 2020 when the COVID-19 pandemic is full-blown on a global scale,\npeople's need to have access to legitimate information regarding COVID-19 is\nmore urgent than ever, especially via online media where the abundance of\nirrelevant information overshadows the more informative ones. In response to\nsuch, we proposed a model that, given an English tweet, automatically\nidentifies whether that tweet bears informative content regarding COVID-19 or\nnot. By ensembling different BERTweet model configurations, we have achieved\ncompetitive results that are only shy of those by top performing teams by\nroughly 1% in terms of F1 score on the informative class. In the\npost-competition period, we have also experimented with various other\napproaches that potentially boost generalization to a new dataset.\n