2020/09/17 by Kumud Chauhan, Chauhan, Kumud
Computer Science · Social Sciences · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Misinformation and Its Impacts #Sentiment Analysis and Opinion Mining #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2009.08590
openalex publication_date 2020/09/17 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Millions of people around the world are sharing COVID-19 related information\non social media platforms. Since not all the information shared on the social\nmedia is useful, a machine learning system to identify informative posts can\nhelp users in finding relevant information. In this paper, we present a BERT\nclassifier system for W-NUT2020 Shared Task 2: Identification of Informative\nCOVID-19 English Tweets. Further, we show that BERT exploits some easy signals\nto identify informative tweets, and adding simple patterns to uninformative\ntweets drastically degrades BERT performance. In particular, simply adding 10\ndeaths to tweets in dev set, reduces BERT F1- score from 92.63 to 7.28. We also\npropose a simple data augmentation technique that helps in improving the\nrobustness and generalization ability of the BERT classifier.\n