Combating the infodemic: COVID-19 induced fake news recognition in social media networks.

COVID-19 has caused havoc globally due to its transmission pace among the inhabitants and prolific rise in the number of people contracting the disease worldwide. As a result, the number of people seeking information about the epidemic via Internet media has increased. The impact of the hysteria that has prevailed makes people believe and share everything related to illness without questioning its truthfulness. As a result, it has amplified the misinformation spread on social media networks about the disease. Today, there is an immediate need to restrict disseminating false news, even more than ever before. This paper presents an early fusion-based method for combining key features extracted from context-based embeddings such as BERT, XLNet, and ELMo to enhance context and semantic information collection from social media posts and achieve higher accuracy for false news identification. From the observation, we found that the proposed early fusion-based method outperforms models that work on single embeddings. We also conducted detailed studies using several machine learning and deep learning models to classify misinformation on social media platforms relevant to COVID-19. To facilitate our work, we have utilized the dataset of "CONSTRAINT shared task 2021". Our research has shown that language and ensemble models are well adapted to this role, with a 97% accuracy.

Fighting hate speech from bilingual hinglish speaker's perspective, a transformer- and translation-based approach.

Many people have begun to use social media platforms due to the increased use of the Internet over the previous decade. It has a lot of benefits, but it also comes with a lot of risks and drawbacks, such as Hate speech. People in multilingual societies, such as India, frequently mix their native language with English while speaking, so detecting hate content in such bilingual code-mixed data has drawn the larger interest of the research community. The majority of previous work focuses on high-resource language such as English, but very few researchers have concentrated on the mixed bilingual data like Hinglish. In this study, we investigated the performance of transformer models like IndicBERT and multilingual Bidirectional Encoder Representation(mBERT), as well as transfer learning from pre-trained language models like ULMFiT and Bidirectional encoder Representation(BERT), to find hateful content in Hinglish. Also, Transformer-based Interpreter and Feature extraction model on Deep Neural Network (TIF-DNN), is proposed in this work. The experimental results found that our proposed model outperforms existing state-of-art methods for Hate speech identification in Hinglish language with an accuracy of 73%.