Financial Sentiment Analysis on Twitter During Covid-19 Pandemic in the UK

Abstract

The surge in Covid-19 cases seen in 2020 has caused the UK government to enact regulations to stop the virus’s spread. Along with other aspects like altered customer confidence and activity, the financial effects of these actions must be taken into account. This later can be studied from the user generated content posted on social networks such as Twitter. In this paper, we provide a supervised technique to analyze tweets exhibiting bullish and bearish sentiments, by predicting a sentiment class positive, negative, or neutral. Both machine learning & deep learning techniques are implemented to predict our financial sentiment class. Our research highlights how word embeddings, most importantly word2vec may be effectively used to conduct sentiment analysis in the financial sector providing favorable solutions. In addition, comprehensive research has been elicited between our technique and a lexicon-based approach. The outcomes of the study indicate how well Word2Vec model with deep learning techniques outperforms the others with an accuracy of 87%.