Categorizing Crypto Conversations: Microblogging Data and Their Differential Impact on Cryptocurrency Markets

Abstract

This study delves into the intricate relationship between social media sentiment and the cryptocurrency market and focuses on the diverse impact across various cryptocurrency types. By leveraging the sectional news on CoinDesk, we employ a novel categorization method to systematically classify microblog sentiments on Twitter into three pertinent categories: market/business, technology, and policy. We examine the influence of the categorized sentiments on the price movement of both established cryptocurrencies and the nascent segment of ‘meme coins.’ Unlike previous studies that rely heavily on manual data labeling, our study uses news content with sectional information as the training data and an ensemble method to increase the performance of the classifiers. Our analysis incorporates lexicon-based algorithms and machine learning techniques to assess the sentiments and reveals that microblog sentiments can exert differential effects on cryptocurrencies, thereby highlighting the nuanced relationship between public sentiment and market dynamics. Our research enriches the field of decision support systems by showcasing the utility of advanced automated sentiment analysis in capturing the sentiment-driven forces at play in the cryptocurrency market. The insights of this study, which considers the multifaceted nature of the influence of sentiment on the volatile and evolving market, can aid investors and market analysts in making better informed decisions.

Keywords: Big Data, Cryptocurrency, Machine Learning, Microblog, Sentiment Analysis