A Scalable Streaming Big Data Architecture for Real-Time Sentiment Analysis

The systems with a short window of opportunity for actions and decisions require developing solutions providing real-time streaming analytics. Real-time big data streaming analytics is a challenging task. In this paper, we propose a streaming big data architecture for real-time social network analysis. As a case study, we investigated the relation between the public opinions on social media about cryptocurrencies and the changes in their prices using lexicon-based sentiment analysis approaches with the goal of assessing the feasibility of predicting the prices of cryptocurrencies. Two different approaches with two lexicons were used for sentiment analysis score calculations to assess the consistency of correlation measures on the collected dataset. Our model indicates that the prediction of cryptocurrency price changes using lexicon-based sentiment analysis methods is not reliable.