Predictive Visual Analysis of Twitter Big Data Originated from Cloud Using Machine Learning Algorithms

摘要

Social networks are one of the main sources that generate streams of big data. This in turn creates challenges due to the fact that the data is being collected from servers that are distributed in different geographical locations, often termed as "Cloud". Hence it is suitable to process the data on the cloud itself. In this work we perform predictive visual analysis of big data originated from social networks by considering the micro-blog "Twitter". We perform both graph and non-graph analytics of tweets collected during the key event IPL10 20-20 cricket final match, using machine learning algorithms (Naive Bayes). Also we perform weighted word cloud visualizations, which give improved semantic insights. The results obtained from our work can also help an advertising company to target the key nodes in the network to maximize its coverage, which in turn helps in viral marketing. Existing visualization techniques in the literature highlight topical changes, concept relationships, or physical locations. But the textual content is either unseen or displayed primarily. The proposed scheme works on visualizing real text content, which helps in exploring the hidden semantics.