Social Media Data Processing Infrastructure by Using Apache Spark Big Data Platform: Twitter Data Analysis

Social media provide continuous data streams that contain information with different level of sensitivity, validity and accuracy. Therefore, this type of information has to be properly filtered, extracted and processed to avoid noisy and inaccurate results. The main goal of this work is to propose architecture and workflow able to process Twitter social network data in near real-time. The primary design of the introduced modern architecture covers all processing aspects from data ingestion and storing to data processing and analysing. This paper presents Apache Spark and Hadoop implementation. The secondary objective is to analyse tweets with the defined topic --- floods. The word frequency method (Word Clouds) is shown as a major tool to analyse the content of the input dataset. The experimental architecture confirmed the usefulness of many well-known functions of Spark and Hadoop in the social data domain. The platforms which were used provided effective tools for optimal data ingesting, storing as well as processing and analysing. Based on the analytical part, it was observed that the word frequency method (n-grams) can effectively reveal the tweets content. According to the results of this study, the tweets proved their high informative potential regarding data quality and quantity.