Telecom Big Data: Social Media Sentiment Analysis

Abstract

Despite Big Data being one of the trending buzzwords, it is possibly one of the least comprehended terms in business. Big Data revolutionized the way business is organized and managed. Nevertheless, massive data is not as clear as it may appear, especially that originated from social media. Big Data is not only concerned with the amount of data that has changed, but also with the velocity and the structural complexity of the data. Thus, the demand has emerged to look for other techniques and procedures that are structured to deal with this vast amount of data as well as drawing insights out of them. One of the most critical and challenging tasks in social media related data is processing the natural language and defining the implicit and explicit meaning behind unstructured text data. This task refers to understanding and analyzing the text sentiment. Sentiment Analysis (SA) is the process of recognizing the intention or the emotional condition of a client articulation. This paper presents a Business Intelligence (BI) framework using social media data to analyze and define the sentiment meaning of data that comes from digital channel feeds. The implemented Business Intelligence framework consists of four main phases: first, ingesting data from multiple sources: structured and unstructured data. Second, transforming, aggregating, and processing the data through an in-memory processing layer. Third, building the Natural Language Processing (NLP) model to define the meaning behind the customer message or complaint. Finally, the visualization phase, which presents the final output and insights in a single dashboard.