Software architecture for social media data analytics

Abstract

Nowadays, social media has grown very rapidly and has a growing number of users making it an attractive data source for analysis. A software that is able to collect data from social media, pre-process and analyze the data to generate knowledge and information as desired is required. Usually, the software is applicable only for a specific social media. The paper proposes an architecture for Social Media Data Analytics such that different softwares for analyzing different social media can be built based on it. The proposed architecture is adapted from Rahman's architecture that originally supports only for analyzing data from Facebook Different form Rahman's architecture, the proposed architecture contains of 4 blocks of new units, i.e. (1) Data Collection and Temporary Storage Unit, (2) Data and Text Pre-processing Unit, (3) Network Analysis and Data Mining Unit, and (4) Knowledge Representation Unit. The software architecture has meet three main aspects of architectural quality attributes conceptual integrity, correctness and completeness and buildability. Based on the architecture, a software has been built. The software developed by analyzing its functionalty for analyzing data from Facebook and Twitter. Thus the software is expanded such that it has functionality for analyzing data from Instagram. In order for that, we define 4 new classes (of 12 classes) extended from the original classes defined in the original software. It shows that the software built based on the proposed architecture can be extended form different type of social media with minimal effort. Using the factory method pattern, the software supports structural configurability and structural flexibility such that the extension can be done using minimal effort.