SocialPulse: a Tool for Extracting Interesting Insights from Social Media

EasyChair Preprint 9276

Abstract

Digital media provides a huge amount of data. This data has rich content and gives us an opportunity to find interesting insights from it. The data consists of texts and other related attributes. The textual data can be used to extract valuable information from it. A shortcoming of the existing approaches is that the structure of the documents is neglected as the primary attribute remains the frequency. This, as a result, loses some of the valuable characteristics of the documents. In this work, we build a framework called Social Pulse that uses keywords to extract live tweets from Twitter and extracts multifold meaningful information from it. It is a complete framework that consists of a data pipeline that fetches and processes tweets, incorporates graph mining, has micro-services to serve data from backend to front-end, and provides a dashboard to visualize the analysis in the form of charts and graphs. At the core of the Social Pulse, we use gSpan, which is a famous and one of the most efficient Frequent Subgraph Mining (FSM) algorithms. We implement a parallel execution of gSpan in which we leverage the multicore processing technique to run gSpan in parallel to improve the execution time. The parallel implementation is imperative because the social media data grows large in size so the sequential run would take a lot of time to process. Our approach uses cooccurrence graphs to represent textual data in graphical form. The tweets’ texts from Twitter are preprocessed and converted into co-occurrence graphs. The gSpan then extracts the frequent subgraphs from the graph database to infer the most common phrases occurring in the texts. Along with the tweet's text, there are multiple other attributes associated with the tweet. We use those attributes to infer multiple meaningful insights from the data.

Keyphrases: Frequent Patterns, Interesting Insights, Social Pulse, text analysis

@booklet{EasyChair:9276,
  author    = {Earass Ahmad and Kifayat Ullah Khan},
  title     = {SocialPulse: a Tool for Extracting Interesting Insights from Social Media},
  howpublished = {EasyChair Preprint 9276},
  year      = {EasyChair, 2022}}