Computational Exploration of Theme-based Blog Data using Topic Modelling, NERC and Sentiment Classifier Combine

Abstract

Our preliminary research results on a unique combination of Topic Modelling, Named Entity Recognition, and Sentiment Classification
for social analysis of blog data are shown here. More than five hundred blog entries on the topic of "discrimination, abuse, and crime
against women" have been compiled. Here, we used topic discovery to using a combination of keyword analysis and a Named Entity
Recognition method based on the 7-entity model, we were able to zero in on the most important topics and people covered in the blog
postings. We then used SentiWordNet to classify all of the blog data as positive or negative depending on the prevailing tone. The
findings produced are fascinating and provide strong evidence for the efficacy of our method for computational analysis of social media
data. This paper's main contribution is the suggestion of a new Text Analytics combination and the subsequent demonstration of its
usefulness for computational investigation of the data gleaned from social media for sociological research.