Contextual estimation using K-Means clustering and SVD to classify microblogging data

Abstract

In order to express their thoughts and feedback with their followers, individuals, celebrities, organizations, and business communities are increasingly using microblogging services. The fundamental and primary concept behind the success of any business to earn the popularity of people through the social media stations is the enormous amount of messages that are being posted every single second. With its massive user base and 280-character limit, $X$ (Twitter) is one of the well-known microblogging services. The annotations, abbreviations, links, photos, hashtags, and emotions of the user are included in the tweets. While still a difficult undertaking, sentiment analysis of these user characteristics can reveal a wealth of useful data about user behavior and client segmentation. In this paper, it is presented that the design of a sentiment-based analysis system that uses machine learning models, particularly the K-Means clustering unsupervised learning model, to extract some context, features, and hidden information from short messages on any social media platform, such as Twitter or Instagram, and then identify a specific category for each type of short messages and place them in an identical cluster.