Prediction Depression Analysis Framework For Twitter Data Using Genetic Algorithm Based Feature Extraction And Naive Bayes Classifier

Abstract

Sentiment Analysis and Opinion Mining is a hotspot for testing and the accelerated growth of platforms of social networks. Social media sites like Twitter, Face book and many more play an extremely important role in today's world. Twitter is a micro blogging site offering a vast quantity of data that can be used for numerous opinion analysis purposes, including forecasts, ratings, campaigns, ads, films and so on. Sentiment Analysis is a method for evaluating the positivity or negative of knowledge gathered from massive quantities and classifying it through multiple groups named emotions. This research examines a detailed view of the methodologies used in the classification of sentiments over Twitter data. This research provides a detailed overview of methodologies used in classifying sentiments over Twitter data, presenting a novel approach to depression analysis using advanced techniques such as stop word removal and lemmatization for preprocessing the Twitter data, elite term score, word2vec, and a modified genetic algorithm for feature extraction. The extracted features are then sent to a classifier to classify the tweets as depression or non-depression, and the performance of the MGA-NB model results in high accuracy, sensitivity, and specificity.