Sentiment Analysis based Recommender System for Reforming Indian Education using Multi-Classifiers
Social media sites have become major source of communication over the internet pertaining to discussion on various subjects. With increase in the growth of internet users, lotsof huge data is generated through social networking sites like Twitter and Facebook. Tremendous amount of data is being generated from educational sector too. Users share their learning experiences like difficulties faced, quality of content and teaching and this resulted in analyzing their valuable sentiments towards Education. This data needs to be mined and classified properly so that knowledge is drawn regarding sentiments. The obtained knowledge from public sentiments can be analyzed before making any decisions in educational reforms. The study aims to use the search key word titled Education in India for extracting tweets related to education from twitter database by means of twitter API. Tweets are pre-processed and further analyzed and classified into three types of sentiments i.e., positive, negative and neutral based on polarity scores. Machine Learning techniques like SVM, Naïve Bayes, Decision Tree, KNN and MLP are employed to predict and classify the tweets by extracting the hidden knowledge. In this work, our foremost objective is to discover the efficient classification technique in order to reform the educational society by considering valuable sentiments and views from the twitter data related to education. Results achieved are evaluated on parameters like Accuracy, Specificity, Sensitivity, Confidence Interval and Kappa Statistics. SVM outstands in predicting students’ sentiments compared to other techniques.