Deep Learning Based Recommender System Using Sentiment Analysis to Reform Indian Education

Abstract

Deep learning is a subset of machine learning, also known as hierarchical learning. It is based on artificial neural network with various stages of representative transforms. Deep neural networks have been applied in different applications like image processing, speech recognition; market-basket analysis and students’ performance prediction to name a few. Now a day’s education is not limited to only the classroom teaching but it goes beyond that like Online Education System, Web-based Education System, Seminars, Workshops, MOOC courses. It’s a big challenge to extract sentiments from the huge data generated which is stored in the environments of Educational databases. Mining on educational databases can be done to extract the hidden sentiments of the students and their views about the education. Analyzing Students’ sentiments and their learning behavior towards the course, difficulties faced, time spent for the course duration in learning the concepts and worries or fears of students like whether they may pass or fail the Final Exam is of prior importance these days in educational institutes. These factors play a dominant role in reforming education. Tweets are gathered from twitter database and found that the obtained are in unstructured form. Preprocessing methods were applied to clean the data set and later classified tweets based on sentiments into classes namely positive, negative and neutral. In this Paper, sentiments of students are analyzed which can be further considered while making reforms in education. In this paper Educational tweets are extracted from Twitter using twitter API and preprocessed. After Preprocessing, clean data is trained and a Model is attained, on this test data is applied. Results are evaluated on few parameters like Balanced accuracy, Sensitivity and Specificity; Prevalence and Detection rate and found that deep learning technique achieves high performance.