Aspect Based Sentiment Analysis of Student Housing Reviews

Abstract

According to a 2016 report by the Indian Ministry of Human Resource Development, there were 39,658 student hostels across India. In recent years, owing to the growing number of students residing in such hostels, there has been an interest in helping students know more about these hostels by providing them with information and reviews from residing students. We aim to categorize these based on various aspects and give greater insights about them using applications of aspect based sentiment analysis. We have used a neural network based approach to pre-process the texts and propose two models, one for aspect extraction and classification and the other for sentiment polarity analysis. Further, we have presented an extensive evaluation of our models and have achieved an accuracy of more than 75% on both the models.