Sentiment analysis and prediction using neural networks

Abstract

Sentiment Analysis or opinion mining have taken many leaps and turns from its starting in early 2000s till now. Advancement in technology, mobile & internet services and ease of access to these services have resulted in more and more engagement of people on the social media platforms for expressing their views and collaborate with people who share similar thoughts. This has led to generation of a large amount of data on the internet and subsequently the need of analysing this data. The sentiment analysis helps different organisations to know how people look to their products and services and what changes are required to improve them. The paper performs sentiment analysis i.e. classification of tweets into positive, negative and neutral on views of a particular product using an inbuilt python library called TextBlob for three platforms i.e. twitter, Facebook and news websites and further it talks about how Artificial Neural Networks (ANN) offer a platform to perform sentiment analysis in a much easier and less time-consuming manner. In this paper Feed-Forward Back propagation neural networks are used to split the data into train and test data and a min-max approach was applied to the data to scale the data and analyse the prediction accuracy of a sentiment using ANN. Precision, recall and accuracy have been calculated to provide a quantitative approach to the results and measure the performance of ANN. We found that such type of neural network is very efficient in predicting the result with a high accuracy.