Sentiment Classification Based on Deep Learning

Abstract

The increase in the usage rate of the Internet and social media around the world has also led to an increase in the text-based contents. Content shared in a social media environment contains important information about the person in many ways. In revealing this information; After extracting the attributes on the text based content, it is possible by machine learning algorithms. Nowadays, in addition to the machine algorithms, the extraction of the feature from the text can be performed automatically with the help of deep learning methods. In this study, binary sentiment classification (positive-negative) operation was performed on messages shared in the Twitter environment, which is one of the most popular social media applications. The classification process is carried out with various machine learning algorithms and Convolutional Neural Networks (CNN) which is a deep learning algorithm. As a result of the classification process, the deep learning algorithm classifies the data set with higher performance than the five machine learning algorithms. We also examined the effect of the batch size on classification process.