Sentiment Analysis Based on Gated Recurrent Unit

Abstract

Sentiment analysis is a term which is used for classification one text is positive, negative or neutral. Sentiment analysis is used many applications like e-commercial in order to make pricing of things, customer churn analysis, brand strategy and advertisement. These goals can be executed as automatic process thanks to machine learning. Sentiment analysis is handled as supervised learning process in machine learning. In order to classification of text as a positive, negative or neutral is used user comments and votes. There are many machine learning supervised algorithms can be used for this purpose such as LSTM, CNN, GRU. In this study, a sentiment analysis was implemented by using Gated Recurrent Unit using 243 thousand line of the Hepsiburada, Turkish e-commercial platform, datasets and obtained 0.95 accuracy.