Sentiment analysis using deep learning on Persian texts
Behnam Roshanfekr, Shahram Khadivi, Mohammad Mahdi Rahmati · 2017
Given the growth rate in the volume of text data and information, text classification has become more practical and handy. Sentiment analysis is one of the text classification applications which can be used in some cases to evaluate products, make market decisions or measure consumer confidence. Most of the methods proposed for this task have concentrated on the English language whereas there have been a few attempts for other languages such as Persian. There are some challenges in Persian Language. For example, it has a wide variety of suffixes. Recently, deep learning approaches have been successfully applied in a variety of NLP applications. Our goal is to evaluate deep learning methods in the Persian language. It can be shown that some of the challenging issues will be addressed when using deep learning methods. We also introduce a dataset of reviews about electronic products in Persian language and evaluate the models on it.