A Survey of Deep Learning for Sentiment Analysis

Deep learning has detonated in the publicresponsiveness, primarily as predictive and analytical productspervade our world, in the form of innumerable humancenteredsmart-world systems, including targetedadvertisements, natural language assistants and interpreters,and mock-up self-driving vehicle systems. In contrast,researchers across disciplines have been including into theirresearch to solve various natural language processing issues. Inthis paper we seek to provide a thorough exploration of Deeplearning and its applications like sentimental analysis andnatural language processing (NLP). Deep learning has an edgeover the traditional machine learning algorithms, like supportvector machine (SVM) and Naïve Bayes, for sentiment analysisbecause of its potential to overcome the challenges faced bysentiment analysis and handle the diversities involved, withoutthe expensive demand for manual feature engineering. Deeplearning models promise one thing - given sufficient amount ofdata and sufficient amount of training time, they can performthe task of sentiment classification on any text class withminimal restrictions and no task-specific or data-specificmanual feature engineering. We hope this survey provides avaluable reference for new Deep learning practitioners, as wellas those seeking to innovate in the application of deep learning