Identifying Hidden Sentiment in Text Using Deep Neural Network

Deep learning has emerged as de-facto standard in computer vision. Due to its wide spread success and popularity with image data almost every problem from other domain is now trying to finding its solution with deep learning and the case is no different for text analytics and language modeling in NLP. In sentiment analysis, identifying users' hidden sentiment in text review is a challenging task. Though there are many rule based and statistical machine learning based approaches for sentiment analysis. Most of these machine learning approaches are probabilistic based model and treats the problem as document classification problem based on the word count and their polarity often missing sentiment. With the introduction of the distributed word vector representation embedding models like word2vec, GloVe and FastText, capturing the meaning of the neighboring words for a give target word in the sentence, NLP expectation with deep learning in sentiment analysis has increased a lot. In this paper we used deep neural network based model to perform sentiment analysis on the IMDB movie review dataset. We compare deep learning based approach for sentiment analysis with other traditional machine learning approaches to evaluate the performance of the model. With experiment result we found that the deep learning based NLP model outperformed the traditional machine learning approaches in several ways.