DEEP LEARNING APPROACHES FOR CLASSIFYING DATA: A REVIEW

Data mining can be considered as the first approach for classification of sentiments. Data mining can be considered as the first approach for classification of sentiments. Later, Machine learning and its techniques were used to analyse sentiments but, machine language-based learning systems find it complex to understand the language of humans. Therefore, we move towards deep learning models to analyse sentiments. The subgroup of machine learning is Deep -Learning; it involves networks, namely RNN (Recurrent Neural Networks), Recursive Neural Networks, Convolutional Neural Network (CNN) and Deep Belief Networks. Neural networks are very useful in the generation of text, the depiction of vector, word assessment, classifying sentences and representation. Sentiment analysis can be determined as a process for identifying the emotions with the help of a series of words which are used in online sites. It can be utilized to analyse the point of view and attitudes, depending on the words. Sentiment analysis is mostly used in monitoring social media to gain information about public opinion on certain trending topics. Sentiment analysis is performed by taking some sentiment examples, the features are extracted from sentiments and then the parameters are trained in our model and in the final stage, the model is tested. In this paper, an empirical survey of the three models of deep learning, namely RecurrentNN, RecursiveNN and ConvolutionalNN are discussed.