TEXT CLASSIFICATION USING DEEP LEARNING

Our daily usage of the internet generates vast volumes of text, much of which is unfiltered. Unstructured data must typically be categorised in order to increase the speed at which a particular text is interpreted. Unstructured text data can be distinguished using a branch of natural language processing called text classification. Because machine learning can generate intricate prediction functions dynamically, it is frequently employed in the categorization of textual data. Similar to how statistical models may explain the relationship between two or more random variables, textual data is frequently classified using statistical models. The challenge of performing sentiment analysis in an e-commerce setting is often difficult. Neural network methods for machine learning Recent studies on text categorization using neural network-based applications have demonstrated promising results. The model still finds it difficult to think about regional characteristics and words that depend on the information in the phrase. This research suggested using deep learning to create more exact sentences that use the classification of earlier texts. A recurrent neural network (RNN) with the architecture of long short-term memory (LSTM) is one of the deep learning techniques employed. As a result of how we use the internet on a daily basis, huge volumes of text are created regularly, and the majority of this text is unfiltered. Unstructured data must typically be categorised in order to increase the speed at which a particular text is interpreted. Unstructured text data can be distinguished using a branch of natural language processing called text classification. Because machine learning can generate intricate prediction functions dynamically, it is frequently employed in the categorization of textual data. Similar to how statistical models may explain the relationship between two or more random variables, textual data is frequently classified using statistical models. The challenge of performing sentiment analysis in an e-commerce setting is often difficult. The performance of Naive Bayes and Decision Tree machine learning approaches is limited when it comes to sentiment analysis. A comparison of recurrent neural networks (RNN) is done in this paper. Support Vector Machine (SVM) is used to categorise consumer product review data according to whether the remarks are favourable or negative. In order to attain the best results, this study prefers to apply Long Short-Term Memory (LSTM) to improve the conventional RNN. The research’s findings demonstrate that RNN, with an accuracy of 87.57.86%, outperforms state-of-the-art SVM, which has an accuracy of 59.67%.