Twitter Data Analysis using Natural Language Processing and Recurrent Neural Network

The essence of this work presented is the conception and construction of a unique model. This model harnesses the power of Natural Language Processing (NLP) and Recurrent Neural Networks (RNN) to decipher the sentiments expressed across various social media platforms. The ultimate goal is to illuminate the emotional landscape of the digital world. In the work presented in this paper the twitter social media posts/messages are retrieved having a particular keyword. Many posts are made using identical message formats which users are unaware about. For effective retrieving of the post identifying the relevant information is required. Many social media post have data which are irrelevant to analyze. For effective analysis removing this irrelevant data is necessary. Once the posts are processed, tokens and keywords are identified. Keywords are important aspects which are used further to identify the sentiment. The processed data is then stored in database and split into two category for training and testing purpose. The polarity and subjectivity of the data is then calculated for analyzing the sentiments using NLP and RNN. The proposed NLP-RNN approach is evaluated using accuracy, loss, and evaluation time required.