Twitter Sentiment Analysis with LSTM Neural Networks

This project delves into sentiment analysis on Twitter using Long Short-Term Memory (LSTM) Neural Networks in conjunction with Global Vectors for Word Representation (GloVe). The study explores the properties of tweets, preprocessing steps, and applying GloVe embedding’s to map words to vectors. The classifier’s design and training parameters are detailed, and the results are compared with baselines, revealing the LSTM’s superiority in handling sequential language data. Furthermore, trials explore how changing the quantity of fully connected layers and LSTM time steps affects accuracy. The findings suggest the importance of sequential processing in natural language processing (NLP) tasks. However, accuracy may benefit from flexible LSTM time steps aligned with tweet length. This project highlights the potential of LSTM neural networks for sentiment analysis on social media platforms like Twitter.