Comparative Analysis of Short Text Classification Models

The role of text classification in various applications like spam and subject classification, detection, and sentiment analysis cannot be overlooked. This research explores the performance of four deep learning (DL) algorithms on short-text categorization tasks: CNN, Dense Neural Networks, Long Short-Term Memory (LSTM), and Gated Recurrent Unit(GRU). Customer complaints about a range of financial services and products, including credit reports, student loans, money transfers, and more, are included in the dataset. Classifying upcoming complaints according to their content is the aim. We assess each model's accuracy and display the findings in tabular and graphical form. The GRU-based model was shown to be useful for processing short text sequences, since it outperformed the other models that were considered in terms of classification accuracy. The results highlight the successfulness of the GRU model with increasing the clarity of text data classification and extracting context. This research further advances our knowledge of the many models that are best suited for short text recognition.