Enhancing Text Classification Using Advanced Natural Language Processing Techniques

Abstract—This study evaluates the performance of var- ious neural network models and a pre-trained trans- former model in the task of quote classification. The models analyzed include Fully Connected Neural Networks (FCNN), Long Short-Term Memory (LSTM), Gated Re- current Units (GRU), Bi-directional LSTM (BiLSTM), and DistilBERT. The goal is to identify the most effective model for the given dataset based on key performance metrics such as accuracy, precision, and recall. DistilBERT, a lightweight transformer-based model, is also assessed for its efficiency and accuracy compared to traditional neural network approaches. Index Terms—Natural Language Processing (NLP), Data Preprocessing, Machine Learning, Quote Classifi- cation, Multi-label Classification, Neural Networks (NN), Transformer Models, DistilBERT.