Large-Scale Text Classification Using Gated Recurrent Units with GloVe Embeddings

The process of large-scale text classification which exists as a core component of contemporary natural language processing systems requires automatic text classification for substantial text data. The ability of traditional machine learning methods to identify contextual meaning in extensive textual material suffers because they depend on sparse data formats and manually developed features. The research introduces a deep learning system which performs large-scale text classification through Gated Recurrent Units (GRU) with pre-trained GloVe embeddings. The embedding layer converts text data into dense semantic vectors while the GRU architecture enables efficient modeling of sequential dependencies through its use of fewer parameters when compared to standard recurrent networks. The research tests three common benchmark datasets which include AG News and DBpedia and Yahoo Answers. The experimental results show that the GRU-based model provides similar classification accuracy results to LSTM but requires less processing power than LSTM which uses more advanced recurrent network designs. GRU combined with semantic embeddings offers an effective method for handling large-scale multi-class text classification challenges according to our results.