Advancements and Challenges in Gated Recurrent Units (GRU) for Text Classification: A Systematic Literature Review

This systematic literature review (SLR) examines recent advancements and persisting challenges in using Gated Recurrent Units (GRU) for text classification. GRUs, a variant of recurrent neural networks, have become essential for processing sequential data due to their ability to handle temporal dependencies effectively. This review synthesizes findings from studies published between 2019 and 2024, highlighting the evolution of GRU applications across diverse contexts, such as sentiment analysis, natural language inference, and topic detection. We discuss how hybrid models integrating GRUs with technologies like CNNs, Transformers, and attention mechanisms have improved performance and adaptability. However, these advancements come with increased computational demands and complexities, such as overfitting, particularly in complex datasets. The SLR provides a comprehensive overview of internal modifications, hybrid approaches that enhance GRU models, and strategies to mitigate their limitations. Future directions are suggested for optimizing GRU efficiency, developing innovative architectures, and improving model interpretability to ensure ethical and effective deployment in real-time analytics and edge computing.