Enhanced News Category Classification Using Hybrid Model

The pervasiveness of digital reporting has created a demand for accurate and efficient information systems. This paper presents an enhanced method for news classification by taking advantage of a set of pre-trained transformer models. A modern approach based on a hybrid model is employed to improve the classification performance further, combining predictions from different models. The proposed architecture has a 2percent improvement compared to the latest work. These results show that this proposed method outperforms individual models, yielding significant improvements in classification accuracy. This hybrid model also reduces class-wise bias and balances error distribution, enhancing generalizability, though it introduces higher computational complexity, highlighting the efficiency-performance trade-off. This study highlights the potential of using multiple pretrained models for complex text classification tasks and provides a robust solution for the automated classification of media content.