BanglaBERT-GCN Ensemble: Advancing Health Misinformation Classification Using Graph Based Method

To prohibit false information from spreading, it is crucial to identify and remove health-related disinformation from online content.The research report suggests a novel approach that combines BanglaBERT-integrated Graph Convolution Networks (BiGCN) with the model's power to enhance the categorization of healthrelated misleading material.The performance of the BiGCN model is compared with that of other text classification techniques currently in use in the study.Extensive testing and analysis findings demonstrate that the BiGCN model outperforms state-of-the-art techniques in identifying and categorizing health misinformation in Bangla text.A combined accuracy of 94.60% is achieved for improved categorization when the structural links obtained by GCN and the semantic understanding from BanglaBERT are combined.The outcomes show how well semantic and structural data are extracted by the BanglaBERT-GCN model, suggesting that it might prove to be a useful instrument in the fight against false health information in Bangla.