A Novel Approach to Multilingual Text Document Classification using Fuzzy Logic and XAI

The fast expansion of worldwide digital content and language-diverse information systems has made multilingual document classification more crucial. Deep learning techniques achieve extremely high classification accuracy, but their opaque nature restricts transparency and trust, especially in critical regulated applications. To classify multilingual documents, this study suggests an explainable fuzzy-based framework that combines explainable artificial intelligence, fuzzy logic, and semantic embeddings. Contextual embeddings are extracted to capture cross-lingual semantics after multilingual texts have been collected and prepared. Interpretable feature representation and reasoning are made possible by modeling linguistic uncertainty and ambiguity using fuzzy logic and membership functions. It incorporates explainability by using both post-hoc analysis and intrinsic fuzzy rules, providing understandable explanations for categorization outcomes. Model performance is improved while maintaining interpretability using an iterative optimization and evaluation loop. The proposed framework aims to balance accuracy, robustness, and transparency, making it suitable for real-world multilingual high-stakes applications.