An Explainable Fuzzy-Logic Framework for Personalized Password Recommendation and Context-Aware Strength Assessment

Authentication remains an important problem due to the "Usability-Security Paradox". In an attempt to create a secure system with a challenging procedure, users' interest decreases, and vulnerabilities increase. Machine learning (ML) techniques have improved in accuracy in the digital age. This paper presents an explainable type-1 Mamdani Fuzzy Inference System (MFS), known as FuzzyGuard, developed for context-aware strength assessment and personalization recommendation. The study enables mapping a vague idea of password strength to an overlapping fuzzy set, using features such as length and complexity, enabling excellent, linguistically interpretable feedback. It incorporates a partly stochastic recommendation algorithm based on "Cognitive Anchors," using slices of data from the user's profile to generate high-entropy passwords that are cognitively aligned with the user's memory. Which was developed using the Django-JavaScript architecture and tested through computational testing and a user study (N = 51). The results showed that FuzzyGuard was 100% efficient and had an 86.3% satisfaction rate, with 94.1% of users certifying that they remember the fuzzy string better than the standard random string. The comparative study showed that FuzzyGuard is more interpretable and personalized than current deep learning and ensemble models. This study demonstrates that the fuzzy-logic framework is effective in bridging the gap between security and usability, offering clear, human-centred direction and setting a new standard for proactive, personalized authentication security.