Purpose: This study aims to design, implement, and evaluate a Fuzzy Mamdani-based book recommendation system for the library of Institut Informatika dan Bisnis (IIB) Darmajaya, addressing the challenge of navigating large library collections by providing personalized, context-aware recommendations that accommodate the subjectivity inherent in user preferences. Methodology: A structured software engineering lifecycle was followed, comprising problem identification, requirement analysis, system design, implementation, testing, and evaluation. The Fuzzy Mamdani Inference System used three input variables, borrowing frequency, book rating, and difficulty level, and one output variable, with triangular membership functions, 27 If-Then rules, and centroid defuzzification, implemented using Personal Home Page (PHP), Laravel, MariaDB, and verified in Matrix Laboratory (MATLAB) R2016a.
Results: Functional testing confirmed successful operation of all system modules, and MATLAB verification demonstrated consistency between manual and automated computations, with a sample input producing a recommendation score of 4.3 via centroid defuzzification.
Conclusions: The system successfully delivered personalized recommendations, outperforming manual cataloguing approaches by providing dynamic, preference-sensitive output.
Limitations: Evaluation was limited to functional testing in a controlled environment, without representative user acceptance testing, and the 27-rule base has not been assessed for scalability to larger collections.
Contributions: This research contributes a domain-specific Mamdani inference design and a validated web-based implementation architecture, providing a replicable model for similar university libraries in Indonesia seeking to modernize information retrieval services through intelligent systems.