<![CDATA[Managing IT support requests in a university environment remains largely manual, leading to untracked ticket backlogs, slow response times, and misrouted requests caused by incorrect manual categorization. This study presents the design and implementation of a web-based IT Helpdesk Ticketing System integrated with an automatic ticket category classification module using Multinomial Naive Bayes with TF-IDF feature extraction, developed for Universitas Kristen Indonesia (UKI) using the Waterfall SDLC on the Laravel 10 framework. The novelty of this work lies in the real-time integration of a Python-based Naive Bayes microservice directly into the Laravel ticket submission workflow. The classifier was trained on 450 domain-labeled IT support tickets across five categories and evaluated using 5-fold cross-validation (mean accuracy 88.7% +/- 0.9%) and a held-out test set (accuracy 88.9%, Macro-F1 87.6%), outperforming SVM (85.6%) and KNN (81.5%). Black-Box testing yielded 100% pass rate. ISO 9241-11 usability evaluation (n=25) produced effectiveness 95%, efficiency 92%, SUS 76.25 (Grade B)]]>
Copyrights © 2026
