<![CDATA[The diverse Linux ecosystem complicates the selection of suitable distributions for users, particularly beginners. This study develops Distromatch, a web-based Decision Support System recommending from 30 Linux distributions using TOPSIS, enhanced with Bayesian shrinkage adjustment and user-level preference penalty as a soft constraint mechanism. Its five-stage pipeline includes criteria weight normalization from user questionnaires, TOPSIS calculation, Bayesian shrinkage, user-level preference penalty, and final ranking. Functional testing confirmed system operation, and domain expert validation for three user profiles (beginner, intermediate, advanced) showed contextually valid Top-5 recommendations, consistent with the Linux ecosystem. Bayesian shrinkage proved to be effective reduced bias for distributions with limited reviews, while soft constraints ensured level-mismatched distributions appeared in rankings with reduced scores. Calculation transparency is provided via an audit page.]]>
Copyrights © 2026
