Strategi Penanganan Ketidakpastian dalam Sistem Pakar dan Decision Support System: Integrasi Pendekatan Kualitatif dan Probabilistik

Abstract

Uncertainty constitutes a fundamental challenge in Expert Systems and Decision Support Systems (DSS) that affects decision-making accuracy. This research aims to analyze various uncertainty handling approaches through qualitative descriptive, comparative, and analytical literature study. The study focuses on four main approaches: Qualitative Confidence Levels (QCL) representing beliefs in linguistic forms, Bayesian probabilistic reasoning updating probabilities based on new evidence, Data Marks as quality indicators for information in knowledge bases, and multi-criteria uncertainty weighting in multi-criteria decision-making. The analysis reveals that each approach has specific strengths and limitations. QCL provides linguistic flexibility but is vulnerable to subjectivity, Bayesian offers strong mathematical structures but requires accurate prior data, Data Marks enhances transparency while still requiring subjective interpretation, whereas MCDA weight uncertainty is robust in multi-criteria conditions despite computational complexity. The key finding emphasizes that uncertainty cannot be effectively managed with a single method. Integration of qualitative, probabilistic, and structural approaches proves more adaptive to real-world complexity. This research contributes theoretically by providing an integrative framework for developing more robust, transparent, and adaptive intelligent systems in handling various dimensions of uncertainty from data, expert knowledge, model structure, to decision preference weights.