An Knowledge-Based Decision Support System for Loan Assessment and Credit Evaluation in Cooperative Societies

Abstract

The financial viability of cooperative societies relies on loan assessment and credit evaluation, yet traditional evaluation techniques employed are heavily dependent on manual judgment, subjective interpretation, and inconsistent application of lending policies. This study presents the design and implementation of a Knowledge-Based Decision Support System (KBDSS) developed using the C Language Integrated Production System (CLIPS) to automate and standardize cooperative loan assessment processes. In order to create structured production rules that evaluate applicants according to factors like income level, debt ratio, collateral adequacy, consistency of savings, and membership tenure, the system captures expert knowledge from cooperative loan officers and institutional policies. Using forward-chaining inference, the system provides transparent and justifiable recommendations which are categorized as Approved, Referred for Review, or Rejected. 50 random past loan cases from Federal Cooperative College Staff Cooperative Investment and Credit Society Limited (FCCIB Staff CICS LTD) were validated by comparing system-generated results with expert opinions. The results showed a 94% decision accuracy rate with notable improvements in processing speed and consistency. The system’s explanation facility enhances transparency by displaying the logic behind each decision, which is consistent with Explainable Artificial Intelligence (XAI) principles and regulatory demands for fairness in credit automation. This study shows that knowledge-based systems can effectively support human decision-making in cooperative finance, thereby, ensuring fair, auditable, and efficient loan evaluation system. The results support the role of symbolic AI in creating intelligent financial tools that are adaptable enough to work in low-resource environments. Possible future improvements include the integration of database, hybridization with data-driven models, and deployment as a cloud-based or mobile application for wider accessibility.