DeCREDIT: Decomposed Prompting for Credit Recognition, Evaluation and Intelligent Transfer

About the Project

The DeCREDIT is a web-based system developed for the Faculty of Computing & Multimedia at Universiti Poly-Tech Malaysia to address a credit transfer process that has historically depended on manual paperwork, fragmented records, and inconsistent syllabus evaluation. The study aimed to digitalize the end-to-end process, centralize academic records, and support course equivalency review through staged syllabus comparison using decomposed prompting. The development method combined questionnaires, interviews, and document analysis with an Agile process covering requirements, design, development, testing, deployment, and review. The implemented system integrates student onboarding, automatic draft CTA creation, transcript extraction, articulation-aware matching and eligibility screening, source syllabus handling, HOP review, RP evaluation, and AAS export within one platform. The stack uses Next.js, PostgreSQL via Prisma, Better Auth, SeaweedFS-compatible storage, Stirling PDF for text extraction, and Ollama-backed text and vision pipelines for transcript processing and staged comparison. Shared syllabus cases and shared review groups allow representative evidence and RP decisions to be reused across related CTA courses and articulation mappings. Evaluation results show that all 72 respondents in requirement analysis supported a fully digital process, all 43 non-functional and 45 functional test cases passed, and user acceptance testing with 55 students produced an average score of 3.89 out of 4.00 with zero negative responses. These results indicate that the system provides a traceable institutional model for credit transfer while preserving human academic judgment.