An AI-Based Decision Support System for Enhancing Academic and Administrative Efficiency at Prince Sattam bin Abdulaziz University

Abstract

Higher education institutions increasingly depend on data-driven tools to manage growing student populations, diversifying degree pathways, and rising demand for responsive administrative service. This paper presents an integrated Artificial Intelligence-based Decision Support System (AI-DSS) for Prince Sattam bin Abdulaziz University that couples two complementary components beneath a shared decision-support layer. A Natural Language Processing (NLP) module, built on transformer encoders (BERT for English and AraBERT for Arabic), performs joint intent classification and slot/entity extraction over bilingual administrative queries, enabling automated triage of recurring requests. A hybrid recommendation module fuses content-based filtering over course descriptions with collaborative filtering over historical enrollment and grade records, subject to a hard prerequisite- and degree-plan-constraint layer, to produce actionable course-planning suggestions. Both modules feed a unified analytics and Key Performance Indicator engine for academic administrators. The primary novelty is architectural: rather than another single-purpose chatbot or recommender, the paper contributes an evaluated, bilingual, constraint-aware integration of both capabilities within one decision-support architecture, together with a cold-start fusion-weight schedule for sparse-history students. The methodology is validated against real, publicly available benchmarks CLINC150, Banking77, and HWU64 for the NLP module and the Open University Learning Analytics Dataset (OULAD) for the recommendation module before an institution-specific deployment layer is described, and it is positioned against a structured, literature-wide comparison of related work rather than an internal model-variant benchmark. We further define a reproducible evaluation protocol spanning intent classification, entity extraction, ranking quality, ablations, and usability study. The framework is offered as a reference architecture for AI-based decision support in Arabic-English bilingual universities.