DEVELOPMENT OF AN INTELLIGENT ADAPTIVE EXAMINATION TIMETABLING SYSTEM FOR NIGERIAN UNIVERSITIES USING A HYBRID CP-SAT AND GENETIC ALGORITHM WITH HUMAN-IN-THE-LOOP OPTIMIZATION

ABSTRACT

University examination timetabling is a complex scheduling problem involving the allocation of examinations to available periods, rooms, invigilators, and other institutional resources while satisfying numerous academic and administrative requirements. In many Nigerian universities, examination schedules are still developed through manual or semi-automated processes. Such approaches can become inefficient as student enrollment, academic programs, examination requirements, room constraints, and departmental interactions increase. This study proposes the development of an intelligent adaptive examination timetabling system that combines constraint programming with the CP-SAT solver, genetic algorithm (GA), and human-in-the-loop (HITL) optimization. The CP-SAT component is intended to enforce critical hard constraints such as examination clashes, room capacity, room availability, and time-slot restrictions, while the genetic algorithm will explore alternative timetable configurations and improve soft objectives such as examination spacing, student convenience, workload distribution, and resource utilization. A human-in-the-loop layer will allow authorized examination officers and administrators to review, modify, approve, or provide feedback on generated schedules without compromising the integrity of hard constraints. This hybrid architecture responds to the limitations of relying exclusively on either exact optimization or evolutionary search. Existing literature identifies university timetabling as a difficult combinatorial optimization problem, while recent research continues to demonstrate the value of hybrid and meta-heuristic approaches. The proposed system is therefore positioned as a practical decision-support solution for Nigerian universities, with emphasis on adaptability, transparency, constraint satisfaction, computational efficiency, and human oversight. The expected outcome is a system capable of producing high-quality, conflict-free examination timetables while supporting institutional decision-makers in handling real-world changes before and during examination periods.

Keywords: Examination Timetabling, University Scheduling, CP-SAT Optimization, Genetic Algorithm, Human-in-the-Loop AI, Constraint Programming, Artificial Intelligence, Combinatorial Optimization, Nigerian Universities, and Adaptive Scheduling.

CHAPTER ONE

INTRODUCTION

1.1 Background to the Study

The effective organization of university examinations is an important component of academic administration because the quality of an examination schedule can directly influence students, academic staff, examination officers, and the efficient use of institutional facilities. Examination timetabling involves assigning examinations to available dates and time periods while simultaneously considering rooms, student populations, invigilators, course relationships, and institutional policies. The problem becomes particularly challenging when many examinations compete for limited resources and when several programs share classrooms, laboratories, lecturers, or student populations. University timetabling is widely regarded as a difficult combinatorial optimization problem because the number of possible schedules increases rapidly as the size of the institution grows.

In a conventional university environment, examination officers may rely on spreadsheets, databases, paper records, or manually coordinated schedules. Although these methods can be useful for smaller scheduling tasks, they can become increasingly difficult to manage when a university has large student enrollment, numerous departments, multiple academic levels, limited examination venues, and several examinations taking place within a short period. A change to one examination may consequently require several other assignments to be reviewed. This creates a cascading effect that can increase the probability of clashes, room over-allocation, uneven examination distribution, and delays in publishing the final timetable.

The increasing availability of artificial intelligence (AI) and optimization technologies creates an opportunity to transform examination scheduling from a predominantly manual administrative process into an intelligent decision-support activity. Modern optimization systems can evaluate large numbers of possible schedules and identify solutions that satisfy predefined constraints while optimizing selected objectives. Recent reviews show that university timetabling has increasingly incorporated meta-heuristic, multi-objective, and hybrid optimization methods because of the complexity of the search space and the need to balance several competing objectives.

One particularly relevant technology is CP-SAT optimization. Constraint Programming provides a formal mechanism for representing rules and relationships within scheduling problems, while CP-SAT combines constraint programming techniques with satisfiability and integer optimization capabilities. In scheduling applications, CP-SAT supports concepts such as interval variables, non-overlap constraints, cumulative resource constraints, and objective optimization. These features make it suitable for representing examination periods, rooms, capacities, and other scheduling requirements.

However, an exact constraint-based method may not always be sufficient when the scheduling environment contains numerous competing preferences and institution-specific objectives. This motivates the integration of a genetic algorithm (GA). Genetic algorithms are population-based evolutionary optimization techniques that generate and improve candidate solutions through processes such as selection, crossover, and mutation. Their flexibility has made them a frequently studied approach for university timetabling. Reviews of timetabling research indicate that genetic algorithms remain among the commonly investigated population-based methods, while hybrid approaches have become increasingly important.

Recent practical research also demonstrates the continuing relevance of genetic algorithms to examination scheduling. A 2026 real-world university examination timetabling study developed a genetic algorithm system for multi-department scheduling involving shared rooms and proctoring resources and reported improvements over manually produced timetables across its evaluated institutional instances. This reinforces the potential value of evolutionary approaches when examination scheduling involves practical institutional constraints rather than purely theoretical benchmark problems.

The proposed study, therefore, combines the complementary strengths of CP-SAT and genetic algorithm optimization. CP-SAT will provide a constraint-driven foundation for ensuring that critical rules are respected. The genetic algorithm will subsequently support exploration and improvement of candidate timetables according to defined quality measures. The combination is intended to reduce the weakness of using either method in isolation. Earlier research has similarly demonstrated the value of combining constraint-based reasoning with genetic algorithms for timetable planning, particularly where candidate solutions must be checked and repaired against constraints.

A further innovation of the proposed research is the integration of human-in-the-loop AI. Fully automated optimization can produce mathematically attractive schedules that may nevertheless conflict with contextual knowledge possessed by examination officers. For example, an administrator may know that a particular venue is temporarily unavailable, that a departmental examination requires specialized facilities, or that an exceptional institutional activity requires adjustment of a particular examination period. Rather than eliminating human decision-making, the proposed system will treat human expertise as an important component of the optimization process.

The Human-in-the-Loop component will allow authorized users to inspect generated schedules, identify practical concerns, introduce approved adjustments, and provide feedback that can guide subsequent optimization. This approach supports a more transparent and accountable form of intelligent scheduling in which the algorithm assists institutional decision-makers rather than replacing them.

The research is particularly relevant to Nigerian universities, where examination administration can involve multiple faculties, departments, academic programs, student groups, examination venues, and administrative units. A locally adaptable system can be designed to reflect institutional realities, including limited room capacity, shared facilities, changing examination dates, departmental preferences, and late scheduling adjustments. Rather than assuming that every university operates under identical conditions, the proposed system will provide configurable constraints and optimization objectives.

The study consequently positions examination timetabling as an intelligent optimization problem involving three complementary layers: constraint satisfaction, evolutionary search, and human decision-making. The CP-SAT layer will focus primarily on feasibility and hard constraints; the genetic algorithm will support optimization of alternative feasible schedules; and the human-in-the-loop layer will incorporate institutional judgment and approval. This architecture is expected to provide a more adaptive approach to examination scheduling than conventional manual methods.

1.2 Statement of the Problem

University examination scheduling is often complicated by the interaction of several constraints. A timetable must prevent students registered for multiple courses from being assigned examinations at overlapping times. Examination rooms must have sufficient capacity, and individual rooms cannot normally host simultaneous examinations. Invigilators and other resources may also have limited availability. In addition, students may benefit from appropriate spacing between examinations rather than having several demanding examinations scheduled consecutively.

When these requirements are managed manually, the scheduling process can become time-consuming and susceptible to human error. A modification to one examination can require repeated checking of other courses, rooms, students, and staff assignments. The resulting timetable may therefore be feasible in one respect but inefficient or inconvenient in another.

Existing optimization approaches also have limitations. A genetic algorithm can efficiently explore a large solution space, but without strong constraint handling, it may generate infeasible candidate solutions. Conversely, a constraint-based model can enforce strict requirements effectively but may require substantial computational effort when numerous objectives and preferences must be considered. Research on university timetabling consequently highlights the importance of hybrid approaches and appropriate constraint-handling mechanisms.

Another problem is that purely automated systems may provide limited opportunities for examination administrators to incorporate contextual knowledge. The absence of an effective human feedback mechanism can make an otherwise optimized timetable difficult to implement in practice.

This study therefore addresses the need for an adaptive examination timetabling solution that can combine rigorous constraint satisfaction, intelligent search, and human judgement within one integrated framework.

1.3 Aim of the Study

The main aim of this study is to develop an intelligent adaptive examination timetabling system for Nigerian universities using a hybrid CP-SAT and genetic algorithm optimization framework with human-in-the-loop decision support.

1.4 Objectives of the Study

The specific objectives are to:

1. identify and model the major hard and soft constraints associated with university examination scheduling;
2. develop a CP-SAT-based model for generating feasible examination timetables;
3. design a Genetic Algorithm for improving the quality of feasible timetable solutions;
4. integrate CP-SAT and Genetic Algorithm techniques into a hybrid optimization framework;
5. incorporate Human-in-the-Loop functionality for timetable review, modification, feedback, and approval;
6. develop measurable criteria for evaluating timetable quality, including conflicts, resource utilization, examination spacing, and computational performance; and
7. Evaluate the effectiveness of the proposed system against conventional or baseline scheduling approaches.

1.5 Research Questions

The study will be guided by the following questions:

1. What hard and soft constraints significantly influence examination timetabling in Nigerian universities?
2. How effectively can CP-SAT generate feasible examination timetables?
3. To what extent can a genetic algorithm improve the quality of CP-SAT-generated schedules?
4. Does the hybrid CP-SAT–Genetic Algorithm approach provide better timetable quality than individual approaches?
5. How can human feedback be incorporated without violating critical scheduling constraints?
6. How effective is the proposed system in reducing examination conflicts and improving resource utilization?
7. What level of computational efficiency can be achieved by the proposed adaptive system?

1.6 Significance of the Study

The study is expected to benefit university examination officers by reducing the time and effort required to construct and revise examination schedules. The system can provide automated conflict detection and optimization while retaining human control over final decisions.

Students may benefit from schedules that reduce examination clashes and improve the distribution of examinations across available periods. Academic and administrative staff may also benefit from improved allocation of rooms and other examination resources.

For university management, the proposed system can serve as a decision-support tool for improving examination administration and resource utilization. For researchers, the study contributes to the growing body of work on hybrid optimization, educational scheduling, and human-in-the-loop intelligent systems.

The study may also provide a foundation for future integration with university information systems, student registration databases, academic management platforms, and real-time institutional scheduling services.

1.7 Scope of the Study

The study focuses on the design and development of an intelligent system for university examination timetabling. It will consider examinations, students, courses, examination periods, rooms, room capacities, and relevant scheduling constraints. The optimization framework will combine CP-SAT and genetic algorithm techniques, while a human-in-the-loop interface will support authorized review and adjustment.

The study does not seek to automate every aspect of university academic administration. Its primary focus is examination timetable generation, optimization, evaluation, and controlled human intervention.

1.8 Operational Definition of Key Terms

Examination Timetabling: The process of assigning university examinations to appropriate dates, time periods, and venues while satisfying specified constraints.

CP-SAT: A constraint-based optimization technology capable of representing scheduling rules and searching for feasible or optimized solutions.

Genetic Algorithm: An evolutionary optimization technique that improves candidate solutions through population-based search mechanisms.

Hybrid Optimization: The integration of two or more optimization techniques to exploit their complementary strengths.

Human-in-the-Loop: A design approach in which human users participate in reviewing, modifying, validating, or guiding an intelligent system's decisions.

Hard Constraints: Rules that must be satisfied for a timetable to be considered feasible, such as preventing student examination clashes.

Soft Constraints: Preferences that improve timetable quality but may be relaxed when necessary.

Adaptive Scheduling: A scheduling process capable of responding to changing data, constraints, preferences, or institutional circumstances.

1.9 Conceptual Direction of the Proposed System

The proposed architecture follows a sequential but interactive optimization process. First, institutional examination data and constraints are collected. Second, CP-SAT establishes feasible candidate schedules by enforcing critical constraints. Third, the genetic algorithm explores and improves candidate solutions according to an objective or fitness function. Fourth, the optimized timetable is presented to an authorized human administrator. The administrator can approve the schedule or identify acceptable modifications. Where changes are introduced, the system re-evaluates the timetable and re-optimizes affected components while preserving critical constraints.

This architecture is consistent with the broader movement toward hybrid optimization in educational timetabling, where researchers increasingly combine complementary approaches rather than relying on a single algorithm.

1.10 Summary of the Chapter

This chapter established the background and rationale for developing an intelligent adaptive examination timetabling system for Nigerian universities. Examination scheduling was identified as a complex combinatorial optimization problem involving multiple resources, constraints, and competing objectives. The chapter proposed a hybrid framework combining CP-SAT, genetic algorithm optimization, and human-in-the-loop decision support. The proposed approach seeks to produce feasible, efficient, adaptable, and practically acceptable examination schedules while maintaining human oversight. The following chapters will build on this foundation by examining relevant literature, methodology, system design, implementation, and evaluation.