ABSTRACT
This research presents a combined rule-based and machine learning approach to improve faculty workload scheduling. It addresses the limitations of traditional manual methods. The system was developed through a design and development research process. It takes into account institutional rules, past workload data, and faculty preferences. Hard constraints are handled with rule-based logic, while machine learning forecasts the best workload assignments under flexible conditions. Testing revealed a significant reduction in scheduling conflicts, a better balance in teaching hours, and high satisfaction with preferred time slots. The results indicate that this hybrid approach effectively improves fairness and efficiency in faculty workload distribution. Future studies may broaden its application and consider additional scheduling factors.
Keywords: Automated Scheduling, Constraint Satisfaction, Educational Resource Management, Faculty Workload Distribution, Hybrid Scheduling Approach, Machine Learning, Rule-Based System, Start–End Time Constraints, Timetabling Optimization, Workload Fairness
INTRODUCTION
Faculty workload distribution in academic settings is still a challenge. Traditional manual and rule-based scheduling methods often lead to inefficiencies and unfair task allocation. These methods usually do not effectively consider start and end time constraints or faculty preferences, which are important for ensuring fairness and boosting productivity [1], [2]. This paper presents a hybrid system that combines rule-based logic and machine learning to improve faculty workload distribution while following institutional policies and accommodating individual preferences [3], [9].
The system was developed using a design and development research framework. It integrates institutional policies, historical workload records, and faculty survey data [1], [4], [5]. Hard constraints were addressed through rule-based logic, while a supervised machine learning model predicted optimal workload assignments based on soft constraints, enabling more flexible and personalized scheduling [6], [9]. This combination aims to align institutional standards with the actual needs and preferences of faculty members [2], [7].
Evaluation results show that the hybrid approach significantly reduced scheduling conflicts by 82% and improved workload equity—reducing the variance in faculty teaching hours from 6.5 to 2.1 hours. Additionally, it satisfied 91% of start and end time preferences and achieved an 89% prediction accuracy in forecasting workload demands [8], [9]. These findings suggest that such a system can greatly enhance efficiency, reduce scheduling errors, and promote fairness in workload distribution [3], [10].
In conclusion, combining rule-based systems with machine learning provides a scalable and practical solution to the ongoing challenges in faculty workload management. It is recommended that institutions implement similar hybrid approaches while ensuring regular system updates, continuous feedback collection, and periodic retraining of the machine learning component. Future enhancements could also include modeling additional constraints such as research commitments, administrative responsibilities, and preferred teaching patterns [1], [6], [7].
RELATED REVIEW
Faculty workload distribution and scheduling have been widely studied due to their impact on institutional efficiency and faculty satisfaction. Johnson et al. [1] emphasize the need for data-driven approaches to redefine workload metrics, highlighting limitations in conventional models that often overlook individual preferences and dynamic constraints. Ludwig-Beymer et al. [2] further evaluated a new teaching workload model, demonstrating improved balance but also identifying persistent challenges in accommodating flexible scheduling demands.
Several studies have proposed decision support and automated scheduling systems to address these challenges. Perez Ortega et al. [3] developed an online scheduling system integrating faculty loading within a decision support framework, which showed promising results in handling complex constraints. Similarly, Campanilla et al. [8] created a web-based pre-loading system that streamlines faculty workload allocation, reducing manual errors and administrative burdens. However, many existing systems primarily rely on rule-based logic and lack adaptive mechanisms to predict or learn from workload patterns.
To enhance scheduling flexibility and fairness, machine learning and optimization algorithms have recently been explored. Austero et al. [9] applied a heuristically enhanced whale optimization algorithm to optimize workloads and room utilization, achieving significant improvements in efficiency. Okahana et al. [7] experimented with randomized trials to address workload disparities, showing that algorithmic interventions can support equitable distribution. Despite these advances, combining machine learning with rule-based frameworks remains relatively underexplored.
The reviewed literature underscores the importance of hybrid approaches that integrate institutional rules with predictive modeling to better accommodate faculty preferences and institutional policies. This study builds on prior work by proposing a hybrid system that leverages rule-based logic for hard constraints and machine learning for optimizing flexible workload assignments, thereby addressing identified gaps and improving workload fairness and efficiency [4], [5], [6], [10].
see PDF attachment for more information