In order to meet the need for scalability in academic advising in higher education settings while ensuring compliance, fairness, and privacy, we propose a governance-ready multi-agent reinforcement learning framework that aggregates the advantages of four highly specialized agents, namely, the Plan Validator, Risk Detector, Resource Recommender, and Conversational Advisor, which function collaboratively with a policy learner and under the supervision of a human. In a quasi-experimental mixed-methods study involving n = 345 participants from different departments, the system yielded a substantial improvement over the baseline in terms of efficiency and quality, where the turnaround time in academic advising was reduced by ≈60% or from 48.6 to 19.3 hours, while compliance was boosted by +11.3% and risk detection recall was enhanced by +0.12 or from 0.67 to 0.79. The stability of the learning process was satisfactory, with a normalized cumulative reward of +0.35 and around 1,200 episodes to converge. For the governance control, the effectiveness of the proofs was satisfactory, as the fairness gap reduced from 7.8% to 3.2%, while the privacy incidents remained zero even with continuous audits. In terms of evaluation, a mixed methods evaluation design was used, which included the performance records of the RL, ANOVA subgroup, and trust judgments to establish validity. From the findings, the MARL solution is technically viable, ethical, and feasible as an advisory mechanism in HEIs. Suggestions are made for future directions to investigate the solution's scalability and dynamic reward shaping for various policies.

The proposed research aims to develop a hybrid learning analytics model that incorporates clickstream and sentiment mining techniques to accurately predict student retention in Learning Management Systems (LMS) in Manila-based HEIs. This proposed research aims to improve the limitations of single prediction modeling. The proposed research utilized a quantitative explanatory research design and was applied to undergraduate students taking general education courses. The proposed research utilized stratified sampling to select a total of 345 students. Clickstream and text-based interaction data were analyzed to obtain sequence and sentiment-based metrics to train the hybrid prediction model using Logistic Regression and XGBoost. The engagement tool was also able to reach Cronbach's alpha = 0.91, which is a good measure of internal consistency. This shows that the click-only model had an AUC of 0.78, but the hybrid model was enhanced to 0.84, with the top predictors being attempts at quizzes per week and negative sentiment ratio. This shows the importance of multimodal analytics in the improvement of early warning systems and proactive interventions. The implication of this can be seen in the administration of the institution and the academic support, which is a basis for evidence-based retention and fairness-based predictive models in Philippine Higher Education.

This paper presents an innovative ensemble method for developing a student policy chatbot that leverages advanced natural language processing techniques, specifically an extractive question-answering approach combined with a large language model (LLM). The study aims to enhance the chatbot’s ability to accurately understand and respond to student inquiries regarding university policies and procedures. Utilizing a dataset derived from the University Handbook and Code of Decorum for Genderbased Sexual Harassment, the chatbot architecture integrates the RoBERTa model for extractive answering and GPT-2 for natural language generation. Evaluation metrics, including automatic assessments and human evaluations, reveal the chatbot’s strengths in relevance and coverage, while also identifying areas for improvement in precision and coherence. The findings underscore the potential of AI-driven chatbots in higher education to enhance student engagement and support, while recommendations for future enhancements are necessary to ensure the chatbot’s effectiveness in addressing the diverse needs of students.

Tinkering activity, defined as the iterative modification of code and debugging, is a key indicator of engagement and problem-solving approaches for novice programmers. In this work, we propose an interpretable machine learning model that relies on feature engineering and regression analysis to identify and measure the extent of tinkering activity in C/C++ programming environments. Based on a feature-dense dataset obtained from compiler log files, we employ binary logistic regression for categorical prediction and multiple linear regression for continuous prediction of tinkering activity. Our model is centered on interpretability, filling the gap of existing black-box models in educational analytics. The model’s key predictors, including attempt rate, corrective moves, and syntactic changes, are found to have high statistical significance. Through the integration of behavioral analytics and regression-based machine learning, this study makes a contribution to the development of intelligent tutoring systems that are capable of providing real-time, personalized feedback. The results of this study highlight the role of interpretable models in improving human-machine interaction in programming education.

This study extended a previous paper that focuses on the acceptability of selected Bachelor of Science in Computer Science (BSCS) and Information Technology (BSIT) students on the use of Intelligent Tutoring System (ITS) as an educational technology tool for C/C++ Programming. A one-shot case study research design was carried out in 5 programming classes taught by the author. A Slovin's formula computation from the population was 35.54. A stratified sampling method was employed with the 4 intervals between students to mitigate bias. The study involved 39 participants, out of which 74.36% were male and 25.64% were female computer science and IT students. Utilizing the Technology Acceptance Model (TAM) as an evaluation tool online enabled importing the dataset into IBM SPSS for finding the correlations and factor loading calculations. Cronbach alpha was conducted by the author with a value of 0.947, which signifies the measure of internal consistency. The seven (7) factors of TAM were analyzed to reveal coefficient values for comparisons and derive their relative implications. Research indicates that every factor significantly influences the acceptance of ITS among BSCS and BSIT students. Interestingly, PerUse→Att has the highest coefficient value (0.883) next in the rank was SocNor→Att by a factor of 0.822 signifying their impact on ITS (Att), leaving SocNor→PerEas ranking last amongst relations with a 0.630 coefficient value. Finally, the results implied CS and IT students are open to the notion of incorporating intelligent teaching tools into their laboratory sessions to supplement their programming activity and increase their efficiency when building console applications.