Tibakanya Joseph
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Tibakanya Joseph has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Ggaliwango Marvin
Department of Computer Science, Makerere University, Kampala, Uganda
6 shared publications - Agaba Ezra Joab
Department of Networks, Makerere University, Kampala, Uganda
3 shared publications - Joyce Nakatumba-Nabende
Department of Computer Science, Makerere University, Kampala, Uganda
3 shared publications - Ampeire Henry Kariisa
Makerere University
2 shared publications - Nakayiza Hellen
Makerere University
2 shared publications - Ahumuza Cedric
Makerere University, Uganda
1 shared publication
Latest Publications
- Open Access
- 24 July 2026
- Published by Elsevier in SSRN Electronic Journal
CoMAD: A Transferable Method for Building Cognition-Aware Assessment Mathematics Datasets for Educational AI Research
Recent advances in Artificial Intelligence (AI), particularly large language models, have accelerated research into automatic item generation, thereby fueling demand for curriculum-aligned, cognitively aware, contextualised datasets. However, there are limited cognitively annotated, contextually grounded, authentic assessment datasets that exhibit linguistic complexity and curriculum-aligned constructs. Moreover, dataset-construction methodologies struggle to integrate established educational taxonomies, linguistic complexity indicators, and FAIR-oriented data management into a single, transparent, and reproducible framework. We present the Cognition-Aware Mathematics Assessment Dataset construction (CoMAD), a transferable, FAIR-oriented framework for creating and curating higher-cognition, curriculum-aligned datasets augmented with computational data engineering. The method is validated using HiCogMath, a mathematics dataset in which cognitive annotations are based on Bloom’s taxonomy and Webb’s Depth of Knowledge, operationalised through Hess’s Cognitive Matrix (CRM). CoMAD is a modular dataset-creation method that is scalable to other subjects, curricula, and contexts, creating datasets suitable for cognitive modelling, data mining, assessment analytics, automated item generation, Knowledge graph construction, and LLM alignment.• Transferable framework for construction of cognition-aware assessment datasets with curriculum alignment and linguistic profiling.• Presents transparent quality assurance for enhancing reliability, replicability and reusability of datasets.• Provides a modular framework that is transferable to diverse curricula, domains and contexts via annotation protocols.
- 24 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessCoMAD: A Transferable Method for Building Cognition-Aware Assessment Mathematics Datasets for Educational AI Research
Recent advances in Artificial Intelligence (AI), particularly large language models, have accelerated research into automatic item generation, thereby fueling demand for curriculum-aligned, cognitively aware, contextualised datasets. However, there are limited cognitively annotated, contextually grounded, authentic assessment datasets that exhibit linguistic complexity and curriculum-aligned constructs. Moreover, dataset-construction methodologies struggle to integrate established educational taxonomies, linguistic complexity indicators, and FAIR-oriented data management into a single, transparent, and reproducible framework. We present the Cognition-Aware Mathematics Assessment Dataset construction (CoMAD), a transferable, FAIR-oriented framework for creating and curating higher-cognition, curriculum-aligned datasets augmented with computational data engineering. The method is validated using HiCogMath, a mathematics dataset in which cognitive annotations are based on Bloom’s taxonomy and Webb’s Depth of Knowledge, operationalised through Hess’s Cognitive Matrix (CRM). CoMAD is a modular dataset-creation method that is scalable to other subjects, curricula, and contexts, creating datasets suitable for cognitive modelling, data mining, assessment analytics, automated item generation, Knowledge graph construction, and LLM alignment.• Transferable framework for construction of cognition-aware assessment datasets with curriculum alignment and linguistic profiling.• Presents transparent quality assurance for enhancing reliability, replicability and reusability of datasets.• Provides a modular framework that is transferable to diverse curricula, domains and contexts via annotation protocols.
- Open Access
- 18 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 48779-48791
Beyond Hype: Will Ugandan Educators Continue Using Large Language Models for Learning Content Creation?
Progress in generative artificial intelligence (AI) and large language models (LLMs) has enabled several educational technology breakthroughs that automate the sometimes tedious and labour-intensive processes of generating and evaluating instructional materials. Despite the hype, there is insufficient information on how task fit, perceived ethical influence, perceived utility, and reported ease of use influence teachers’ continuance intention to utilise LLMs for teaching and learning content creation. We sought to extend the Technology Acceptance Model (TAM) by integrating the new constructs of Task-Technology Fit (TTF), Social Influence (SI), and Perceived Ethical Influence (PEI) with its long-established constructs of Perceived Usefulness (PU) and Perceived Ease of Use (PEU) to predict Continuance Intention (CI). A quantitative questionnaire survey was administered to Ugandan educators, yielding 231 responses. The data collection and analysis phase took two steps. In phase one, descriptive statistics were computed using IBM SPSS version 23, while the measurement model’s reliability, convergent, and discriminant validity, model fitting, and hypotheses were tested using ADANCO. The study’s findings underscore the significant influence of TTF and SI on both PU and PEU, which in turn strongly influence Ugandan educators’ continued intentions to use LLM technologies in their instructional endeavours. Nonetheless, the ethical dimension, which appears prominent in theoretical discourse, had no substantial effect. In phase two, we explored the unexpected null effect of ethical influence in a follow-up survey on awareness of LLM ethical concerns. Results from 105 participants showed a significant impact of ethical issues on educators’ trust in LLMS. The study contributes to the ongoing discourse on the responsible adoption of AI in education and underscores the need for targeted ethical literacy in AI-powered instructional practice.
- 18 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 48779-48791
Open AccessBeyond Hype: Will Ugandan Educators Continue Using Large Language Models for Learning Content Creation?
Progress in generative artificial intelligence (AI) and large language models (LLMs) has enabled several educational technology breakthroughs that automate the sometimes tedious and labour-intensive processes of generating and evaluating instructional materials. Despite the hype, there is insufficient information on how task fit, perceived ethical influence, perceived utility, and reported ease of use influence teachers’ continuance intention to utilise LLMs for teaching and learning content creation. We sought to extend the Technology Acceptance Model (TAM) by integrating the new constructs of Task-Technology Fit (TTF), Social Influence (SI), and Perceived Ethical Influence (PEI) with its long-established constructs of Perceived Usefulness (PU) and Perceived Ease of Use (PEU) to predict Continuance Intention (CI). A quantitative questionnaire survey was administered to Ugandan educators, yielding 231 responses. The data collection and analysis phase took two steps. In phase one, descriptive statistics were computed using IBM SPSS version 23, while the measurement model’s reliability, convergent, and discriminant validity, model fitting, and hypotheses were tested using ADANCO. The study’s findings underscore the significant influence of TTF and SI on both PU and PEU, which in turn strongly influence Ugandan educators’ continued intentions to use LLM technologies in their instructional endeavours. Nonetheless, the ethical dimension, which appears prominent in theoretical discourse, had no substantial effect. In phase two, we explored the unexpected null effect of ethical influence in a follow-up survey on awareness of LLM ethical concerns. Results from 105 participants showed a significant impact of ethical issues on educators’ trust in LLMS. The study contributes to the ongoing discourse on the responsible adoption of AI in education and underscores the need for targeted ethical literacy in AI-powered instructional practice.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 116359-116375
A Semantic-Aware Multimodal Data Warehouse for Crowdsourcing and Contextualizing AI-Generated Educational Content in Fragile Contexts
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 116359-116375
Open AccessA Semantic-Aware Multimodal Data Warehouse for Crowdsourcing and Contextualizing AI-Generated Educational Content in Fragile Contexts
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 101616-101640
From Templates to Agentic Reasoners: A Systematic Review of Automated Item Generation for Mathematics
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 101616-101640
Open AccessFrom Templates to Agentic Reasoners: A Systematic Review of Automated Item Generation for Mathematics
- 13 July 2025
- Published by Springer Nature
- p. 425-442
Retrieval-Augmented LLMs for Culturally Sensitive Learning Content in Refugee Education
- 13 July 2025
- Published by Springer Nature
- p. 425-442
Retrieval-Augmented LLMs for Culturally Sensitive Learning Content in Refugee Education
- 23 May 2025
- Published by Springer Nature
- p. 213-242
Explainable Real-Time Sign Language to Text Translation
- 23 May 2025
- Published by Springer Nature
- p. 213-242
Explainable Real-Time Sign Language to Text Translation
- Open Access
- 8 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 243-255
Explainable Machine Learning and Graph Neural Network Approaches for Predicting Employee Attrition
- 8 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 243-255
Open AccessExplainable Machine Learning and Graph Neural Network Approaches for Predicting Employee Attrition
- Open Access
- 8 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 213-225
Interpretable Machine Learning Techniques for Students Grade Prediction
Predicting student grades within an academic cycle is a remarkable opportunity to improve academic results. Robust prediction methods such as use of machine learning allow educational leaders to allocate enough resources and personalized instructions where necessary, fostering improved outcomes in Academia. We have presented various machine-learning models that have helped predict students' grades as an early intervention to determine how well the models achieve the required objectives using a combination of input features and past data or grades. We obtained a student dataset with various demographics, economic, educational factors, and grades from different subjects with their corresponding averages where we came up with 15 models including decision trees, random forest, linear regression, k-nearest neighbor, AdaBoost, Gradient Boosting, Support Vector Machine XGBoost, Lasso, Ridge, Elasticnet and among Deep Neural Networks. Our findings show that Linear Regression based model is the best with a MAPE of 8.14 and R-squared of 0.2536, and the Graph neural networks performed worst with a MAPE of 47.2 and R-squared of -83.7. We achieve computational outcomes indicative of the model's predictive performance. Our results show promising accuracy and generalization capabilities and that these models can predict very well, meaning we can rely on them to figure out how students might perform in academia.
- 8 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 213-225
Open AccessInterpretable Machine Learning Techniques for Students Grade Prediction
Predicting student grades within an academic cycle is a remarkable opportunity to improve academic results. Robust prediction methods such as use of machine learning allow educational leaders to allocate enough resources and personalized instructions where necessary, fostering improved outcomes in Academia. We have presented various machine-learning models that have helped predict students' grades as an early intervention to determine how well the models achieve the required objectives using a combination of input features and past data or grades. We obtained a student dataset with various demographics, economic, educational factors, and grades from different subjects with their corresponding averages where we came up with 15 models including decision trees, random forest, linear regression, k-nearest neighbor, AdaBoost, Gradient Boosting, Support Vector Machine XGBoost, Lasso, Ridge, Elasticnet and among Deep Neural Networks. Our findings show that Linear Regression based model is the best with a MAPE of 8.14 and R-squared of 0.2536, and the Graph neural networks performed worst with a MAPE of 47.2 and R-squared of -83.7. We achieve computational outcomes indicative of the model's predictive performance. Our results show promising accuracy and generalization capabilities and that these models can predict very well, meaning we can rely on them to figure out how students might perform in academia.