Generating Questions and Feedback from Ontologies
Generating answerable questions and feedback automatically from ontologies can support teaching and learning activities and therewith alleviate teachers' workload. One may represent the same knowledge using different modelling styles resulting in distinct ontologies, which existing studies do not consider. So their algorithms are ontology-dependent, static, and suitable for restricted cases. Moreover, the existing studies only deal with predefined question types, and there is not yet a question-generation method to provide good-quality questions, limiting generalisability. Feedback generation also lacks generalisability, and the resulting feedback is still relatively basic. To address these problems, we take an NLP-driven approach combining both parsing and generation. Then, we combine this approach with a semantics-based method with template variables using ontology entities to develop an architecture implemented as a proof of concept. In our architecture, we separate the main components into different modules to support diverse cases. Our analysis and use cases show that our ontology-based approach overcomes the generalisability limitations, and our human-based evaluation proved that the NLP-driven approach generates good questions regarding the syntax and semantics with multiple ontologies. Thanks to our architecture and our algorithm that extracts the specific axioms from abstract axiom patterns, our approach can also generate feedback using the ontology's relevant contents. The current work is to improve feedback generation.
Publications
An Architecture for Generating Questions, Answers, and Feedback from Ontologies (pdf)
Toky Raboanary, Maria KeetIn: Garoufallou, E., Vlachidis, A. (eds) Metadata and Semantic Research (pp. 135-147). MTSR 2022. Communications in Computer and Information Science, vol 1789. Springer, Cham. https://doi.org/10.1007/978-3-031-39141-5_12.
Towards the generalisation of the generation of answerable questions from ontologies for education.
Toky Raboanary, Steve Wang, Maria KeetInternational Journal of Metadata, Semantics and Ontologies 16(1), 86–103 (2022)
Generating Answerable Questions from Ontologies for Educational Exercises (pdf)
Toky Raboanary, Steve Wang, Maria KeetIn Metadata and Semantic Research: 15th International Conference, MTSR 2021, Virtual Event, November 29–December 3, 2021, Revised Selected Papers (pp. 28-40). Cham: Springer International Publishing.
The list of all my publications is here.