Скачать или смотреть Toky Hajatiana Raboanary, HPI Research School at UCT 2021

Toky Hajatiana Raboanary
Hasso Plattner Institute Research School at UCT
PhD Student, Computer Science
Supervisor: Maria Keet

Generating Answerable Questions from Ontologies

Abstract:
This work aims at generalising the generation of answerable questions from ontologies. Generating well-structured questions from ontologies with respect to fluency and adequacy is important to be able to interact with the human being. It can be useful for education, validation of authored ontologies and so on. We have done a preliminary study which is about generating educational questions from 3 different ontologies. We considered 10 types of questions with their 72 variants. Our experiment shows interesting results. Our NLG (Natural Language Generation)-enhanced templates approach reached over 80% very good syntactic; and 73.7% as good or very good semantic quality. If it works for 3 different ontologies, can we claim that our approach is a generalisation? Does our approach have the same performance for other ontologies? Since there are many ways to formalise the same knowledge; probably, our previous algorithm cannot cope with this diversity. For instance, one can adopt different representation styles (28 different traits are partitioned over 10 dimensions); Foundational Ontologies such as DOCLE, Yamato, BFO, GFO, UFO and so forth; and Ontology Design Patterns and so forth. Each decision choice design affects the axiom prerequisites for each type of questions and the NLG process for generating questions. Hence, to enlarge the scope of the generalisability of our approach, we will take into consideration this plurality of knowledge representation. This work will open up to other research studies such as reusing ontologies, translating ontologies from representation styles to other ones and developing practical educational exercises systems.

Bio:
I am Toky Hajatiana Raboanary. I am a PhD student at the University of Cape Town in Computer Science. My supervisor is Associate Professor Maria Keet. My work is about generalising the generation of answerable questions from ontologies. I aim to contribute to the field of Ontology Engineering. Before continuing my studies in Cape Town, I studied and worked at Institut Supérieur Polytechnique de Madagascar (ISPM).


About the Hasso-Plattner Research School at UCT:
The Hasso Plattner Institute Research School at UCT, in Information and Communications Technology for Development (ICT4D), serves as a gateway for educating sub-Saharan African information technologists, and in the grand scheme, to become a leading centre of excellence in designing and understanding computational solutions (both algorithmic and user centric) and systems to address the UN Sustainable Development Goals. Learn more here: http://www.sit.uct.ac.za/sit/research...

The school is financially supported by the Hasso Plattner Institute for Digital Engineering.