Epaminondas Kapetanios has spent many years investigating ways to improve the tasks of interpreting, explaining, and understanding computational artifacts (for example, algorithms, data, knowledge) with a variety of user types and personas. Some notable results of this explorative journey have been the design and implementation of a metadata-driven visual query language, a human-computer interactive automaton and parser for predicting user query intent, and a language and model for adaptive ontologies.
His research culminated in prototypes applicable to a variety of disciplines and real-world projects, including the retrieval of scientific and statistical databases for ozone hole research over the Arctic, natural language-based querying and processing, as well as web-based decision trees as an e-consultation system in healthcare. Epaminondas has also applied web and text mining techniques for competitive business intelligence within the context of an initiative funded by Innovate UK.
Epaminondas is currently affiliated with the School of Physics, Engineering and Computer Science, at the University of Hertfordshire. His work focuses on responsible and trustworthy AI. Specifically, he is investigating human-oriented explainable and interpretable artificial intelligence (AI) and machine learning, such as natural language (conversational, dialogue) based systems and knowledge discovery from source code mining.
Epaminondas has been a reviewer for Computing Reviews since 2014.