Abstract
EExplainable artificial intelligence (XAI) is concerned with the interpretability of AI systems to ensure that AI systems can provide relevant explanations that can make humans understand the basis and rationale for the results they produce. The explainability/interpretability of AI systems is crucial to make their decisions comprehensible, interpretable, transparent, retraceable, and reproducible, providing a basis for increased trust and reliability of their results. So far, the existing methods of XAI that are designed to enable the interpretability of intelligent systems are deficient in terms of providing:
● Accurate explanations;
● Convincing basis for justifications;
● Consistency of explanations in different but similar scenarios; and
● human-centric explanations
This research will investigate how the infusion of data semantics and semantic web approaches could help alleviate the challenges of existing methods in the context of explainable decision support systems. It will also investigate the applicability of the concept of Responsible AI in the context of explainable decision support systems (ExDSS) in healthcare, business, and social media.
The research is expected to lead to the design and development of novel semantics-based algorithms and methods for XAI and afford a deeper understanding of the application of explainable decision support systems. The project will yield tangible deliverables such as academic publications, software artefacts, research-based training, and degree awards for postgraduate students.
Vacancies for funded full-time masters and PhD positions exist in the SEDESS Project.
For more details contact me at: wande.daramola@up.ac.za