Grant: Discovery Launch Supplement

Data analytics plays a pivotal role across various academia and industry domains. The importance of data ethics is highlighted by the General Data Protection Regulation (GDPR) and its similar counterparts. Notably, GDPR introduces the "right to explanation," granting individuals the right to request an explanation for data analysis outcomes that significantly affect them. However, achieving explainability in many data analytics models, particularly those relying on neural networks, remains a significant challenge. In the long term, this project aims to enhance the explainability of data analytics through three-way methodologies, establishing a realm of explainable three-way data analytics. Three-way data analytics (3WDA) covers a wide range of topics founded on three-way thinking and computing, such as using three parts, levels, facets, and views. The use of threes is a simple yet powerful idea observed in many practices and theories, underscoring its cognitive advantages. This implies the potential contributions of three-way methodologies to explainable data analytics, which has not received adequate attention. This proposal works around the following short-term objectives: (O1) Developing fundamental theories for grounding explainable 3WDA     (O1.1) Motivations and cognitive advantages of three-way thinking     (O1.2) General frameworks and formulations of explainable 3WDA (O2) Constructing specific models for building explainable 3WDA     (O2.1) Models in the lifecycle of data analytics     (O2.2) Modes and multi-faceted views of explainable 3WDA (O3) Implementing practical applications for realizing explainable 3WDA     (O3.1) Explainable intelligent systems     (O3.2) Explainable decision support systems The objective (O1) justifies WHY three-way methodologies can contribute to explainability and forms the basis of explainable 3WDA. (O2) studies HOW explainability can be achieved using three-way methodologies in diverse facets of data analytics. (O3) showcases WHAT can be realized in real-world scenarios by explainable 3WDA. These three objectives together lay the groundwork for constructing a paradigm of explainable 3WDA, characterized by robust foundations, meaningful models, and useful applications. The promising results from my past research activities will serve as starting points, ensuring the feasibility of achieving the objectives of this proposal. With an in-depth exploration of explainable 3WDA, this proposal will make the following key contributions. (a) In academia, it will highlight the roles of three-way methodologies and approaches in improving the explainability of data analytics. (b) In industry, it will generate explainable and transparent data products and tools. (c) In HQP training, it will equip HQPs with comprehensive training for successful careers. (d) Finally, it will foster an EDI environment by promoting interdisciplinary research and the development of ethical data products.