ARCHIVES
VOL. 11, ISSUE 1 (2026)
From interpretability to knowledge: A meta-analytical conceptual framework for explainable data mining in structured and unstructured big data
Authors
T Karthikeyan, Dr. Sashank Swami
Abstract
The rapid growth of structured, unstructured, and hybrid big data has led
to widespread adoption of complex machine learning and deep learning models,
which often operate as black-box systems. Although explainable artificial
intelligence (XAI) techniques aim to improve interpretability, existing
approaches remain fragmented across data modalities and frequently fail to
translate explanations into actionable knowledge. This study presents a
meta-analytical and conceptual synthesis of explainability across structured,
unstructured, and hybrid data mining systems. The analysis identifies key
patterns and trade-offs, including the shift from intrinsic interpretability to
approximation-based methods and the increasing disconnect between model
performance and knowledge extraction. The findings highlight a critical gap
where explanation does not necessarily imply understanding, causality, or
decision relevance. To address this limitation, the paper proposes the
Explainability-Driven Knowledge Discovery (EDKD) framework, which integrates
data, models, explanations, and knowledge into a unified structure. The
framework extends traditional data mining pipelines by explicitly incorporating
knowledge as a distinct analytical layer, enabling more effective and
human-centered data-driven decision-making.
Download
Pages:108-116
How to cite this article:
T Karthikeyan, Dr. Sashank Swami "From interpretability to knowledge: A meta-analytical conceptual framework for explainable data mining in structured and unstructured big data". National Journal of Multidisciplinary Research and Development, Vol 11, Issue 1, 2026, Pages 108-116
Download Author Certificate
Please enter the email address corresponding to this article submission to download your certificate.