Abstract
Data exploration methods that rely on manual intervention may be limited in their ability to autonomously discover insights from large, unanalyzed enterprise datasets. To address this, systems and methods are described that can employ a self-improving, multi-agent architecture. This system can feature an exploration layer for generating hypotheses, an evaluation layer for validation, a temporal knowledge graph (TKG) to serve as a persistent memory of insights, and a meta-learning layer. The meta-learning layer can observe analytical outcomes stored in the TKG to refine future exploration policies. This closed-loop process can enable the system to autonomously generate, validate, and learn from its analyses, which may allow data exploration to function as an evolving, self-optimizing process for proactively identifying insights with reduced human steering.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Kommuri, Sai Charan Tej and Sukoon, Madhup, "Self-Improving Autonomous Data Exploration System with Meta-Learning and Temporal Knowledge Graph", Technical Disclosure Commons, (December 15, 2025)
https://www.tdcommons.org/dpubs_series/9040