Abstract
A knowledge graph and graph neural network framework is proposed herein to replace flat, entity-by-entity infrastructure risk reports with a relationship-aware view of how risk emerges across configuration, software, hardware, topology, and other operational dimensions. The framework preserves relationship types during representation learning, so that risk arising through an entity's relationships is distinguished from risk recorded on the entity itself. The learned representations are composed into a prioritized remediation ordering, and a language model translates the resulting findings into clear, operator-readable narratives bounded to the supplied facts. The result is a repeatable, scalable, and explainable workflow that surfaces at-risk entity groupings that current rule-based tools consistently miss.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Dhure, Akash, "A KNOWLEDGE GRAPH AND MACHINE LEARNING FRAMEWORK FOR RELATIONSHIP-AWARE RISK INTELLIGENCE IN ENTERPRISE NETWORKS", Technical Disclosure Commons, (September 09, 2026)
https://www.tdcommons.org/dpubs_series/11670