Abstract
Proposed herein are techniques to enhance an agentic Artificial Intelligence (AI) platform by integrating a dynamic, on-premises network topology knowledge graph with a Retrieval Augmented Generation (RAG) system through graph-based techniques. Unlike traditional text-based RAG, the approach segments the network topology into semantically meaningful subgraphs called topology chunks, which are embedded into a vector space for efficient similarity search and retrieval based on user queries. These retrieved topology chunks then augment the prompts that can be sent to an agentic AI platform's large language model (LLM), enabling precise, context-aware reasoning about complex network environments.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Barton, Robert and S Desai, Vishal, "INTEGRATION OF LIVE NETWORK KNOWLEDGE GRAPHS WITH RAG FOR ENHANCED AI-DRIVEN NETWORK OPERATIONS", Technical Disclosure Commons, (March 09, 2026)
https://www.tdcommons.org/dpubs_series/9478