SEGRA: AI Agent for Gremlin Querying on Enterprise Knowledge Graphs
A team of researchers has introduced SEGRA, a new agent aimed at enhancing text-to-Gremlin question answering for enterprise IT support knowledge graphs. These graphs map out relationships among cases, users, devices, symptoms, categories, root causes, and past solutions. However, effectively querying them in Gremlin requires a good grasp of graph schemas and traversal rules. SEGRA uses techniques like intent routing, schema-based query creation, and a skill library that builds on proven query patterns. In tests against a benchmark for enterprise IT support, SEGRA outperformed traditional methods by achieving a judge score 7.0 times higher. Moreover, its skill library reduces both LLM calls by 20% and costs by 18%. This study was shared on arXiv with the ID 2607.22713.
Key facts
- SEGRA stands for Structured Experience-Guided Graph Reasoning Agent
- It is designed for Gremlin-based question answering on enterprise IT support knowledge graphs
- Enterprise IT support knowledge graphs contain relationships among cases, users, devices, symptoms, taxonomic categories, root causes, and historical resolutions
- SEGRA integrates intent routing, schema- and taxonomy-grounded query generation, multi-shot decomposition, execution-aware verification, and a curriculum-bootstrapped skill library
- On an enterprise IT support benchmark, SEGRA achieves a 7.0× higher mean judge score than backbone-only chain-of-thought prompting
- The skill library reduces LLM calls by 20% and dollar cost by 18% relative to SEGRA without skills
- The paper is available on arXiv with ID 2607.22713
- The announcement type is new
Entities
Institutions
- arXiv