ExecGraph Research
Applied research on industrial relationship intelligence. The question is not whether a model finds another name. It is whether network evidence helps identify a relevant, credible route to the right person.
EXG-RN-001 · Graph learning / Link prediction / Energy
Recorded relationships grew faster than the population. GraphSAGE improved more on ROC AUC, but semantic similarity still led. A dated comparison with the evaluation limits left visible.
GraphSAGE test ROC AUC
Offline research note · Not peer reviewed. Metric gains are not customer outcome measurements.
These are research directions, not additional completed studies. New results appear only after evidence and disclosure review.
Research and data partnerships
ExecGraph welcomes inquiries from AI labs and research teams studying graph representation learning, relationship prediction, temporal robustness, or graph-assisted retrieval in industrial domains.
Proposals can cover controlled evaluation or a potential licensed training study. Describe the research question, intended data use, required fields, evaluation design, and security arrangements. Availability depends on source rights, third-party terms, privacy review, and a separate written agreement. No underlying dataset or model-training license is granted by these pages or by ordinary platform access.
Inquiries go to support@execgraphenergy.com. No raw records, proprietary feature definitions, or model artifacts are published here.