# UniDataAgent: China Unicom's Ontology-Grounded Enterprise Q&A Agent Cuts Report Time from Days to Minutes

> ChinaUnicom's UniDataAgent pairs expert-reviewed enterprise ontologies with question-to-report execution, per its arXiv paper, claiming 95.0% strict accuracy versus 72.5% for document RAG.

- **Topic**: Research
- **Published**: 2026-09-24T07:49:53.458Z
- **Canonical URL**: https://highsignal.sh/stories/unidataagent-china-unicom-s-ontology-grounded-enterprise-q-a-agent-cuts-report-t-0ee40ca1

## Why It Matters

The paper reports report generation drop from several working days to minutes and ontology build from about a week to a few hours across 27 enterprise tables, suggesting ontological grounding, not just retrieval, may be key for structured analytical queries.

## Key Findings & Analysis

### What happened

China Unicom researchers posted UniDataAgent on arXiv, an ontology-grounded agent for enterprise question-to-report automation. Per the paper, the system splits work into an Ontology Acquisition and Validation (OAV) stage that builds versioned enterprise ontologies from metadata, business knowledge, and supporting materials via expert-authored business skills, constrained generation, question verification, and selected expert review, and a Question-to-Report Execution (QRE) stage that retrieves semantic contracts per question, coordinates skills and data tools, validates results, and produces evidence-linked reports.

### Reported performance

The paper claims ontology grounding hit 95.0% strict accuracy on real business questions versus 72.5% for document RAG, especially on structured and compositional tasks. It reports ontology construction took a few hours instead of roughly a week manually, and report generation took a few minutes instead of several working days, across 27 enterprise tables and roughly thousands of metric types. The authors say the system has been deployed and generates cost savings, and could be replicated at other enterprises.

### Why it matters

The paper's framing is that enterprise data agents need to preserve organization-specific semantics rather than just translate questions into queries. The reported gap over document RAG suggests retrieval alone may be insufficient for analytical tasks with structured or compositional requirements. The accuracy and timing figures come from the authors' own evaluation and deployment; the paper does not describe an independent benchmark.

## Primary Sources & Citations

- [UniDataAgent: An Ontology-Grounded Agent for Enterprise Question-to-Report Automation](https://arxiv.org/abs/2609.27257) — *ArXiv CS.CL (Computation and Language)* (Reporting)

---

[← Back to front page](https://highsignal.sh/) | [Daily Brief](https://highsignal.sh/brief) | [All stories](https://highsignal.sh/latest)