Ontology guided llm agents improve enterprise data analysis accuracy
AgentOnto: Ontology-Guided LLM Agents for Enterprise Data Analytics
Artificial IntelligenceDatabases
Summary
Analyzing company data is hard because the information is scattered across many places and written in different formats. The authors organize this information into a structured map called an ontology, which helps a smart language model agent understand and perform data analysis more accurately. The system also learns from feedback and improves itself over time. Tests show it works better than other methods, and it has been successfully used in real business settings.
What this means in practice
- •For enterprise data teams: Enable data analysts to get more reliable results by linking diverse company data sources using an evolving structured knowledge map.
- •For business intelligence developers: Build analytical tools that adapt to feedback and improve accuracy by incorporating a formal ontology guiding large language model agents.$Commercial implications: This paper enables creating enterprise analytics products with higher accuracy and adaptability by combining ontologies and language models.
Authors
Dayan Pan, Bincheng Zuo, Xiaojian Wang, Daogang Zhang, Jingyuan Wang
Abstract
Enterprise data analysis requires agents to connect analytical requests with knowledge distributed across databases, business documents, and existing example tasks. Without an explicit representation of this knowledge, an agent may produce incorrect analytical results and cannot reliably incorporate feedback. We present AgentOnto, an ontology-guided LLM agent for enterprise data analytics. AgentOnto organizes heterogeneous enterprise knowledge into an ontology that directly guides analysis and incorporates confirmed feedback. Iterative Ontology Preparation further improves the ontology through self-generated analysis tasks. We evaluate AgentOnto on two public data sources. AgentOnto substantially outperforms representative methods, while module-level evaluations confirm the effectiveness of ontology construction, analysis, and updating. A real-world deployment further demonstrates its practical value in enterprise analytics. These results demonstrate that explicitly constructing, using, and updating enterprise knowledge supports reliable and adaptable data analysis.