Enhance and IBK Industrial Bank of Korea Launch Ontology-Based AI Data Analysis Pilot
TECHWORLD ·
✦ AI Summary
Enhance and IBK Industrial Bank of Korea announced on the 1st that they have launched an ontology-based AI data analysis pilot.
The pilot focuses on assetizing in-house work knowledge and tacit knowledge into knowledge assets that can be used by AI agents and verifying whether data can be queried and analyzed using only natural-language questions.
Enhance will deploy AgentOS on IBK Industrial Bank of Korea's internal network and establish a Semantic Layer, then conduct a quantitative evaluation using a golden set through November.
Enhance and IBK Industrial Bank of Korea announced on the 1st that they have launched an ontology-based AI data analysis pilot. The pilot focuses on assetizing in-house work knowledge and tacit knowledge into knowledge assets that can be used by AI agents.
Enhance said the purpose of the pilot is to ease the burden on frontline employees who request materials from the IT department and to reduce the burden of directly handling existing queries. Through this, the company will examine an environment in which employees rely less on separate data requests or query work.
The pilot will verify whether needed data can be queried and analyzed using only natural-language questions. It will check whether employees can query and analyze the data they need with natural-language questions alone.
To that end, Enhance will deploy its in-house AI agent platform, "AgentOS (AOS)," on IBK Industrial Bank of Korea's internal network. It will also establish a "Semantic Layer" that connects business meaning and relationships to information-system data used for the bank's reporting and analysis work.
Enhance said that, to improve the performance of data-utilization technologies such as Text-to-SQL and RAG, it is important to reflect a company's unique business context. Accordingly, it proposed an ontology as the applied technology, with the ontology taking on the role of systematizing business concepts and relationships. Enhance plans to organize frontline work knowledge and tacit knowledge into a knowledge dictionary, then connect that knowledge dictionary with the data schema so that the AI query interpretation standard reflects the bank's business terminology and decision criteria.
The scope of verification will range from simple lookups to root-cause analysis and insight generation. The verification targets include complex analysis, and it will also check whether reference data can be presented for each answer and whether the reasoning path for each answer can be shown. In addition, explainability, which describes the basis and derivation process of answers, is also included in the verification targets.
The pilot will run through November. Performance evaluation will be conducted using a quantitative evaluation method based on a pre-agreed golden set. The golden set has the character of benchmark data for evaluation.
IBK Industrial Bank of Korea expects this pilot to be the starting point for realizing Ready Data for AI. The purpose of Ready Data for AI is to assetize in-house work knowledge and tacit knowledge and establish a framework for using AI agents.
Lee Seung-hyun, CEO of Enhance, said that in the financial sector, it is important to build a data semantic layer so that domain knowledge such as business terminology and decision criteria can be used by AI. He added that this pilot has proved that even on a bank's internal network, analysis results and their grounds can be presented together, and said the company will expand data utilization by frontline employees based on this.
Source: TECHWORLD · Lee Gwang-jae
Original: https://www.epnc.co.kr/news/articleView.html?idxno=407602
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.