Reducing External Dependence and Minimizing Security and Regulatory Risks
Model to Be Released on Global Open-Source Platform

On July 2, BC Card announced that it has improved the accuracy of its generative artificial intelligence (AI) search, specialized for the financial sector, by utilizing its self-developed embedding model.


Seoul Jung-gu BC Card Headquarters building. BC Card

Seoul Jung-gu BC Card Headquarters building. BC Card

View original image

An embedding model is a technology that analyzes the meaning of documents to find and deliver the most relevant information to users.


According to BC Card, existing embedding models had limitations in fully understanding domestic financial terminology and context, such as merchant and card payment-related terms. To address this, BC Card developed its model using a dataset of 1.82 million cases that had previously been released on an open-source platform.


The model is available in two versions: a lightweight version and a high-quality version.


Notably, the lightweight model ranked number one worldwide in state-of-the-art (SOTA) evaluation categories.


BC Card plans to release the model on the global open-source platform Hugging Face. Starting with integration into KT Group's AI services, the company also plans to expand its financial AI search platform business to government agencies and financial institutions.



Sungsoo Oh, Executive Vice President and Head of AI Data Division at BC Card, stated, "By opening and sharing our self-developed financial-specialized model and dataset, we aim to contribute to the broader adoption of AI transformation (AX) across the financial sector and to the establishment of a Korean-style financial AI ecosystem."


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing