Benzinga Launches Korean Translation Engine and Financial Dataset for AI Training

Benzinga, a leading provider of real-time financial news and market data, announced the launch of its Korean translation engine and a large-scale Korean financial dataset built for AI training, localization, and data licensing applications.

Real-Time Korean Translation for Financial Markets

Benzinga’s translation engine delivers U.S. market news in Korean in real time, designed for brokerages, banks, fintech platforms, and trading applications where accuracy at the headline, ticker, and event level is critical.

The system is optimized specifically for financial content, helping preserve context, market sentiment, company attribution, and industry terminology that are often lost in general-purpose translation systems.

A Large-Scale Korean Financial Dataset for AI

In addition to the live translation feed, Benzinga is making its historical news archive available as a Korean-language dataset derived from its proprietary English and Korean financial news corpus spanning more than a decade.

The dataset contains more than 400 million words of Korean financial text aligned to real market activity, creating one of the largest commercially available Korean financial datasets designed for AI model training and financial language applications.

General-purpose translation datasets often struggle with financial terminology, entity attribution, causality, and sentiment. Benzinga’s dataset is designed to preserve those market signals across both languages.

Read More on Fintech : Global Fintech Interview with Baran Ozkan, co-founder & CEO of Flagright

Structured for APIs, Workflows, and Model Training

Both the real-time feed and historical dataset are delivered in machine-readable formats with structured metadata, timestamps, and entity tagging for direct integration into APIs, data pipelines, analytics systems, and AI training workflows.

“Benzinga is embracing AI use cases for financial news, including model training, and we’re excited to see what builders create with structured market data that reflects how markets actually move,” said Michael Saad, AI Licensing Lead at Benzinga. “Most models still underperform in non-English financial domains because the underlying training data simply doesn’t exist at scale. This changes that for Korean.”

Translation Performance Optimized for Finance

Internal evaluations showed the translation engine outperforming leading general-purpose systems on financial-domain BLEU and ROUGE-L benchmarks, with stronger performance in entity alignment, ticker accuracy, and headline interpretation.

“Our approach applies a proven machine learning principle: combining specialized models produces stronger results than relying on a single system alone,” said Aiden Harris, AI Engineer and Data Scientist at Benzinga. “The engine runs multiple fine-tuned models in parallel and synthesizes their outputs through a separate aggregation layer, delivering state-of-the-art performance on financial translation tasks where precision directly impacts decision-making.”

Catch more Fintech Insights : Real-Time Payments and the Redefinition Of Global Liquidity

[To share your insights with us, please write to psen@itechseries.com ]

The post Benzinga Launches Korean Translation Engine and Financial Dataset for AI Training appeared first on GlobalFinTechSeries.