New Feature Added to XMAT® Machine Translation Platform: Leveraging RAG Technology to Reference Linguistic Assets

XMAT Update Details

Product details page: https://ldxlab.io/xmat

New Feature: Implemented a function to reference linguistic asset data when performing document translation using generative AI. *Linguistic asset data: Glossaries and translation memory used in translation.

In XMAT's document translation, in addition to translation and grammatical correction by generative AI, users can execute translations using their own custom prompts. While this prompt function allows for detailed specifications of terminology and style, it can be time-consuming to list all terms and expressions to be unified. RAG (Retrieval-Augmented Generation) is the technology that solves these challenges.

XMAT now utilizes RAG technology to allow registered linguistic assets to be used as reference information for AI processing. Just like selecting a translation engine or language, users can simply select the necessary linguistic assets to reflect their content in the output. Combining this new feature with the existing document translation function enables more accurate and consistent translations.

What you can do with Document Translation × RAG Technology

1. Unifying expressions using past translation data By referencing expressions from previously translated documents, you can prioritize the same phrasing and terminology as before. This prevents inconsistencies and contributes to stable quality.

2. Accurate translation reflecting glossaries Automatically references product names, technical terms, and company-specific expressions from glossaries to accurately reflect specified translations. This is expected to significantly reduce correction work and review man-hours.

XMAT will continue to evolve by actively incorporating the latest technologies and reflecting user requests to contribute to business efficiency.