Argos Multilingual, a leader in the global content solutions market, has announced the release of a significantly upgraded version of its AI-driven Translation Memory (TM) cleanup service, which enables automatic analysis and cleanup of content stored in TMs at scale. These enhancements add new features and automations to this solution, first launched in 2022, and introduce intelligent ongoing TM monitoring and improved terminology cleanup using new AI capabilities.
The AI-driven Translation Memory cleanup service developed by Argos Multilingual has already proven to be of great value for clients managing multilingual assets. This is because clean data help improve the leverage TMs can provide but are also critical for the development and fine-tuning of custom MT engines and large language models.
The initial release last year addressed the key challenges of outdated entries, terminology errors, untranslated or mistranslated segments, as well as inconsistencies in Translation Memories.
The upgraded service introduces several major enhancements. These include the intelligent TM health score (TM Vanguard) that has been introduced to continuously monitor TMs and identify issues that require attention over time. This proactive approach enables clients to stay on top of their TM maintenance and ensures that TMs are consistently optimized for linguistic quality.
The terminology cleanup module has been significantly improved through the application of advanced AI technologies made available recently. The enhanced AI algorithms reduce false positives and intelligently identify real terminology issues within TMs. This results in more accurate and reliable terminology management, and significantly reduces the costs and efforts needed to root out terminology issues.
Jan Bares, Chief Technology Officer at Argos Multilingual, explained the motivation behind these enhancements: “Many clients lack the time and resources to maintain and clean their Translation Memories effectively. But everyone needs clean, reliable multilingual data for MT or generative AI applications, and TMs are a real goldmine for most of our clients that they can now better leverage. So we’re excited to release another innovation that builds on our long-term investment into internal AI capabilities, with a range of these data quality checks powered by our own LLM.”
The benefits of the upgraded AI-driven Translation Memory cleanup service are substantial. Clients can expect an 80% reduction in the efforts required for TM reviews compared with purely manual methods. Furthermore, linguistic quality is consistently improved, and machine translation output can achieve a 90+% increase in accuracy.