KantanMT Releases Improved Stock Data in KantanLibrary™
KantanMT aims to improve efficiency and productivity with KantanLibrary™, a repository of stock training data to improve and customise translation engines
KantanMT is pleased to announce the launch of KantanLibrary™, a repository of stock training data to improve and customise translation engines quickly. KantanLibrary™ makes it easy for Project Managers and MT engineers to navigate through KantanMT’s high-quality data catalogue in various language pairs and domains.
KantanLibrary™ data is available in various domains, including: Legal, Technical, Financial, Medical, and Automotive. The bilingual training data also consists of more generic domains, which can be used while building non-domain specific engines, to improve the quality of the engine output.
“We have thoroughly cleansed, updated and improved our stock training data, which makes it easier and more efficient for our clients to train high-quality engines,” says Pat Nagle, Project Manager at KantanMT. “KantanLibrary™ is immensely helpful to clients who do not have enough training data to build high quality MT engines and the data in our repository can be used to complement company-specific training data.”
The KantanLibrary™ data is publicly available and are Intellectual Property Rights (IPR) cleared by the KantanMT Professional Services Team. In addition to KantanLibrary repository, KantanMT also provides a fleet of pre-built engines called KantanFleet™. KantanLibrary™ is different from KantanFleet in that the KantanLibrary™ can be used to boost the quality of the translation engine, when the user already has some data available. However, KantanFleet is the pre-built domain specific MT engine that is used when users need an in-domain engine to start translating documents immediately.
KantanLibrary™ is available to all KantanMT clients. They can use KantanLibrary™ data to build their own engines or supplement their existing Custom Machine Translation (CMT) engines, as long as they have an account on the KantanMT platform.