Kirti Vashee, a well-known Machine Translation veteran and independent MT consultant, is currently writing a series on expert MT systems in his blog eMpTy pages. The in-depth posts and interviews by Kirti not only highlight the MT buyer’s expectations, but also stress what the Expert MT Developers are doing differently.
In his blog Kirti informs and introduces the reader to “competent MT technology alternatives available in the market today.” To date he has spoken about tauyou, Iconic and KantanMT. As Kirti points out, our client base consists of Language Service Providers as well as multinational enterprises. What makes KantanMT attractive to both of these diverse client bases is its extremely customisable, bespoke solution, which can be tailored according to the requirements of each client. Our clients can easily build their own Custom Machine Translation (CMT) engines, or they can opt for our Professional Services team to do it for them.
At KantanMT, the Professional Services Team comprises of Solution Architects, Project Managers, Product Trainers and Engine Developers. This team also has the support of our SRE (Site Reliability Engineering) Team, which ensures that KantanMT solutions stay running 24x7x365 on our cloud.
Adding MT Research into the Mix
In the last few decades, MT has evolved drastically. New technologies are being combined to make MT more effective. At KantanMT, we agree that MT research is critical to improving our solution. With this mission in mind, we established KantanLabs, the Research and Development department of KantanMT, which aims to combine world-class academic research with KantanMT’s leading translation technology. Since its establishment in April 2016, KantanLabs has successfully completed a project in collaboration with the ADAPT Centre and Dublin City University (DCU).
Dr Dimitar Shterionov, MT Researcher at KantanLabs is currently focusing on hybridisation of Statistical Machine Translation (SMT) and Neural Machine Translation (NMT). He recently published an article on Multilingual where he talks about the scope of Neural Machine Translation in the industry today.
Read why Neural Machine Translation might be the future of MT.
Collaboration is the Key
Strong collaboration between enterprises, MT developers, Language Service Providers, translators and academic research is vital to the success of MT’s real-world application. With the explosive amount of content now available online, and our ability to instantly communicate with people from all over the world, we need access to quick translations.
While MT can help translate high volumes of content, translators are the key to developing high-quality MT engines. Their expertise with Language Quality Review (LQR) helps to retrain engines properly and improve quality.
But, to date the LQR process in translation projects have often been rife with collaborative difficulties for the Project Managers (PMs), thanks to the proliferation of spreadsheets and a lack of standardised KPIs. To answer this need, we launched KantanLQR, a new tool that facilitates collaboration between Project Managers and Reviewers, and formalises the LQR process, allowing PMs to download valuable metrics that drive a deeper understanding of how your engine will perform in production.
Selecting the right expert MT provider
As most MT experts will tell you, selecting the right MT solution for your company is simply a question of recognising your requirements and communicating them to your potential MT provider. If their solution matches your need they are already ahead of the curve!