KantanMT is presenting a two-part blog series on the history of Machine Translation to give our readers a better understanding of the industry and where KantanMT fits in within the grand scheme of things. In our last post, The History of MT Pt.1 KantanMT presented the key stages in the history of MT during 1945-1979. In this post KantanMT highlights the major developments from 1980 to the present day.
The 80s and 90s…
While the EC’s Machine Translation project EUROTRA continues, Japan launches a state supported Machine Translation research programme in 1982. Japanese manufacturing witnesses a surge in the number of English-Japanese MT projects as a result. In Stuttgart in 1984, Trados is founded and becomes the first company to roll out translation memory technology (Multiterm (1992) and Translator’s workbench (1994)). In the same year IBM begins research on using “slot” grammars for Machine Translation. In 1987 the first ever Machine Translation Summit is held in Tokyo, Japan and at a conference in 1988 IBM reports on its experiments in Statistical Machine Translation (SMT) with the Canadian Hansard corpus. The feasibility of SMT becomes a major research topic and represents a break from the traditional Rule-Based methods.
The 1990s begin with Michael Blekhman establishing the first university course on Machine Translation at Kharkov State University. In 1996 Systran offers free translation of small text segments on the internet. iTranslator, the first commercial internet Machine Translation service is launched by Lernout & Hauspie in 1998 while in Dublin, ALPNET launches one of the localization industry’s first language technology integration services.
The 00s and beyond…
In 2001, the US National Institute of Standards and Technology introduces its Open Machine Translation evaluation system. The aim of OpenMT is to help improve Machine Translation technologies. Language Weaver is established in California in 2002 to produce Statistical Machine Translation systems and in 2003 the ISI team wins DARMA’s speed MT competition with, you guessed it, a Statistical Machine Translation engine. In 2004 TAUS is established and the state funded OpenTrad project is rolled out in Spain – the scheme aims to develop Machine Translation engines for the different languages in Spain.
In 2006, the European Commission launches EuroMatrix which aims to develop Machine Translation engines for European language pairs and in 2007, Moses, the Statistical Machine Translation system is launched and incorporated into EuroMatrix. 2008 sees the introduction of text/SMS translation for mobile phones by NEC and in 2009, Apptek combines Statistical Machine Translation with traditional Rule-Based models to produce a hybrid MT system.
There is also large-scale development in cloud technologies after the turn of the century and in 2012, KantanMT is launched as a cloud based Statistical Machine Translation platform.
By providing a Statistical Machine Translation Service in the cloud, KantanMT is drawing from developments throughout the rich history of the Machine Translation industry and carrying the torch into the future.