-
-1930: BEFORE COMPUTER
In the mid-1930s, a French-Armenian Georges Artsrouni and a Russian Petr Troyanskii filed patents for "translating machines". Troyanskii's originality was to propose both a method for an automatic bilingual dictionary, and also a scheme for coding cross-linguistic grammatical roles (based on Esperanto) and an outline of how parsing and synthesis might work.
However, Troyanskii's ideas were not known until the late 1950s. At that time, the computer was invented. -
Period: to
Historique de la Traduction automatique
Traduction automatique : est le processus qui consiste à utiliser l'intelligence artificielle pour traduire automatiquement un texte d'une langue à une autre sans intervention humaine. La traduction automatique moderne va au-delà de la simple traduction mot à mot pour communiquer le sens complet du texte de la langue originale dans la langue cible. Elle analyse tous les éléments du texte et reconnaît comment les mots s'influencent mutuellement. -
The Translation memorandum
Warren Weaver's 1949 "Translation memorandum" proposed computer-based machine translation based on information theory, code breaking, and theories about natural language. Picture: Dr. Warren Weaver, taken January 9, 1940. Cooksey-14 -
Period: to
RBMT
Rules Based Machine Translation (RBMT) systems were the first commercial machine translation systems and are based on linguistic rules that allow.
* Direct Machine Translation
* Transfer Based RBMT
* InterLingua Machine Translation -
FIRST SUCCESSFUL RESEARCH
Yehosha Bar-Hillel was the first to research MT, followed by Georgetown MT team in 1951.
On 7 January 1954 the Georgetown-IBM experiment was held in New York at the head office of IBM. This was the first public demonstration of a machine translation system. The demonstration was widely reported in the newspapers and garnered public interest. The system itself, however, was no more than a "toy" system. It had only 250 words and translated 49 carefully selected Russian sentences into English. -
CREATION OF ALPAC
MT research programs began in Japan and Russia (1955) and the first MT conference was held in London (1956). Researchers continued to join the field: the Association for Machine Translation and Computational Linguistics was formed in the United States(1962) and the National Academy of Sciences formed a committee (ALPAC) to study MT(1964).
However, rough translations produced were sufficient to get a basic understanding of the articles. They were sent to a translator if confidential or discarded. -
The end of the optimism era
Researchers faced semantic barriers, leading to disillusionment. By 1964, the US government sponsors had become increasingly concerned at the lack of progress; ALPAC finally concluded in 1966 that MT was slower, less accurate and twice as expensive as human translation and that "there is no immediate or predictable prospect of useful machine translation." It saw no need for further investment in MT research; and instead it recommended the development of machine aids for translators. -
The Aftermath of the ALPAC report
The ALPAC report was widely condemned as biased and short-sighted: Soviet Union, Canada & Europe persevered.
USAF (1970) striked back with the use of the Systran system and shortly afterwards by the Commission of the European Communities for translating its rapidly growing volumes of documentation (1976). In the same year, another successful operational system appeared in Canada, the Meteo system for translating weather reports, developed at Montreal University. -
Period: to
EBMT
The Example-based machine translation (EBMT) approach to machine translation is often characterized by its use of a bilingual corpus with parallel texts as its main knowledge base, at run-time. It is essentially a translation by analogy and can be viewed as an implementation of case-based reasoning approach of machine learning. -
The wide availability of microcomputers
The 1980s witnessed the emergence of a wide variety of MT system types, and from a widening number of countries.
* Systran(En-Ru)
* Logos: Ge-En & En-Fr
* PAHO: Es-En
* Metal system: Ge-En
* and major systems for En-Jp from Japanese computer companies.
Other microcomputer-based systems appeared from China, Taiwan, Korea, Eastern Europe, the SovietUnion, etc.
The 90s most notable projects: GETA-Ariane(Grenoble), SUSY(Saarbrücken), Mu(Kyoto), DLT(Utrecht), Rosetta(Eindhoven), Eurotra, CMU project -
The early 1990s
The end of the decade was a major turning point.
1. A group from IBM published the results of experiments on a system(Candide) based purely on statistical methods. Also, certain Japanese groups began to use methods based on corpora of translation examples.
2. Traditional rule-based projects continued.
3. The 1st translation memory systems came to the market(Trados), enabling translators easy access to previously translated texts.
4. Start of research on speech translation : ATR JANUS Vermobil -
Period: to
SMT
Statistical machine translation (SMT) is a machine translation approach that uses large volumes of bilingual data to find the most probable translation for a given input. Statistical machine translation systems learn to translate by analysing the statistical relationships between original texts and their existing human translations. The most important components in statistical machine translation are the translation model and the language model. -
The late 1990s.
The use of MT and translation aids in the 1990s grew significantly due to increased demand for non-translators, met by new systems and downsized, improved versions of previous mainframe systems, particularly software localization. Moreover, MT has grown rapidly in direct Internet applications, where the need is for fast real-time response with less importance on quality: emails, webpages...
It became a mass-market product with MT online
services: first BabelFish and later Google Translate -
~Since 2000 | 1996 - 2012
SMT is now the dominant framework of MT research due to the availability of large monolingual and bilingual corpora, open-source software, and widely accepted metrics. However, rule-based approaches still have relevance for certain aspects of MT, such as syntactic analysis, morphologically rich languages (Russian, Finnish and agglutinative languages), transliterated names(Chinese) , and discourse relations(treatment of pronouns. Hybrid approaches combining SMT & rule-based start to be adopted. -
2013-NOW
Google, Microsoft, SDL, Yandex's researches into neural machine translation imply an optimistic future for the industry. Machine translation is moving from being an untenable high-speed, high-quality option to offer a reasonable alternative to low-visibility content translation. The race for a competitive quality advantage is heating up and machine translation vendors are beginning to take differentiated approaches to "improving" the quality of their systems. -
Period: to
NMT
Neural machine translation (NMT) is an approach to machine translation that uses an artificial neural network to predict the likelihood of a sequence of words, typically modeling entire sentences in a single integrated model. -
Transformers model presentation at IMT-Atlantique
2023: A team of students from the ITM Atlantique demystifies machine translation while highlighting its social and ethical implications.Since the technology itself is not the problem but rather its use, it is important to note that taking into account cost and speed of execution, here is what you potentially stand to gain:
* Increased brand awareness
* Identical service in all territories
* New business opportunities