Machine translation (MT) is the process of automatic translation from one natural language to another by a computer. One of the very earliest pursuits in computer science, MT has proved to be an elusive goal, but today a number of systems are available which produce output which, if not perfect, is of sufficient quality to be useful in a number of areas, and to assist human translators.
On 7 January 1954, the first public demonstration of a MT system was held in New York at the head office of IBM. The demonstration was widely reported in the newspapers and received much public interest. The system itself, however, was no more than what today would be called a "toy" system, having just 250 words and translating just 49 carefully selected Russian sentences into English -- mainly in the field of chemistry. Nevertheless it encouraged the view that MT was imminent -- and in particular stimulated the financing of MT research, not just in the US but worldwide.
| Table of contents |
|
2 Linguistic approaches 3 Users 4 See also 5 Free software 6 External Links |
Translation is anything but simple. It's not a mere substitution
for each word, but being able to know "all of the words" in a given
sentence or phrase and how one may influence the other. Human languages
consist of morphology (the way words are built up from small
meaning-bearing units), syntax (sentence structure),
and semantics (meaning). Even simple texts can be filled with ambiguities.
It is often argued that the problem of machine translation requires the
problem of natural language understanding to be solved first.
However, a number of heuristic methods of machine translation work
surprisingly well, including:
Statistical-based methods (the last two) eschew manual lexicon building
and rule-writing and instead try to generate translations based on
bilingual text corpora, such as the Canadian Hansard corpus, the
English-French record of the Canadian parliament. Where such corpora are
available, impressive results can be achieved translating texts of a similar kind, but such corpora are still very rare.
Given enough data, most MT programs work well enough for a native
speaker of one language to get the approximate meaning of what is
written by the other native speaker. The difficulty is getting enough
data of the right kind to support the particular method. The large
multilingual corpus of data needed for statistical methods to work
isn't necessary for the grammar based methods, for example. But then,
the grammar methods need a skilled linguist to carefully design the
grammar that they use.
It was recently revealed that in April 2003 Microsoft began using a hybrid MT system for the translation of a database of technical support documents from English to Spanish. The system was developed internally by Microsoft's Natural Language Research group. The group is currently testing an English – Japanese system as well as bringing English – French and English – German systems online. The latter two systems use a learned language generation component whereas the first two have manually developed generation components.
The systems were developed and trained using translation memory databases with over a million sentences each.
Introduction
Linguistic approaches
In general terms, rule-based methods (the first three) will
parse a text, usually creating an intermediary, symbolic
representation, from which it then generates text in the target
language. This approach requires extensive lexicons with
morphologic, syntactic, and semantic information, and large sets of
rules.Users
Despite their inherent limitations, MT programs are currently used by various organizations around the world. Probably the largest institutional user is the European Commission, which uses a
highly customized version of the commercial MT system SYSTRAN to handle the automatic
translation of a large volume preliminary drafts of documents for internal use.See also
Free software
External Links