Soget has created a workflow that integrates machine translation and computer-aided translation, the latter based on the use of translation memories created by professional translators.-->
agenzia di traduzioni soget
a
Computer Aided Translation, machine translation, workflow, translation memories, CAT <!--Odescription--> Soget has created a workflow that integrates machine translation and computer-aided translation, the latter based on the use of translation memories created by professional translators<b>.</b>a
aprint this page print this page  
Translations agency   a
Translations agency
a

 
 
a

a

 

 

 

 

   Computer Aided Translation
and machine translation
in a single workflow

a		  

Computer aided translation and machine translation go hand in hand.
a

[The exponential increase in texts to be translated into traditional and emerging languages ​​requires the use of technologies that dramatically reduce lead times and costs]


Until a few years ago, machine translation, known as MT, was a complex and expensive technology, restricted to government institutions and used primarily for the purposes of military and strategic intelligence: the U.S. film "Three Days of the Condor", filmed in 1975, documents its use for the translation of Russian texts during the "Cold War".

For years this technology has sought to reproduce the syntactic patterns of languages, and continued to be used only in restricted areas, without achieving particularly successful results.
 
Things have changed with the transition to a statistical model, based on the comparison of thousands of phrases, made ​​possible by the Internet. A huge amount of multilingual data is available, already in digital format, through the universe of websites, e-books, online documents and regulations.
Google deserves the credit - favoured as it is by its leading position in the collection and classification of web data - for bringing machine translation into the reach of all users, allowing experts and laymen alike to evaluate its characteristics, risks, benefits and quality.
 
Ease of use and absence of charge have undoubtedly helped to promote the technology and, paradoxically, to make it worse. Statistical models give preference to texts with the greatest number of occurrences. Machine translation systems, therefore, are likely to feed on their own errors, providing incorrect translations which, if used uncritically (for example the machine translation of websites), may eventually result in the favouring of incorrect translations over correct ones.
 
A real risk for users of free machine translation services is the lack of data confidentiality. The use of free online translation tools, such as editors, translation memories, etc., offered by Providers, involves assigning the latter the right to use the original and translated text. Consequently, sensitive, confidential or private information enters the Provider's "translation memory" and becomes "public". This aspect should not be neglected as professional translators are often unaware they are using public translation memories, and hence unwittingly add the texts of their equally unaware customers.
 
The usefulness of this technology, however, is undisputed. With the expansion of markets, it is becoming important to translate contents into dozens of languages. Members of social networks communicate and disseminate information on an international scale … assistance, training, culture, and so on, generate such a huge amount of content that there may be insufficient professional resources, and the costs and times involved in standard production flows may be impracticable.
The "quality" of machine translation is still poor. Machine translation is more efficient than human translation as it always uses its resources in the best way possible: it produces fast (time) and cheaply (cost-effectiveness). However, we could say that it is never entirely effective as it "ignores" established objectives. In cases where effectiveness is a primary objective, machine translation can even be a hindrance or a waste of time: promotional and marketing campaigns, particularly communicative texts, training material, etc.
Not all translations are equal. Quality is the result of several factors: selective use of sources based on domain, sector, type of document, and fundamental post-editing activities conducted by specialized translators or revisers.

To produce multilingual texts rapidly and cost-effectively, Soget has incorporated machine translation into a specific workflow.
To improve quality, we have created a workflow that integrates machine translation and computer-aided translation, the latter based on the use of translation memories created by professional translators. Texts obtained in this way are always post-edited by revisers, who carefully reread and adapt same to meet customers' needs.
Technology and expertise thus strike a balance, offering rapid, cost-effective products suited to a multitude of communication objectives.  

Creative Commons License
This article is published under a Creative Commons License.


  Contacts
a
  Preventivo on-line No obligation online quote
within 24 hours!

 



a
a
a
a
© 2012 Soget srl. All rights reserved.
Contacts
a