Kamis, 04 April 2019

Computer Aided Translation (CAT) & Computer Translation (MT)

Computer -Aided Translation (CAT) Tools are software application that assist in translating content from one language to another. 
Translation tools, also known as CAT (computer-aided translation), enhance the productivity and consistency of translators. They usually include several component technologies in a single integrated workbench, such as document editors, terminology management, and translation memory. The tools have evolved along with the computing and networking industries, first as stand-alone software to be used on a single computer, then client-server tools to be used on a company network, and most recently to cloud-based tools delivered via the internet.
A computer-assisted translation tool facilitates the translation process by resolving the text into smaller, translatable segments. It organizes these segments of text in a manner which makes it easier for the translator to translate the text effectively, and makes the process of translation time-efficient. The segments can be recalled later on and thus the translator ensures that the terminology and writing style of the original is followed. It also provides savings when the material that needs translation is similar to previously translated material – you only pay for part of the sentence that has changed.
A CAT tool further reduces any chances of errors by saving the translated segments along with the source phrase. A translator can easily access any of the translated segments at any given time to ensure that the segment has been translated adequately. Computer-assisted translation tool has been developed to allow the translator to be able to quickly search and alter-if needed- a particular segment of the text. It has been programmed to assist the translator with the revision in a timely manner.
The computer-assisted translation tool saves the translated segments and source segments as distinguished translation units. It saves these segments in a translation data base, referred to as the translational memory.
Translation units stored in the translation memory can be accessed at any given time to be re-used either in the same document or a different one. Computer-assisted translation tool is equipped with special search features to allow the translator to access segments of translated fragments even when two segments do not match completely.
A computer-assisted translation tool leads to the effective use of time and energy resources during the process of translation. It also helps to create a translation memory, which can be utilized later on as training corpus for machine translation engines.
The availability of translation units in the translation memory also ensures consistency in the usage of certain terms. This adds authenticity to the translated document.
A computer-assisted translations tool can provide with much needed assistance when translating texts of considerable sizes. It effectively aids the translator in the processes involving correct use of terminology, maintaining consistency and proof reading.
At times it happens that the translator is required to translate content which is highly technical in nature. In such cases, or otherwise, complying with a given list of specific terminologies to be used accurately in the translated document can increase the work load for the translator. Going back and forth to the provided list to tally your document with the given list of terminologies can also be quite time consuming. Computer-assisted translation tool can be programmed to automatically identify any of the places in the document where a specific terminology can be used suitably.
Computer-assisted translation tool works in compatibility with the requirements of the digital age. In most cases we can made to work in paper-less environments and computer-assisted translation tool can be of great help while working in such surroundings. It comes with the technology to provide backup support to a translated document ensuring that in case of any mishap, a copy of the document can be retrieved.
Computer-assisted translation tool has added greatly to the efficiency of translation process while ensuring the availability of quality translation. It is an essential tool for both translators and buyers of translation services.
CAT Tool Features
CAT tools are similar to a text editor, such as Microsoft Word only with additional features useful for a translation professional. There are many products on the market with different sets of features included. Below is a list of common features that are can be found in CAT tools:
Term bases are translation glossaries that are built from frequently occurring words or phrases, such as technical terms and brand names. They are used to pre-translate recurring words and phrases, and to assist translators in maintaining consistency.
Translation memory is an aligned record of previously created translations. When identical or similar segments are found in a new translation project, translation memory allows the reuse previous translations with or without modification.
Quality assurance relies on a combination of technology and processes to prevent errors from creeping into translation projects. The QA process starts before a project is sent for translation, continues throughout translation and editing, and lasts until after finalizing the new text in the final format.
Resource lookup gives access to online and offline resources, such as dictionaries and reference materials.
Word counts and match analysis provides number of words or characters along with any TM or internal matches. This forms the basis of quotes for translation services.
Term extraction allows the extraction of term lists from translation memories or translated and aligned documents to use in termbases.
Conversion of files between translation-specific (.tmx, tbx, xliff, etc.) and source formats.
Alignment is the process of matching segments in the source text with their translated renditions in order to create new translation memory files.
Machine Translation integration allows connecting an external MT engine which displays machine-translated words and phrases to the translator, who can accept or reject them.
Concordance searches are manual searches in a translation memory for a particular word or phrase.
Predictive typing suggests complete words and phrases to translators as they type based on the content of dictionaries, termbases and translation memories.
Spell check is a more advanced form of the tool typically seen in simple text editors. It allows the selection of not only language and flavor, but also specific grammar and style rules.
Software localization features allow translators to work on software files directly and see the results of their work in a mirrored rendition of the user interface of the software being translated. Once translation is done, the localized strings are automatically reintegrated in the software.
Controlled authoring prevents writers from using inconsistent terminology, highlights common errors, and produces standardized output that is search engine friendly, and translation ready.
CAT tools are generally used by individual translators and editors. More robust systems used by translation providers and translation departments inside corporations that allow for management of linguists, management of translation workflows, invoicing and billing are known as Translation Management Systems (TMS).
Computer Translation (MT) Machine translation is accomplished by feeding a text to a computer algorithm that translates it automatically into another language. That is, no human is involved in the translation process.
The advantages of machine translation include cost and speed. Computers can process a machine translation almost instantly. There are free programs such as Google Translate that can translate relatively short texts instantly, but if you need to translate a very long document, you can purchase software that can process an unlimited amount of text at the cost of the software alone. There is also software available that can be integrated with other computer and online tools, providing instant translations in various contexts.
The major disadvantage is lack of accuracy. If you’ve ever used Google Translate to attempt to understand a text in a foreign language, you will know that this method does not produce a particularly natural-sounding or accurate translation. Language is highly complex and dynamic, and while this type of translation technology has improved greatly over the years, it will never be able to completely accurately identify the nuances of each language and transfer them into another language.
It is possible to hire a “post-translation editor” to look over the translation and correct errors, but it can be harder to correctly deduct the meaning of a sentence from its machine translation than from its original language. Translators hired to “smooth out” such translations sometimes end up asking clients to send them the original text because the translation was unintelligible. This is a big waste of everybody’s time!
The best use for machine translation, then, is when you need to understand the general gist of a text. If you need an accurate translation that anyone can understand, you’ll want to opt for a computer-assisted translation.

MT Systems

Generic MT usually refers to platforms such as Google Translate, Bing, Yandex, and Naver. These platforms provide MT for ad hoc translations to millions of people. Companies can buy generic MT for batch pre-translation and connect to their own systems via API.
Customizable MT refers to MT software that has a basic component and can be trained to improve terminology accuracy in a chosen domain (medical, legal, IP, or a company’s own preferred terminology). For example, WIPO’s specialist MT engine translates patents more accurately than generalist MT engines, and eBay’s solution can understand and render into other languages hundreds of abbreviations used in electronic commerce.
Adaptive MT offers suggestions to translators as they type in their CAT-tool, and learns from their input continuously in real time. Introduced by Lilt in 2016 and by SDL in 2017, adaptive MT is believed to improve translator productivity significantly and can challenge translation memory technology in the future. 

MT Approaches

There are three main approaches to machine translation:
  • First-generation rule-based (RbMT) systems rely on countless algorithms based on the grammar, syntax, and phraseology of a language.
  • Statistical systems (SMT) arrived with search and big data. With lots of parallel texts becoming available, SMT developers learned to pattern-match reference texts to find translations that are statistically most likely to be suitable. These systems train faster than RbMT, provided there is enough existing language material to reference.
  • Neural MT (NMT) uses machine learning technology to teach software how to produce the best result. This process consumes large amounts of processing power, and that is why it’s often run on graphics units of CPUs. NMT started gaining visibility in 2016. Many MT providers are now switching to this technology.


References : 

Tidak ada komentar:

Posting Komentar