|
| Site Translator : Configure Translation Memory |
Configure Translation Memory
Translation Memory is designed to improve the quality and efficiency of the machine translation process. If used effectively, the Translation Memory feature can improve consistency; add localization to the translated content, and reduce the overall translation time. The Translation Memory contents are updated with every translation. New records are added to the Translation Memory, while less frequently used records are replaced.
You can configure the manner in which records are stored and retrieved from the Translation Memory by using the following options:
- Maximum number of words in a sentence
This option allows you to specify the maximum number of words allowed in the sentence that is to be added to the Translation Memory automatically. The range for this value is 5-20 words.
For example, the value is set to 10 and you are trying to translate an English document to German.
The site translator will scan through the document and only the English sentences with 10 words or less, along with their corresponding translations (irrespective of the word count in the target language) will be added to the memory. The next time the same sentence is encountered by Site Translator, it will directly take the translation from the Translation memory, instead of using the remote translation server. The value you should specify for this option greatly depends upon the nuances of the type of document and the language being translated. For example, technical documents in English are usually characterized by small sentences. It would be better to use a word count of 10-12, so you can add more sentences for future use. However, in content having casual writing, fiction or normal prose, sentences are usually long. If the value is low, very few sentences would be added to the Translation Memory.
To change the value:
1. In Site Translator, on the Options menu, click Configure Translation Memory…
2. Drag the slider to specify the Maximum number of words in a sentence.
- Translation Memory Percentage Match
This option allows you to define the percentage accuracy to be considered when translated text is to be retrieved from the Translation Memory. The Site Translator first compares the text in the document to be translated with the records present in the Translation memory. If a match is found, the record in the Translation Memory is used. Using this option, you can specify the percentage of match. The value ranges from 70% to 100%. Let us consider the following example:
English – They were all searching for the Temple of the king. Spanish – Eran todos que buscaban para el templo del rey.
The percentage is calculated word-wise. If the English document being considered for translation to Spanish contains the text, “They were all searching for the Temple”, and if a 100% match is specified, the text would not be found in the translation memory, since it is only a 70% match. (The example has ten words, and only seven are matching). However, if a 70% match is specified, the text “They were all searching for the Temple” is translated to “Eran todos que buscaban para el templo del rey” as the match is within 70% accuracy range.
It is best to keep the Percentage match high (90% to 100%) for Technical documents or documents with a lot of local flavor. A low value in such cases might result in inaccurate results. An advantage of keeping a low value is that most of the Translation would be done using the Translation memory, saving time. However, it is always recommended to keep a high value to avoid inaccuracies creeping in to the translated text.
To change the Translation Memory percentage match:
1. In Site Translator, on the Options menu, click Configure Translation Memory…
2. Drag the slider to specify the Translation Memory Percentage Match.
|