Difference between revisions of "Machine Translation"

From Lingoport Wiki
Jump to: navigation, search
(How the process works)
(Supported Parser Types)
Line 17: Line 17:
 
== Supported Parser Types ==
 
== Supported Parser Types ==
 
Machine translation supports the following parser types
 
Machine translation supports the following parser types
*
+
* [[LRM_JSON_Support#json_parser_type | ''json'']]
 +
* [[LRM_msg_Support#msg_parser_type|''msg'']]
 +
* [[LRM_Properties_Support#properties_parser_type|''properties'']]
 +
* [[LRM_rjs_Support#js_parser_type|''js'']]
 +
* [[LRM_strings_Support#strings_parser_type|''strings'']]

Revision as of 17:03, 9 July 2020

LRM and Machine Translation

How the process works

Once you've setup your LRM project to use a Systran vendor for translation then the following occurs when there is a change in a base resource file:

  • The changed text is machine translated to each target locale language
  • In each of the corresponding target resource files the corresponding machine translated text will replace the current text. If the change to the base resource file was a new key/value then this new key/value will be added to the target resource file
  • The changed target resource files will be pushed to the repository.
  • An email will be sent out only if some type of error or modification occurred during the machine translation process. These include:
    • An email is sent if the connection information is incorrect or some other unexpected connection issue occurs.
    • An email is sent if one of the target locales is not supported (see below for information on setting up the locale mapping file)
    • An email is sent if, for some unknown reason, the text could not be machine translated. In this case, LRM will pseudo-localize the text and the email will list those files and keys where the values were pseudo-localized.
    • An email is sent if, for some unknown reason, the machine translation added or deleted some of the parameters in the text. In this case, LRM fixes the text by either adding missing parameters to the end of the line or deleting extra parameters. The email will list those files and keys where parameter manipulation occurred. Example:
      • Text sent for machine translation is {0} people have been invited to {1} party
      • Text returned from machine translation for French is {0} personnes ont été invitées à soirée
      • To prevent a parameter mismatch error, LRM will add the missing parameter to the end of the text ({0} personnes ont été invitées à soirée {1}. Since this may not be syntactically correct, the email contains the information about the modification.

Supported Parser Types

Machine translation supports the following parser types