Leveraging Step

From Okapi Framework
Revision as of 19:19, 4 June 2016 by Ysavourel (talk | contribs) (1 revision imported)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Overview

This step leverages existing translations into extracted text units.

Takes: Filter events. Sends: Filter events.

This step allows you to send the text units (or the segments if the text units are segmented) to a translation resource and get a corresponding translation, if one is available. The translation resource can be a TM engine or an MT system. For more information see the "Connectors" page.

  • The text units that are flagged as non-translatable are skipped.
  • The text units that are flagged as approved (i.e. the approved property is set to yes) are skipped.

You can chain several steps if you want to leverage against several translation resources, for example: a TM first, then for all entries not leveraged, use an MT engine.

The matches that have a score equal or above a given threshold are placed into annotations that are attached to the target container or the target segments of the text unit. Optionally, you can also specify to copy the translation of the best match directly into the content of the target, if the score of the match is equal or above another given threshold.

Parameters

Leverage the text units — Set this option to enable the step. This option allows you to keep the step in your pipeline and to set it on and off.

Translation resource — Select a translation resource connector. It can be a TM or an MT engine, the list may vary based on your installation. A summary of the current options for the selected resource is displayed below the resource. Note that Internet-based resources may be slow and result in lengthy processing time. For more information see the "Connectors" page.

Settings — Click this button to edit the settings associated to the translation resource you have selected. Note that some translation resources have no settings.

Do not query if there is already a candidate with a score equals to or above — Select the no-query threshold. If an entry has already a candidate with a score equals to or above the given value, no query is done for it. This allows you, for example, to no use a for-fee translation resource if the entry has already a candidate good enough to use. Use 101 to always allow the query to be done.

Accept matches that are equal or above this score — Select the threshold value under which a match is not taken into account.

Downgrade identical exact matches — Set this option to downgrade by 1% top exact matches that are identical. Some connectors may return a set of exact matches that are identical except for their translation. Because some components may trigger automated processes when the best match is an exact match, setting this option ensure that nothing is triggered in the case there are several identical best matches.

Copy the source into the target if the source content has no text — Set this option to copy the content of the source entry into the target if the source has no text (i.e. it contains only white spaces and/or codes). The copy overwrites any existing target content.

Fill the target with the best translation candidate — Set this option to fill the target with the best translation candidate, if its score is above or equal to the specified threshold.

When the best match is equal or above this score — Select the threshold value equal or above which the best match is copied into the target content. If the score of the match is lower than this threshold, the match is not copied into the target content, but it is still added to the list of potential matches for the given text unit.

Only if the target is empty — Set this option to not overwrite existing translation.

Or if the target is the same as the source — Set this option to overwrite target content that is the same (text and codes) as the source. This option is enabled only when the previous option is set.

Add this prefix to the leveraged translation — Set this option to add a prefix at the front of the translated text, if a match is found and copied into the target, and depending on its score.

When the best match is equal or below this score — Select the threshold value equal or below which the prefix is copied at the front of the translation.

Generate a TMX document — Set this option to create a TMX document with all the entries for which a match has been found. A single document is created for all the input files. Enter its full path in the edit field. You can use the variables ${rootDir} and ${inputRootDir}, as well as any of the source or target locale variables (${srcLoc}, ${trgloc}, etc).

If needed, add a MT prefix to the source text — Set this option to add a MT-specific prefix to the source text if the enttry is coming from an MT engine. This ensures the entry will be only used as fuzzy match in tools that cannot identify the origin of the entry as MT using the creationid attribute.

Limitations

None known.