This action identifies source texts that have been translated in different ways. It can for example be applied to a project to find inconsistently translated source strings.
Consider the following segments and their translations:
...
lets you find identical occurrences of source texts and to flag them (by setting bookmark or status).
Optionally, you can opt to flag translations if they are identical or similar. See examples at the bottom of this page.
Use cases are:
- Flag repetitions of a source text. This permits to see all the different translations and potentially amend them.
- Flag repetitions of a source text where the translation is identical. Useful to find redundant entries in translation memories or term bases.
- Flag repetitions of a source text where translations are similar. Same as before but accepts minor differences in translation.
Please note that the source text must be identical to be considered a repetition.
The action is configured as follows:
type | " |
FindRepetitions" | string, Mandatory | |
| localeFrom | The language of the original texts. This is typically the source language of the segments. | string, Mandatory |
| localeTo | The language of the translations. | string, Mandatory |
| similarity | Optional similarity threshold. Default is |
0. |
Defines how different translations of the same source text must be to be flagged as inconsistent.
The default value of 100% expects translations to be fully identical.
A smaller value such as 90% would still accept minor differences.When 0: Flags any source text repetitions, without looking at the translation. When 100: Flags any segments where both source and translation are identical When 90: Flags any segments where source text is identical and translations are very similar | int, Optional |
| ignoreCase | Optional flag to disregard case differences in translations. Default is false. If true then case differences in translations are disregarded |
when being compared. | bool, Optional |
Examples
Example 1
Consider the following segments and their translations:
| 1 | Hello world | Bonjour monde |
| 2 | Hello world | Bonjour MONDE |
When choosing ignoreCase = true then no inconsistency is and similarity = 100 then both segments are flagged.
When choosing ignoreCase = false and similarity = 50 then both segments are flagged. You can see that both properties can achieve a similar effect.
Example 2
Consider this example:
| 1 | Hello world | Bonjour monde |
| 2 | Hello world |
By setting similarity = 90, the system disregards the punctuation difference and thus not flag any of the 2 segments0 the system flags both segments. Please note that the similarity relates to the translations.