ApSIC Xbench Forum

Grouping errors

Hi everyone,

I would like to know if there is a way of showing all the possible mistakes in a segment altogether. That is, for example, if a segment has several key term mismatches, the QA would show the sentence with all the warnings above, all grouped. This way, all the mistakes can be corrected in a go. In particular, this method is needed when reviewing TMs. With the current visualization, there is always some danger of correcting each time only one error and then generating multiple translations for the same source sentence.

Thanks in advance,



I second Almudena’s suggestion: it would be great if it were possible to group errors in a more flexible way, so that we could see all errors (even of different types) in each segment.
I think that a workaround is already possible: export the QA report to Excel, and use Excel’s capabilities to resort the errors – but the process would be manual and fairly time-consuming. If it could be automated in Xbench, it would be very useful!


@RRLL_LS in general it is a good practice to do a final Xbench QA pass to ensure that no source/target inconsistencies were introduced during earlier passes.

The rationale behind the current presentation is to allow skipping over false positives a lot faster (for example a misconfigured checklist item or unsuitable check can produce many false positives of the exat same kind). Also, source/target inconsistencies affecting or example 20 segments are a lot harder to handle if they appear duplicated in 20 segments that if they appear in single one block with 19 segments already correct that you can skip over and 1 segment incorrect.

1 Like

Thanks for your reply, Pcondal. I’d like to have more flexibility on this, allowing the user to decide how to group.
My use case: as a language resources coordinator, I use to check translation memories against key terms and checklists and then export results to an excel file for native speakers to correct issues. You can imagine that some segments in TMS contain more than one problem very often. Hence the need for grouping on segments, even if only, for segments posing more than one issue. Would it be possible at all?

Best regards,