Empty pages count as "not ready for MT" in dashboard

digitalfabrik / integreat-cms

Simplified content management back end for the Integreat App - a multilingual information platform for newcomers

https://digitalfabrik.github.io/integreat-cms/

Apache License 2.0

55 stars 33 forks source link

Empty pages count as "not ready for MT" in dashboard #2317

Closed timobrembeck closed 1 year ago

timobrembeck commented 1 year ago

Describe the Bug

The value of "ready for MT" in the dashboard calculates the percentage of all pages with HIX above the threshold in relation to all translations. This is incorrect since e.g. empty pages without a HIX score are allowed for MT (but do not increase the percentage in the dashboard).

Steps to Reproduce

Go to a region dashboard
See percentage of pages that are ready for machine translations
Check that the percentage gets lower for every empty page

Expected Behavior

An empty page should be considered ready for MT by default

Actual Behavior

An empty page is calculated as "not ready for MT" in the overall percentage counter in the dashboard

Additional Information

Traceback

``` ```

seluianova commented 1 year ago

In addition to empty pages, we also have pages that are not empty but do not have HIX. For example, if TextLab API returned an error. Should such pages be considered as ready for MT or not ready for MT?

Currently such pages also decrease the percentage and it's not quite obvious, because they are not listed in the table.

UPD. processed at my discretion in PR 2320

osmers commented 1 year ago

Maybe it would already help if we closed this issue: https://github.com/digitalfabrik/integreat-cms/issues/2296 I just had the case today that basically all pages have a HIX above 15 but the percentage says only 85% are ready for translation: https://admin.integreat-app.de/odenwaldkreis/ check the dashboard here

Maybe we can even separate it further and say: XX of YY pages are ready for translation. ZZ pages have no HIX value bcs they have no content. They can be translated. ÄÄ pages have a hidden HIX value and cannot be translated.

XX = pages with HIX above 15 YY = pages with a HIX value ZZ = pages with only a title ÄÄ = pages with hidden HIX value

seluianova commented 1 year ago

Maybe it would already help if we closed this issue: https://github.com/digitalfabrik/integreat-cms/issues/2296

@osmers yes, that's basically what we are doing.

But we use a slightly different calculation. The message is just: XX out of YY pages are ready for automatic translation. Where: YY = all pages with HIX enabled (including empty pages and pages with HIX errors) XX = YY - (pages with HIX below 15)

There was a discussion about it in PR #2320, you can take a look.