I am trying to export a project as .xslx, but unfortunately for some odd reason it does not export all sentences. The concerned project is this one: textada
It has 10 documents with 100 sentences each. When I download, it only exports 931 sentences. I remember running into a similar problem a while ago, but am unsure how to fix it. Curiously, it seems to affect only the first document – all other 9 documents are fully exported.
Running into the same problem in another project. This time, the export returns only 795 out of 1000 sentences. The issue seems isolated to the first three documents, which is where my coder was doing the manual labelling. In subsequent documents, where she mainly reviewed AI annotations, the export does correctly return 100 sentences per document.
Thanks for looking into this, I’m a bit puzzled as to what could explain this and how I could avoid it.
thank you for pointing that out! This has to do with how our engine processes our data, so the explanation is a little complicated:
Internally, our AI always adds selections to all sentences in one document - even if they are empty*. The only exception is if a document has (and always had) the option “AI can add annotations” disabled. Then the AI is not allowed to add any (even those empty) selections to a document. So my guess is, that the option was always been disabled for the first** document in your project, thus preventing the engine from adding those empty selections. The .xlsx-export just exports all selections grouped by document and ordered by the sentence position. But since the AI never added annotations to the first** document, only the manual annotations are exported for those.
For all the other documents (which partly have the option disabled by now), I guess that the option was at one point enabled when the AI added its selections.
That much for the behavior, I hope the explanation clears that up for you.
Of course, now that you mentioned it, we realized that this solution is not intuitive at all and will work on improving it in the future.
Best regards
Moritz
*then they are marked as empty but are still there
**first three documents for your second mentioned project