Zip file format for uploading pre-annotated data in Watson Knowledge Studio

Question

I am trying to figure out how to include a pre-annotated model in Watson Knowledge Studio. I have followed the information found here but it doesn't seem to generalize. As a start I have tried exporting an annotated set from Knowledge Studio to re-upload (using the "Import corpus documents and include ground truth" option). If I re-upload the exported zip as-is this works but if I unzip the folder and then recompress it I get the following error:

A file could not be imported: The imported ZIP file is not in the expected format. Check whether the file was exported from another project. The type system from the same project must be imported first. (You selected 'Import corpus documents and include ground truth').

I have tried using the zip command in Linux (both with and without the -k flag which tries to force to MS-DOS style naming) and also used the compress utility in Windows but I get the same error each time. This is without making any changes to the contents of the folder.

Any help would be greatly appreciated!

Update: I managed to get around this problem by adding and removing files from the zip without uncompressing it. To remove, I used: zip -d corpus.zip "gt/*" and to add (or modify) I used zip corpus.zip documents.json — lomizod

Hiroaki Komine Hiroaki Komine · Accepted Answer · 2017-02-21T01:35:11

Would you please check internal structure of your created ZIP with comparing the original ZIP ? Sometime I got the similar trouble report and found that their created ZIP contains root folder in ZIP structure. WKS expects the same folder structure in the ZIP file.

Zip file format for uploading pre-annotated data in Watson Knowledge Studio

1 Answers