Box Skills: Searching Card Metadata?
So we recently implemented a Box Skill using the Transcript metadata card template. We're writing OCR data to it ultimately, because we'd like full-text searching that isn't limited to 10k characters as Box's current document cracking provides. We send documents off to Azure Computer Vision, and write the extracted sentences back to Box with the transcript card, and use start/end properties to denote page number. This seems to be working as we can see the Skills pane & transcript card through the UI.
However searching that content doesn't seem to actually be implemented.
Expected Behavior
Given I am logged into Box's website as our enterprise service account
When I search for a unique term that appears in the Skills pane transcript card for a given file
Then I expect the document that contains that search term in it's Skill transcript to be returned in the search results
Actual Behavior
''
''
No matches are returned via Box's search functionality.
This is really making me question the utility of Box Skills if I can't search the transcript card metadata that is written back to Box. To be clear, we're talking specifically about skills card metadata, not the SDK metadata API (which requires specific templates to be searched for specific property values).
Am I doing something wrong or is this literally just not functional...? I was told on a sales call that it worked that way. Any help would be appreciated. Otherwise I'm off to start building our own search index entirely outside of Box, which I'd really prefer not to do.
-
Hello ,
The text that is applied to files through a transcript card should be searchable. I just tested this on my end after applying a transcript card to a file. I searched for some of the transcript text in the Web Application logged in as the user that owned the file. From there, I searched for a snipped of text within the transcript and the file returned. This was the case when searching for the content when logged in as the user that owns the content as well as when searching in the Global Content manager in the admin console.
Note that it can take around 10 minutes for content to be searchable. If you try and search for this text again, are you still not seeing the results that you expect? Perhaps the text was previously not indexed just yet.
-
I will follow up an say it is now working, but what we experienced yesterday and the day before was absolutely no typical latency.
Sure looks like it magically started working after chat support, and calls/emails to my account rep. If y'alls indexing was down for a while, I'd prefer to just know that.
edit: Checked my chat logs and squared with the file creation time: uploaded Tuesday 1:40PM. As of 3:33PM, still no search results returning.
I stand by that not being a latency issue.
-
Thank you for your reply! We did not have any known issues with our indexing (reference: https://status.box.com/) during that time. It appears you've opened a case with our support team, so please continue to follow up with relevant details there so we can further assist. Please note, that since it appears to be working now, there may not be much we can do, but try to confirm if there was a larger issue. However, if you see this behavior again please let us know immediately via a support ticket so we can investigate!
Best,
Kourtney
Box Technical Support Engineer
Please sign in to leave a comment.
Comments
4 comments