We wanted to follow up with a more detailed description of how the document categorization works under the hood 

Available Today

Document categories, as shown above, are available for projects with data modeled using asset-centric data models. The categorization functionality is the same as that available in Data Explorer; 

• The categorization rules are based on the NORSOK standard* for document categories.
• Files are first loaded in Search, then checked against the NORSOK categorization rule set.

  1. We check if the doc_type metadata has a match in the categorization rule set. If yes, then that file is categorized accordingly.

  2. If the metadata does not exist we check the title metadata. The title is dissected by the ‘-’ (dash) character. All of the dissected parts are checked for a match against the categorization rule set. The first match is used to set the category. 

  3. If no match is found in either doc_type or title, then the file will be placed in the Not categorized category. 

Coming up next

If your data is modeled as an industrial knowledge graph, document categories will become available in the next official release on December 3. Categorization will be based on how you have populated the CogniteFileCategory property. 

*Upon request, our team can share the NORSOK standard. 

Best, 

Sofie Berge, Product Manager
 

Pretty cool….can see categorized documents!!
@Sofie Svartdal Berge  Will it be possible to manually override a document category?

Hi @Aditya Kotiyal !
It is not possible to overwrite the document category for Files in asset-centric data modeling, but when working with Files in Data modeling, the flexibility will be complete

For Data modeling, categorization will be based on how you have populated the CogniteFileCategory property. 

Best, 

Sofie

Hi Sofie! We are very excited about this development and have already been sharing with our users.

A request I shared with Diego Antonino yesterday is to improve how the files are loaded. Our Assets often have 300-400 files associated, and the default view limits the number of files shown (1st image). I have to click “Show More” about 10 times before I can see all of the categories & documents (2nd image).

I know loading all 300-400 files at first view might decrease performance, so my suggestion is to load all the categories as empty “folders”, and then when the user expands a category, load all the files within that category.

Also, we recommend the categorized view to be the default, as that is what people use most often. 

Lastly, we requested in Zendesk ticket 12997 that files be listed in alphanumeric order. On 10/23 Jay Van Sickle mentioned Cognite was “targeting to include this in the release of Document classification in Search,” but it likely was not included due to the swift deployment of Document classification. Will listing files in alphanumeric order be included in the Dec 3rd release, or sooner?

Thank you for this thoughtful feedback, @Emily Viola! 

• For how the files are loaded, I understand how the current way it works is cumbersome. I’ll bring it back to the team and see if we can improve it. 
• About making categorized view the default - we agree with you. This will be coming!
• The Zendesk ticket about listing files in alphanumeric order; sorting is already available, also when in categorized view. Have you seen the ability to sort on name (and any other metadata)? Let me know if this does match your request, I believed it would See example in gif below. 

Best, 

Sofie