Best practices on document categorization

[JoergPalmer]

I have used OpenKM for a couple of days now, but still very new to it. Actually, after "playing" a little with all the nice features I think it's time to get serious

Could someone give me guidance on how to set up the taxonomy and configuration structure in a clever way? I find myself repeating the same steps, e.g. have a category with the same name like a taxonomy folder. Speaking of taxonomy, I take it that the document names must be unique? Does that mean unique per folder? Isn't this going to be a nightmare with bigger numbers of documents? I mean, do I really need to memorize the existing names before being able to import a new document? How does the thesaurus come into play? Again, I understand technically what it means to set up thesaurus, but what is the most clever way of using it?

In the documentation I could find explanations how to physically use this features. But my question rather is, are there best practices to set them up? Or maybe examples you consider being a good use? Any hint is appreciated!

[jllort]

Is not possible explain in post all you should need to know for categorization in your system, anyway I will try to give you some clues.

1- First of all, try to do some table with files types, and who can access.
2- Create forlder structure based on security access and file types ( normally is good solution, not always ).
3- If you got for example invoices you could store in a folder structure like
/okm:root/invoices/2013/01
/okm:root/invoices/2013/02
...
/okm:root/invoices/2013/12

Or you could do by client
/okm:root/invoices/customer1/2013/
/okm:root/invoices/customer2/2013/
.....
/okm:root/invoices/customern/2013/

Or ( where A-Z is first customer letter )
/okm:root/invoices/a/2013/
/okm:root/invoices/b/2013/
...
/okm:root/invoices/z/2013/


Take it only as suggestions how to get an acceptable segmentation of the data. WHAT's IMPORTANT is to not get a lot of folders, documents ( objects ) under the same parent, otherside the UI performance will not be so good. A good taxonomy should get less than 200 child nodes ( take it as a reference ). If you for example are rendering 1000 child nodes you'll see less UI performance, that will be caused for rendering 1000 table rows in browser ( take in mind you're on browser app, and rendering time is also a paramaters ).

You can make automatic catalog with metadata http://wiki.openkm.com/index.php/Proper ... nistration and scripting or crontab task. Consider take a look here http://wiki.openkm.com/index.php/Crontab and here you can get some good ideas http://wiki.openkm.com/index.php/Utilities

[kumar4004]

Hi

I have two questions regarding loading of folders in Taxanomy.

1) My current taxonomy has a tree structure something like the below.

Okm:root
----Main Folder
-----Subfolder ( If we place a document in sub- folder, it will be shown in right hand side of the openkm desktop).
But we would like to have a template in such a way that the document placed in the sub-folder needs to be on taxonomy and when we click on the document on taxonomy(Left-hand side), it should display the document preview on the right hand side. Can you suggest us how to handle this situation. Below is the Current behavior
1) During startup, root Node and one immediate level children are fetched and displayed with Root as navigation node.
2) During mouse click, the next level children of the node receiving the click event are fetched and displayed with that child node as root for navigation.

2) Can OpenKM handle millions of documents in Taxonomy? I could see the current community version is very slow in response when it tries retrieving more documents that are being uploaded. All the users logged in will have the access to millions of folders. So, we would require something like adding all the folder names from database to taxonomy(Paginating/Scrolling the folders is also fine instead of loading all the documents). Users generally traverse the documents either through taxonomy or search the document/file .

Can you explain me why it takes more time in loading all the folders names? Does the tool stores all the document’s binary data in memory or only it fetches the folder names, file name and fetches the document upon clicking on preview button? Can you please suggest a better approach to handle this.

Thanks
Kumar

[jllort]

OpenKM can store millions of docs. Good taxonomy for UI performance should have less than 1000 object per folder ( 100-200 is the optimal ). Arrived at this point take in mind OpenKM UI is a general purpose UI, is right for almost people but the idea in some cases is build your own. For it last months our efforts has gone in direction to building SDK ( php, java , .net ) what allows you easily build your own UI ( it's other possibility ).

In professional version we're working on file plan ( will be released in 1-2 months ). What automatically does file cataloging, control security, metadata, etc. ( all what you should expect of a file plan ). For a huge repositories with a lot of users this is sure the right direction ( that not mean with community you can get some approach of it ). The OpenKM file plan is a radical change from actual OpenKM concept. OpenKM will be able to get running in traditional way or with fileplan. When file plan is enabled some typical action like user folder creation, user security changes etc. are not under user control, file plan controls in a restrictive way: how, when etc.