How to Make Content "Intelligent"

By Mike Clarke posted 06-15-2010 12:08


I really like the AIIM "Intelligent Information Management" theme.  I would like to take that a bit further and propose a "wouldn't it be nice if" scenario, and then explain what I think it would take to get there.  This means that the content becomes "intelligent" by itself.  It will take an industry association or collaborative project or an open source project to pull it off but it won't ever happen if it doesn't start somewhere so here goes.

In a recent conversation with some colleagues someone said "Wouldn't it be nice if the documents would just figure out where they were supposed be instead of us creating code to analyse great swaths of content files and then push them around in workflows."  This will become increasingly important with Enterprise 2.0 initiatives in the workplace.  So, I agreed, it would be nice and started to think about how that might happen.

Here's what I think needs to be put in place:

1) Global taxonomy project.  This will take analyzing billions of documents over a long period of time.  There are known taxonomies out there, most proprietary, but they only work for structured and single-purpose document types.

2) Expansion of the CMIS standard to include taxonomy and classification capabilities into XML file structures to enable content re-use and auto and self-classification in context of file/content creation.  Standardized file formats need to be able to incorporate metadata from taxonomies and taxonomy projects.

3) The ability to create a document as an object which would operate autonomously in a standards based program framework environment.  This is the tricky bit.  I am not talking about embedding binary code into documents but encapsulating a document in a java object where it would analyse itself and associated metadata and then compare that to it's context and purpose and initiate standardized code in services which would move it to a workflow, change it's status, declare it a record, announce itself for review or publish itself to the web.  This will require a standard services and api framework for the java objects which could then be present in a variety of different systems from various vendors, either installed inside the firewall or available in the "Cloud".

4) Some sort of AI or rules engine based solution that is either available as a service or embedded in the objects, or both.

Sounds like a tall order, but one never knows...  Any thoughts?

#intelligent #enterprise2.0 #content