I believe that metadata, with its rigorous information models, might be the salvation for knowledge management at least for IT...
The problem with most KM solutions I have seen is that too much depends on the Taxonomy Gods. That is the all-powerful committee tasked with subdividing the knowledge domains and coding the knowledge resources so that people can retrieve documents. While this is a necessary function, it has hardly proven sufficient -- the uptake of KM as a capability is pretty dismal in a lot of organizations that have tried it.
I think the problem here is the inherent subjectivity of taxonomy creation, and people's unwillingness to invest time in climbing the learning curve of another's knowledge categorization philosophy.
I don't have a general solution for this, but there is a particular section of KM that dovetails with metadata: how to manage the volumes of unstructured documentation generated in building and running IT systems. Source control repositories don't cut it; they are no better than file systems as far as retrieval goes. However, a metadata repository can be an ideal solution. Since metadata is generally objective and unambigous, and does not have to be shoehorned into a hierarchical taxonomy, it provides a more robust framework on which to hang knowledge assets.
For example, a metadata repository might have a record for a given server (assuming the repository is handling technical metadata). This server is a known fact, one that everyone in the IT shop can agree on. Therefore its record is an ideal consensus point on which to hang documentation, because people will go looking for the server by an unambiguous name and will probably be interested in whatever they can find pertaining to it.
The same goes for databases, queues, components, batch jobs -- all the major elements of the IT infrastructure.
Like I said, it doesn't solve the general crisis of KM. But for IT shops trying to leverage their masses of Word and Visio documents, a metadata-driven KM framework is an intriguing possibility.
-ctb
Thank you. I have always been saying that using categories such as "business" and "technical metadata" does not give metadata justice. Most people in the world (IT included!) do not realize that they create metadata on an hourly basis...and most of it is not used for anything other than its original intent...
Wouldn't it be nice if all of this metadata became as reusable as it should be? Consider the fact that when we create files, the file type and its "access procedure" are reusable for life....I am always interested in knowing why corporate information and its associated access does not share the same benefits....
Meta-meta - the secret to success!
Posted by: Adrienne Tannenbaum | October 14, 2003 at 10:29 AM
OMG has formed a SIG on "Legacy Transformation"; see http://legacy.omg.org/ltwg_info.htm.
Such work will definitely rely on loads of metadata for - I quote from their meeting minutes - "automatically extracting
knowledge from existing software assets" (http://www.omg.org/cgi-bin/doc?lt/03-06-02)
Should be interesting. But we needed this 5 years ago. Well, maybe for the Java and .Net legacy now being developed...
Posted by: Gabriel Tanase | November 17, 2003 at 11:27 AM