Auto-classification and metadata –the missing link.

I am doing some research on auto-classification and would sincerely appreciate your input, thoughts, and insight. This is the third in a series on auto-classification. The other two postings are Auto-classification – Has it time come? Too bad no one knows what it does! The second blog is Auto-classification, the only way out.

Auto-classifiers certainly help in organizing and managing content with an automated approach. Pretty basic. I think the key to accurate classification is the associated metadata. The types of classification metadata include intrinsic, administrative/management, descriptive, and semantic. All are useful, however, the more descriptive the metadata, the better the auto-classification will be.

Why is metadata so important? Here are some reasons:

• The Challenges of Content Overload
  • 80% of enterprise data is unstructured (IDC)
  • 60% of documents are obsolete (e.Law)
  • 50% of documents are duplicates (equivio)
• The benefits of automatic semantic metadata generation
  • Elimination of costs and errors associated with end user tagging
  • Identification and protection of secure content assets from unauthorized access and portability in accordance with compliance procedures
  • Automatic identification and tagging of documents of record
  • Normalization of content across functional and geographic boundaries
  • Integration with search
• Ability to apply policy consistently across diverse repositories and
  environments

Given the fact that accurate and meaningful metadata is often the missing link in auto-classification, my question is why more organizations aren’t jumping on the bandwagon? In fact, in a recent AIIM survey, that we co-sponsored, it appears that the term is understood, but not how to get there.

What do you think?