Eating our own (auto-classification) dogfood and loving it
It’s always been my contention that we folk in information management should use our own products. So that we can experience the same pain as our users are feeling.
So here at Synercon we use our own a.k.a. software to build metadata and classification schemes which we deploy across our business systems and into SharePoint. It works pretty well – we can do faceted search and filtering, and we create multiple views of our data information depending on who’s using it. We use the same terms across all systems so our staff know what to look for.
But like most systems it only works well when you have the time to capture metadata. In practice we don’t have that much time, so we have a significant backlog of unclassified email and legacy documents. And just like you, we feel pain because our productivity is affected, and we can’t afford to bring on more staff just to manage information.
Therefore, I’m overjoyed to tell you that this year is different. With the help of our business partners Pingar we’ve had all of our legacy client records auto-classified and appraised.
Was it difficult?
Not really. We already had all of our metadata and most synonyms managed in a.k.a.. From this foundation we built a new ontology (more about that in another post) which we deployed into the DiscoveryOne autoclassification engine. A few days of rule building and revisions, and then over a few short hours all of our client documents were tagged with 5 metadata classes that we know are critical to our business – Account Name, State, Country, Document Type and Synercon Business Entity.
But wait there’s more!
We also tagged the documents with Functions, Activities and Records Class IDs – not concepts that you would expect to extract from the content. It’s all about the linked data model and given that we incorporated the BCS and retention schedule into our ontology, we were able to able to additionally tag with recordkeeping metadata.
How accurate was it?
For the first pass it was really good. We’re already discovering some really valuable IP that I thought was lost. We’ve identified all of the stuff that we can get rid of. And I love that we now have real visibility of all our client information.
It’s not perfect yet, but I can see where improvements are needed – mostly tweaking of acronyms and synonyms and filling in some gaps. I can’t see it taking too much time because we had good foundational metadata. And some great software at our disposal!
As with everything in information management, best practice evolves over time. What I really like is how the ontological model can be used to blend together the classification needs of business users with record-keeping and data protection requirements.
It might have been a simple project for us but it’s delivered huge business benefits already!