Many companies have become dismayed at the prospect of building and maintaining a comprehensive enterprise model. The sheer volume of meta data and the rate of change discourage all but the most stalwart, and overwhelm many who do undertake it.
But our experience gleaned from recent work with four large enterprises, and leveraging over a decade of work in the application of semantic technology, we’ve come up with a general approach that is reversing the tide. We call it “Elegance” and it involves focusing on similarity as well as difference. It eschews structure at the conceptual level. It centers on the capture and application of “enduring business themes;” those aspects of the model that will survive your next package implementation or outsourcing project unchanged. It relies on reuse of properties and attributes rather than indiscriminant multiplication.
The net result is that for the enterprises we’ve worked with between 500-1000 concepts and 200-300 properties or attributes are all that are needed to cover anything that is important. Yes there will be taxonomic differences but they are difference of degree and not type. These organizations typically have tens of thousands or more attributes in their existing repositories, so at first it’s surprising that this level of economy could be achieved.
In this talk we will describe our experience and outline a methodology for practitioners to follow to achieve these kinds of reductions in complexity.