The challenge of metadata conception

Linear vision of the data

Information technologies historically have developped a linear vision of the data. That means a two-dimensional vision represented as a board. Requesting a database will always end up as a two-dimensional view. The consequence of that is a reduction of our analysis capacity to envision another universe...

Where are we?

The BigData is operational and massively acquiring data or measure stemming from the production tool. Likely, algorithms were developped and deployed in the enterprise. Those rely on multiple data from the BigData to deliver the expected intelligence.

Probably, each individual application owns its own perimeter of meta decription (the meta describes the data) out of the whole BigData perimeter. Are functional areas of asset knowledge being kept in some experimented field people mind ? But, where are we?

- How to find out the (single or several) measure corresponding to any functional use case or context?

- Are the applications sub perimeters of meta description aligned ?

- How are the meta datas used for dashboards and key performance indicators governed ?

- How to know if any measure is already acquired in my BigData ?

A globalized business vision describing my BigData is not available, how to reach such a goal ?

Performance management use cases do not fit into the linear vision

Indeed, the performance management use cases will rely on data or measures in a multidimensional holding asset perspective. The metadata cannot be represented into a mere two-dimensional vision, it will be contained into a tree whose complexity depends on various factors. Resulting in that any measure can be found at any node into the tree. The consequence is that any single data or measure inherits its own characteristics plus the full set of its parents characteristics. What's more, the use cases rely on types and super types of holding assets. And the measure may also be any specific type depending on the functional BigData topology.

Meta describing the BigData is not an easy task...

Yet, it is essential to "linearize" (or render generic) those metadatas so they can be used by the IT tools.

Reconcililing the irreconcilable

Deagital conceived a system that consumes your own BigData datas metamodel and all the information you insert into it. Those two elements are designed by you according your needs.

The Deagital solution wil automatically generate a lexical of functional axis for requesting into a linear vision. Next you just have to request the datas or measures (but also the asset) into their specific types with the whole set of holding assets.

The goal ?

Any request on the metadata matches a functional use case or any context related to your needs of performance management.

The goal is to match the strict set of data from the BigData required for any use case or context.

The goals range from any simple reporting up to operational (anticipation and real time supervision), predictive (correlating the past with damage events)  and introspective (analysing the past to discover effect cause relations) intelligence.