The 1960’s saw the fourth generation of the mathematicians who collectively wrote under the name of the Bourbaki. Led by twentieth century pioneers such as, Jean Dieudonné , Charles Ehresmann , Szolem Mandelbrojt , André Weil , Laurent Schwartz , and Alexander Grothendieck , the Bourbaki and other “structuralists” in linguistics, anthropology and other fields redefined our whole approach to thinking about the world. Those who grew up on the so called “new math” know well the impact of this abstract and model driven approach to thinking.
In the 1980’s the remnants of structuralist philosophy transformed the IT world with the introduction of the concept of an object. Until that point, the word data had meant little more than the most primitive types such as integers, floating point numbers and strings. In fact this view survives today in misguided discussions of metadata, where everything but the numbers or pixel values (e.g. listen to image metadata people) are treated as metadata.
The idea of structuralism that developed in the 1930’s was to look at the world in terms of a collection of abstract elements and the relationships between those elements, with a special emphasis on those relationships. Understanding the world meant to understand the hidden abstract structures that underlay the apparent disorder of physical experience and empirical data.
These notions of the importance of structure and object relationships are clearly reflected in the object oriented concepts that have been developed in the world of information technology starting in the 1980’s with Smalltalk, and continuing through C++, Java and UML.
From the 1990’s onward, objects had become sufficiently main stream that when we spoke of data we meant objects. Even in the case of sensor data we can now take that point of view, since measurement of any quantity is really only well defined when we also express that measurement in terms of measured quantities be they scalar or vector.
Defining our data objects (i.e. our data) then enables us to make sense of metadata. It is then truly “data about the data” meaning things like who created the object, when was it created and so on.
Over the past several years we have seen the “structuralist” approach applied also to the modeling of built environments with the emergence of Building Information Models and more recently cityGML. Rather than looking at the built environment in terms of drawing elements of CAD, we look at it in terms of the structural elements (buildings, rooms, doors, windows) and the allowed or possible relationships between these elements. This is now being extended even further so that the “model” includes the semantics of planning, development and operations.
Data modelers much like their structuralist heirs are indeed engaged in developing a model or theory of their application of interest, whether that is oceanography or real estate assessment.
Next time to look at a database table or a UML diagram or ponder the complexities of business process engineering think of debt we all owe to the pioneers of structuralism and the Bourbaki.