« Back from FITC | Main | NASA's Earth Observatory up for a Webby »

What Shape Is Your Data?

I've come to think of data as coming in two basic shapes: linear or globular. Linear data can also be branched, but it can fall into a fairly simple data structure or XML schema. Rendering this data is fairly straightforward and tends to require simple loops (nested or not).

Globular data is more challenging to work with. An example of globular data would be mathematical notation and MathML . Why globular? The term came into my vocabulary in molecular biology classes, when describing protein shapes. When thinking about data, a path through the data bends and folds back on itself just like a strand of globular protein. Rendering this data requires multiple passes through the schema and recursive loops.

I also see data as having fluid mobility, when handled properly. A catch-phrase to describe this would be "client-side logic" for rendering data. Like water, the data takes the shape of the container that holds it. This could be a database, with each data point in it's own little compartment of an ice-cube tray. With a query, I can melt many cubes, use client-side logic, and combine them into a comfortable view for the user interface. The same data set can be used to create many different views and can be collected from many different inputs. Data flows in, data flows out.


Interesting distinction, linear vs. globular data. But shouldn't the distinction be not about the data itself but in how that data is processed? For example, well-formed HTML or MXML would be linear for purposes of traversing the DOM tree, but would be globular when being rendered, like determining the widest element in a column and adjusting the widths of all other items accordingly.

hello kristin,
nice ideas. have you heard of "pile"? http://www.pilesys.com
they built an engine only based on pure relations between "atomic" data items. its kind of connetec to your thoughts.

eugen (germany)

Mike, oh yeah, it's the processing of the data that makes it globular or linear and more complicated data structures will have regions of globular and linear shapes.

It's not just the data name/type/values themselves that make the shape, but how those values are processand and rendered.

Eugen, I'll have to check that out. Thanks!