Monday, March 16, 2009

Open (scalar) Data

Sean (no relation) over at the always-interesting Park Paradigm is thinking about the relationship between the semantic web and financial (numeric) data : Semantic shemanticrich open data is what we want.

The world is literally awash with very, very useful scalar data types. The big gorilla being integer and fractional[1] quantities that change value over time. I firmly believe that in order to make these things first class members of the Web, they need to live *on* the web.

Simply put, numbers need URIs with RESTian APIs for management. Lets put that layer in place first. Then we can make RDF statements about numbers (and numbers at idempotent points-in-time). We can also provide feeds that describe time series changes using things like XBRL...

That is what the www.timetric.com is all about.

[1] Financial markets love fractions, as do market makers of all forms horses, two-flies-on-a-wall etc.

7 comments:

Bruce said...

I really like the idea. But I'm curious how you will model the data. Might be worth a future post?

stelt said...

Nice, even great if they replace the Flash with SVG.

Danny said...

See also: scovo. It was designed for statistical data, but there's no reason other numerics couldn't be modelled in it.

Sean said...

Giving numbers URIs is something that Hypernumbers is trying to do. Will be interesting to see if they get traction. They are a bit stealthy so not sure how they are progressing since winning seedcamp in 2007.

Sean said...

Sean,

Yes, hypernumbers is one of the entities in this space. I'm hoping that soon now, the number of entities live in the business of giving numbers homes on the Web will reach the recombinant growth threshold number R.

R is a term we use for to capture the economic/biological concept of "recombinant growth" (http://tinyurl.com/czfya2)

R is, I think, in the low single digits in this case :-)

It is straightforward to see how entities can independently host numbers that collectively can be mashed-up to create new host numbers. Less obvious[1] is how the fabric of the Web can be made to support the concept of calculations on those numbers so that "calculating" is as easy as "hosting" is today.

Then of course, there is the minor sounding but hugely thorny issues of naming the numbers (the URIs), namespaces, normative copies yada, yada.

Hard and interesting problems one and all. Problems well worth solving for the value they will bring.

Sean

Bruce said...

Going back to my question on modeling, your mention about "naming the numbers" suggest you understand the numbers as resources, rather than as a property of something else (an instance in a sequence of, I don't know, temperatures, or dollar values, or whatever). So which is it?

BTW, I was also scratching my head about the reliance on Flash. There are some JS-based libraries around for graphing these kinds of data using SVG and/or Canvas.

Sean said...

Bruce,

Yes. I see numbers as resources in the RESTian sense. A number is just a representation of something else. That "something else" is not necessarily on the web at all. Resources are an abstract concept. The only things that gets turned into bits-on-the-wire are representations of resources - not resources themselves.

See http://adactio.com/journal/1469/.

The "something else" is a noun, an object, a thing. It is picked out from the surrounding reality by having some attribute that differentiates it from the rest of reality. That thing is what we observe, what we measure. Humans measure things with their senses. I can measure smell, color, touch...but for the purposes of Timetric, the interesting form of measurement is quantitive - a number.

Now it is no secret, that computers don't do words, or pictures or colors or smells. They only do numbers...Consequently giving numbers homes on the Web creates a happy playground for encoding all sorts of measurable phenomena that are not traditionally considered numbers. For example, to me "red" and 16711680 (FF0000 in hex) are the same thing really. Red is an encoding of reality my mind makes. FF0000 is an encoding of reality my computer makes. The former is a word, the latter a number.

Sean