Tuesday, April 20, 2010

Lithology logs: using Psicat XML to create SVG outputs

PSICAT is a very nice tool to create lithology logs for sediment cores. One of the most interesting features of PSICAT is it's own XML export format. The format is very simple, it basically consists of two element: 'model' and 'property'. 'Model' is just a simple container for data and can represent an Interval, Bed or Lithology etc.. A 'model' can contain several 'property' elements, which represent simple key value pairs. Here is an example:

<model id="158cc6_1250de53dc5_15" parentId="1899213_1244fb3b1d8_1" type="psicat.core.interval.Interval">
<property name="depth.top">0.00</property>
<property name="depth.base">0.60</property>
<property name="grainsize.base">-1.56</property>
<property name="grainsize.top">-1.56</property>
<property name="contact.type">sharp</property>
Last week I was playing with this format and was surprised how easy to handle it was. In an attempt to do something useful with it, I tried to create an SVG (scalable vector graphics) output based on PSICAT data. And it worked! The first version is able to plot Intervals and Beds quite nice. You can try it here: http://www.stratigraphy.net/psi2svg

UPDATE: The code and tool has been moved, please visit http://www.lithologs.net !

Tuesday, April 13, 2010

Boring latin names are banished

This funny link just came in via the taxacom list:

Dudley UnEarthed

On the lower floor the museum has its major interpretative displays in a recently refurbished gallery called ‘Dudley Unearthed’. The gallery features two time lines, one showing the place of the famous Silurian Geology of the area in the history of the Rocks of Britain and the other showing the place of the industrial revolution in the human history of the area.
Boring latin names are banished but if you want more information just ask !
Well... generally applied, this would solve much of our taxonomy troubles ;)

Monday, April 12, 2010

Twittering data - what to do with dit.li?

I started using twitter last year and was quite impressed by the number of tools offering extra services on top of it.

Onfortunately there is nothing out there which is suited to send data.
You can of course simply link data to tweets via a short url (e.g. tinyurl), but it is impossible for your followers to identify what's behind a shorturl before you click on it. Could be data could be anything else..

Also data dois won't work unless there is something like a short-doi as twitter will shorten doi links to tinyurls.

But wouldn't it be much cooler to have all the twittered data at one -trusted- place? Twitpic is a good example how this works for twittered images.
To mimic a service like this for data and to allow to identify twittered data, the first thing needed is an own short url.

Fortunately, I got dit.li, one of the rare, left, semi-meaningfull three letter domains and so I started some trials:
My first approach was to allow manual data entry in a textarea or *.txt file upload to save data at dit.li. I also added some nice javascript charts to display the data. The twiter API is quite simple, so it was easy to allow sign in via the twitter authentification service. Sending the data by twitter was also very easy by using the twitter API. Some example how sent looks like is : http://dit.li/018. I also started to play with data dois, so you can now also send data by just entering a data doi.

All good so far, but after having fun programming all this I lost all confidence in this project. Is it be useful to send data via twitter anyway?
Any feedback is welcome!