2011-04-05 07:25:47Update on the data visualisation front
Paul D

chillcast@googlemail...
82.18.130.183

Thought I would post a link to the current state of affairs regarding the data visualisation project.
Today I got the circle clustering code working, although it will probably need some tweaking. You can see it working at this link:

http://homepage.ntlworld.com/podnoid/SKS/cluster14.html

Be aware that the data it is using is not live, although it is a copy from Johns server.

(Note for John, the first year entry in my copy has zero for the year, something to have a look at???)

I'm going to have a go at speeding the animation up, there is a sort algorithm that could be improved a lot, but I'll discuss that in the technical section.

The next stage is to implement the rollover labeling of the circles and then code up the slider. Also the positioning of the clusters relative to each other is something to work on, although a relatively minor task.
I have had some thoughts about how the slider will interact with the circles, should be interesting.

I have also had some thoughts today about an additional feature to add to a later version.

Regarding what graphics library/tecnology to use. Actually this should be a minor issue, since so far the actual graphics drawing code is about 6 lines and I don't envision it growing by a great deal.
I hope it wouldn't be to difficult to convert it to html5/canvas in the not to distant future.

2011-04-05 08:45:02
Paul D

chillcast@googlemail...
82.18.130.183

Quick update!

Added mouse rollover tool tip that shows year and number of papers:

http://homepage.ntlworld.com/podnoid/SKS/cluster16.html

2011-04-05 11:52:11Cool mouseover effect
John Cook

john@skepticalscience...
60.231.60.165

I wonder if the best approach would be rather than display all the papers upon mouseover, what if instead, clicking on the year opens a window with the papers listed. Eg - I could generate a page that contains the year and bias in the URL. Very quick and easy for me to set up. Eg:

www.skepticalscience.com/papers.php?b=1&y=2010

Just a thought.

2011-04-05 12:18:09
Alex C

coultera@umich...
67.149.101.148

I haven't been keeping up with the progress of this new idea, so perhaps this has come up: will there be another sort interface too that makes it easier to find and click on a year?  If I, say, wanted to find the proponent papers from 1961, I don't want to go through the tedious process of moving the mouse around through a sea of tiny circles that don't appear to be ordered in any chronological sense.  Just a thought.  I like what you have done, but I think there should also be a more user-friendly interface if users are going to... use it.

2011-04-05 17:47:49
Paul D

chillcast@googlemail...
82.18.130.183

Hi Alex.

Yeah, the graphic display isn't really meant to be an interface to search or find papers. It is intended to show the numbers of papers.

The new additional idea I have is to have a search box, so you can type some keywords, then a list of papers would come up (probably below the graphics) and the display would clear and circles representing just those papers would be displayed. Haven't thought about this idea a great deal, just popped into my head yesterday.

Maybe a range of years could be a feature??

I'll get the basic app working, then we can think about additional useful features!
I think there are enough ideas to produce a number of progressively better versions of the app.

2011-04-05 18:02:13
Paul D

chillcast@googlemail...
82.18.130.183

Clicking on a circle and displaying a list of papers was an idea that I intended to implement in the second iteration of the app.

My thoughts are:

Phase 1. Get a basic app working that displays the numbers of papers graphically.
Phase 2. Add useful features that use the paper titles, links etc. Ways of finding information, this can also feed back into the display. So that your searches for papers can be reflected graphically.

2011-04-05 21:44:24
Paul D

chillcast@googlemail...
82.18.130.183

Improved animation...

Managed to find a much improved sort algorithm using the javascript sort method. This version should have noticeably quicker animation than cluster16 linked to above. (actually, I'm not sure there is much difference, although the sort code is more compact)

http://homepage.ntlworld.com/podnoid/SKS/cluster18.html

Would appreciate if people would test it in your fav browser.

BTW I don't have access to IE9 so if anyone can test it in that, it would be helpful.

2011-04-05 21:57:16
Rob Painting
Rob
paintingskeri@vodafone.co...
118.93.226.229

I'm using Chrome, and the animation is twitching/flickering quite a bit. Wouldn't want to be an epileptic watching it at the mo.

2011-04-06 00:48:53
Paul D

chillcast@googlemail...
82.18.130.183

Really!

I have Chrome on my PC and the animation is really quick and smooth (i'll check the version I have).

What version/platform/processor speed are you using.

I know it is sluggish in IE7 and Safari on a Mac.

2011-04-06 00:56:09
Paul D

chillcast@googlemail...
82.18.130.183

Oh f***!

Your correct Rob.

Try the cluster16 version in the link above, that is fine.

Looks like the sort method is messing it up in 18.

Thanks Rob.

I'll return to my original home grown sort code for now and take another look at that later.

2011-04-06 05:33:23
Rob Painting
Rob
paintingskeri@vodafone.co...
118.93.217.218

Ville, earlier version is fine on my screen. And don't ask about my computer, she's very sensitive about her age!.

2011-04-06 13:38:39Order
James Wight

jameswight@southernphone.com...
112.213.179.164

The circles seem to enter the animation in no particular order. Wouldn’t it make more sense if they entered in chronological order? I was hoping the animation would depict how the consensus has grown over time.

2011-04-06 18:39:07
Paul D

chillcast@googlemail...
82.18.130.183

The time line slider will add and remove circles. So when you slide it back to 18** it will show one or two circles, then when you move it forward in time circles will be added. At the moment all the circles are drawn in one go.
The original idea wasn't to show any years, the original idea was to just indicate the totals.

The Time line slider is the next feature I will be working on.

2011-04-07 02:10:44
Paul D

chillcast@googlemail...
82.18.130.183

Actually as a result of Robs comment and a 'bug' in Johns XML file. I have uncovered a bug in the sort algorithm code, which results in the earlier years (statistically) ending up in the centre and the recent years on the perimeter. That's why all the small circles end up in the centre, because the database has fewer papers per year for the early years.

2011-04-07 08:58:01
Paul D

chillcast@googlemail...
82.18.130.183

Fixed two bugs. I think this has resulted in the animation speeding up and significant a reduction in circle overlaping:

http://homepage.ntlworld.com/podnoid/SKS/cluster21.html

I'm going to focus on the slider now.

2011-04-07 09:59:26
Rob Honeycutt

robhon@mac...
98.207.62.223

Whoa!  That's freaking cool, Ville!!

2011-04-07 10:59:41
Alex C

coultera@umich...
67.149.101.148

Hehehe, I like pressing the button just so I can see the smaller circles in the green section try to get past the big ones :)

Very cool, love the animation!  May I ask which color is which bias?  Green = pro-AGW, red = skeptic, blue = neutral?

It's pretty interesting too seeing now how large the groups are getting - a good indicator that a couple people are taking the book competition seriously :)

2011-04-07 11:14:15Yes re bias colours - Green = proAGW, red = skeptic
John Cook

john@skepticalscience...
60.231.60.165

I presume the final version will include some explanation

The groups have been growing very impressively in recent days. Would be great if a few others could try to take on Rob and Ari, would really swell out the database, but the efforts of those two have been heroic so far.

2011-04-07 17:49:18
Paul D

chillcast@googlemail...
82.18.130.183

Yeah. Haven't got around to the details yet, eg. text explanations. I like watching the circles jiggle about settling down.

2011-04-07 19:55:34Entertainment
John Cook

john@skepticalscience...
60.231.60.165
Yes, we all enjoy watching the jiggling circles :-)