MIT's ConceptNet: http://csc.media.mit.edu/conceptnet offers a very large dataset of concepts and their relationship to other concepts across 27 different relationship types.  For example, a relationship is pictured below, stating that the class "fish" has the relationship of "being located at" the other class "water."  Each of these relationships has a score assigned to it based on how well other members of MIT's OpenMind initiative judge the relationships veracity, the pictured score of 151 is relatively high, indicating that this is a truthful relationship.

Using the score metric of each edge on the graph (the 151 above), we can sum the incoming scores to each node.  For example, if the relation "people drink water" is represented as a directed edge towards water with a score of 49, the water node would have a total incoming score of 200 (151 49).  We can see the top 10 scores for concept in-degree below:

1. fun (1,385)
2. water (1,010)
3. house (923)
4. store (896)
5. city (837)
6. kitchen (773)
7. sleep (762)
8. animal (751)
9. eat (732)
10. learn (720)

Graphing the distribution of score in-degree, we see the expected long-tail effect.

The long tail effect seems to take place after in-node score decreases below 100.  On this note, we should also consider the score relationships of the edges (relationships).  When we rank the relationships based on the sum of the score weights assigned to them, we see the following ranking:

1. Is A (81,444)
2. At Location (65,080)
3. Has Property (60,671)
4. Used For (60,504)
5. Capable Of (30,117)
6. Has Prerequisite (24,761)
7. Has Subevent (23,795)
8. Conceptually Related To (19,968)
9. Causes (17,797)
10. Has A (17,753)
11. Motivated By Goal (14,297)
12. Receives Action (8,177)
13. Desires (5,872)
14. Part Of (5,465)
15. Causes Desire (4,871)
16. Located Near (4,813)
17. Defined As (3,781)
18. Has First Subevent (3,663)
19. Has Last Subevent (2,528)
20. Made Of (1,906)
21. Similar Size (1,102)
22. Created By (706)
23. Symbol Of (155)
24. Has Pain Intensity (78)
25. Instance Of (65)
26. Inherits From (50)
27. Has Pain Character (43)

Using the long-tail class cutoff, we can visualize ConceptNet with these top 540 "connector" nodes not found on the long-tail, yielding the following, highly interconnected graph:

However, if we constrict the graph only using the "Is A" relationship, it becomes much less connected:

The "Located Near" relationship yields a disconnected graph:

And the "Causes" relation:

While there is obviously a lot more to learn from ConceptNet than making these graphs, I hope that this offers a good introduction and motivation for anyone else interested in ConceptNet.

Today I added visualization for the language area of this site.  If you look at the word definition for the word examination, for example, you'll see that a few definitions exist.  The WordNet database from which these definitions come from provides an "is-a" hierarchal representation of nouns and verbs, enabling visualization of conceptual parent-child relationships.  It sounds complicated, but helps demonstrate the ambiguous meanings that words can have out of context, and the hierarchy to which we speak and think about the world in.

I've seen a lot of websites with live chat features, usually relying on customized software or pay-software such as Olark.  I've always considered creating a live chat system for myself or some of my clients, but have put it off because of the time and/or money involved.

Yesterday, I read about Google's solution to live chat, and got around to installing it today.  It took an astonishing 3 minutes to setup and integrate with my Google Talk account, IM client, and this website.  Below are the steps I used; they should be straightforward enough to get you started with your own live chat system.

1) Register for a Google Account / GMail if you don't have one already, and sign into Google.

2) Visit http://www.google.com/talk/service/badge/New and copy the HTML code it gives you.

3) Paste the code into the HTML source of your webpage where you would like the badge to appear.

4) Log in to your GMail to access your Google Talk account and test your system, sending an IM from your website.

5) If you would like to be available to your visitors whenever you're online (without having to log into GMail), then you can download and install one of these programs, enter your Google account information, and let the program run in the background.

• Adium (Mac)
• Pidgin (Linux)
• Google Talk (Windows)

You can also find a program for your phone so your visitors can get a hold of you 24/7!

Well, I did it.  I finally got around to writing my own blog application.  This application is an evolution from one I wrote nearly a year ago for Fans United Football Club, which was just updated a few days ago with some world cup fun.

In the next week or so, I'll most likely get around to integrating this application with my ontology research, enabling this application to self-write itself with derived database schemas, controls, presentation, etc. based on an ontological architecture to suit a user's requirements.  This is the general direction for my research, to which I'll post more updates on.