Catchup on Updates

I finally got around to installing Ubuntu 8.10 and Eclipse Ganymede. I'm pretty much starting over with Ubuntu, as it's going to be a server project in parallel with OpenSolaris. So far I'm still planning to do the bulk of my development in mach-BSD, that is OS X.

Still working on the Swing app with Netbeans. UI v2 will feature new embedded browser panes and more controls.

Predicates Chosen

Orthogonal technically means "at right angles" but I've heard the term used loosely to mean what I was driving at: providing maximum coverage of the problem set with the least amount of overlap. Conceptually, logical right angles might allow one to compose a two-element query where each element can be adjusted independently with linear variation in the results. Conceptually linear? That's why it's more a contemplation, a notion, rather than a rigorous proof. These thoughts lead to what I can call my best educated guess.

After much contemplation, I have my predicates.

My ideal MacBook

Sweet new macbooks, faster bus and memory, new trackpad, and what i really need in the dark, a backlit keyboard. Sure would like to gets me one with 4GB RAM and the largest HD option. Then I'd install some stuff...

First VMWare, Windows XP and Ubuntu 8.10 (desktop and server), bringing the number of operating systems on this laptop to 4.

I'm partial to BSD, so I'd pretty much have my development tools in OS X.

MySQL for the data store...

Eclipse with JBoss, Hibernate, Jena, Jersey... also in Eclipse, pydev and Oxygen plugins.

Netbeans with Glassfish and those same framework libraries...

XCode with the iPhone SDK.

Glassfish and JBoss on the Ubuntu server, where I deploy the applications I build. The Ubuntu desktop iscuz i can haz a learneding enviroment for Ubuntu.

Windoze just in case, and until I find a suitable open source replacement for Visio.

Okay, I just described my current black macbook, but the 'lumin' ones are sweet. That's what I'd do with it.

Orthogonal Predicates

A first foray into semantic web could be a small set of initial predicates. For example, a museum might ask what are the most notable ways two artifacts, events, or people can be related. Person A invented artifact B, and so on. The question then arises, what is the most useful small set of predicates for a given application? How do you minimize their number while maximizing functional versatility?

RDFa and Its Subsets

For any semantic web site, RDFa or similar tags will be added to web pages multiple times for multiple reasons. Families of predicates can work together, and predicate relationships not used by a particular application are effectively invisible to it.

For example, a person looking at the web page, i.e. the visual markup, would not see the non-visual RDFa markup. Likewise, one app might use location-centric relationships and not care about accompanying information. Another app might look at the same page and grab only the bibliographic connections. Part of installing a semantic web application, then, is to initialize the pages with necessary markings. The surface area of a semantic web application is the complete set of tags that it reads and writes. Tags are added, modified, and deleted as the business evolves changes.

Non-interference is an important property.

Interoperability of Domain-Specific Ontologies

Families of predicates can work together, and predicate relationships not used by a particular application are effectively invisible to it.

As the simplest possible example, compare visual markup and RDFa mixed together on the same XHTML page. A person looking at the web page is utilizing the visual markup, but would not see the non-visual RDFa markup. Likewise, one app might use location-centric relationships and not care about accompanying information. Another app might look at the same page and grab only the bibliographic connections. Domain-Specific Ontologies can work independently in this way, each with its own XSLT data extraction.

Now that Oracle Bought BEA

This is their application stack road map or whatever. The former BEA products are circled.

Oracle has more information on their road map.