Sunday, July 24, 2005

Off to AAMAS

Tomorrow I'll be attending two days of workshops at AAMAS. Agent-Mediated Knowledge Management on Monday, Service Oriented Computing and Agent-Based Engineering on Tuesday. Sadly I can't stay for the main conference due to conflicting family engagements. AAMAS is usually a fun conference, and it's nice that it's located practically on the doorstep this year. There's a direct flight from Bristol to Schiphol on KLM, then a 40 minute train journey to Utrecht - assuming my lack of Dutch language skills doesn't get me lost!

Friday, July 22, 2005

Java Universal Network/Graph Framework

Christoph Kiefer mentioned on jena-dev a Java graph layout package I've not come across before: JUNG - Java Universal Network/Graph Framework. The example screenshots look pretty good and the jws examples all worked (good sign!). I've been looking for a graph layout package to use with RDF and OWL data sources. JUNG could be just the ticket.

Tuesday, July 19, 2005

FAQ: ontologies in business

A colleague (thanks Dave!) spotted a rather nicely done FAQ on ontologies in a business context. Not sure that I agree with everything it says, but at least it tries to approach what is often a rather abstract, technocratic subject with a down-to-earth sensibility.

Thursday, July 14, 2005

Strategic retreat

Disappointingly, I'm giving up (for now) on continuing writing my thesis using open-source tools and DocBook. I still think that there are good reasons for having the source code to my thesis text encoded in an open standard language, and I do like the idea of being able to generate both pdf and xhtml from one source. The basic structures of DocBook - books, chapters, sections, paragraphs, lists, etc, all work fine. Where I ran into problems was with:

  • citations,
  • formulae and equations, and (to a lesser extent)
  • diagrams
Sure there are open-source tools that help with all of these requirements, but they just don't work well enough. Over the past many weeks, I've spent more time wrestling with getting software working, getting the right entries in my XML catalog, tweaking php files and MySQL tables, and grokking complex XSLT stylesheets than I have actually writing. So much so that it's getting quite desperate now, deadline-wise! I have to say, as well, that the default output from the DocBook stylesheets is pretty dull, so I was expecting some heavy-duty xslt hacking to get a reasonable end result anyway.

So, my plan is to stage a strategic retreat and use OfficeXP for the time being, which as drawing tools and equation editor built-in, and integrates nicely with EndNote and ProCite. Sure, professional bibliographers complain about EndNote, and doubtless it has its flaws. But right now, it's an attractive proposition for not being as much in my way as the existing crop of open-source tools.

I characterise this as a retreat, not surrender, on the basis that I expect OpenOffice to catch up with Word for equations and citations eventually. I have tried OpenOffice for this project, but it the 2.0 beta kept crashing on me and the 1.1.3 version is too primitive. In time though, I expect to be able to convert my document to an open XML standard, either DocBook or another comparable format.

I do thank those who've helped me get open-source programs working (or tried to) over the past couple of months. It's not that I don't appreciate the work and commitment that goes into open-source development, it's just that 80% working isn't enough that I can work with the tool rather than on the tool.

Monday, July 11, 2005

Fixing yum

I have two Fedora Core 3 systems, one in the office and one at home. Yum always works very smoothly on my home system, but at work I'm frequently bugged with:

[Errno -1] Metadata file does not match checksum
leading to
[Errno 256] No more mirrors to try
Googling around, this seems to be a caching issue. Various HTTP caches (on our network or between us and the backbone, I'm not sure) are caching either the metadata or the checksum file, leading to a mismatch. I've tried various suggested solutions for this, with no success so far. However, my colleague Steve Pearson suggested a fix which works in our office, since we have a SOCKS proxy available. I simply:
  • installed the dante socks client rpm
  • removed my http_proxy environment variable
  • run socksify yum update, or whatever yum command I need.
The reasoning is that the SOCKS protocol doesn't, I'm told, use any caches. Anyway, it works for me and hopefully may be of some use to other people faced with the same problem.

Wednesday, July 06, 2005

London olympics: should be a great event

Big congratulations to Seb Coe and the London Olympics bid team on winning the competition to stage the 2012 Olympics. My friends who went to watch the 2004 Olympics in Athens had a wonderful time, so I'm looking forward to catching some of the events live in seven years time. Wonder if it's too soon to book hotel rooms now?!

Updated: changed the title after the awful events on Thursday.

Monday, July 04, 2005

Space scientists have all the fun

NASA consistently does good work in AI. I've blogged before about the Earth Observing Satellite project. There was autononmous fault diagnosis on Deep Space One, and the project to organise ISS astronaut's daily schedules. I've heard from my friend Simon Thompson that there's a very impressive NASA paper at the AAMAS industrial track this year. Now they're installing speech interfaces on the ISS, as astronauts have trouble holding a printed manual and performing complex maintenance tasks while floating about in microgravity (no kidding!). Cool stuff.