October 18, 2009
SCARSDALE, New York. This blog aims at a global audience, so why am I writing about a conference in Central Pennsylvania? Let me explain.
I’ve been going to Linux User Group meetings for over ten years now, so of course I’m on the NYLUG announcement list. Last April I got a notice of “a small, low-cost, one-day conference about all things Open Source,” CPOSC 2009 in Harrisburg, Pennsylvania. Open source is important to my work in many ways, Harrisburg has a special significance for me from Antioch College days, and it’s a manageable 3 1/4 hour drive from my home.
Most importantly, I am on the lookout for ways to introduce my two sons to the world of work. My 17 year old son is not particularly a computer person, but I thought this might be a good experience for him, and he liked the idea, so we spent yesterday together at the Harrisburg University of Science and Technology conference center—incidentally one of the best venues of it’s kind I’ve ever seen.
This is how it happens that I am reporting to you now on some of the things I learned yesterday at CPOSC 2009.
The broadest lesson is that what’s of interest in Central Pennsylvania is likely to be of interest around the world, and in particular to many of the people I work with in statistical offices in developing countries. Such is our globalized world.
Here are some highlights that may interest readers of this blog.
Survey of Fuzzy String Matching Algorithms, by Kyle Burton. Ever wonder how Google comes up with it’s “Did you mean …?” messages? Fuzzy string matching algorithms, that’s how. In statistical work, these algorithms are used for record matching, as in matching census and censsus Post-Enumeration Survey (PES) records. During the question period I asked about matching for non-western alphabets, such as Amharic and Mongolian. The reply was that string similarity metric methods would be the best approach because they are independent of the glyphs used by the language. Kyle noted further that the Unicode Consortium has implemented unicode for every known language.
Jump Start Django: The Web Framework for Perfectionists with Deadlines, by Rob Yates. I manage several websites, including, of course, this one. For many years I worked exclusively with static HTML pages hand-coded with a text editor. Then I learned a bit of PHP, and most recently I’ve been using WordPress. WordPress is powerful, has a huge user base, a vast number of available themes (though I’ve never found one I liked out of the box), amazingly many useful plugins, a decent administrative interface, and gets the job done. I don’t like the complexity of the themes, however, and I’ve found customizing it inconvenient to torturous. I like the look of Django and may do some experimenting with it.
Lighttpd Several dozen books (mostly or perhaps exclusively O’Reilly books) and other goodies were raffled off during the conference (my son won one in the first session—off to a good start!). Browsing the books on the raffle table I discovered Lightttpd, a fast web server with a small memory footprint used by Youtube and Wikipedia. Youtube also uses Apache; seems the two servers complement each other.
SQLite Then there are the things one learns by conference follow up. Checking out Django, for example, I discovered SQLite a “self-contained, serverless, zero-configuration, transactional SQL database engine” that claims to be the most widely deployed SQL database engine in the world. Doesn’t “zero-configuration” sound good?
Screencasts and ShowMeDo Searching for Django tutorials, I came upon ShowMeDo, a website hosting and promoting the production and use of video tutorials. Founders Kryan Dale and Ian Ozvald are convinced that “web-videos can be a great way to quickly and efficiently acquire knowledge,” and I think they’re right.
I’ve been thinking for several months of doing video tutorials for training in demographic methods, such as a screencast version of the component projection tutorial I posted last week. I was going to put it on Youtube, but now I’ve more or less decided to put it on ShowMeDo.com. Stay tuned!