For my collaborative course in American Studies there's high demand for building interactive historical timelines into our Web site. There are plenty of instances, especially in Flash, on the Web, but I couldn't find tools to build those suckers.
After a bit of trawling, I've discovered a Java based timeline creator, freely available, from the John Hopkins University Center for Educational Resources. It's not the most visually polished thing in the world, but something's better than nothing.
Link parkin': Randall Hyde on "Why learning assembly language is still a good idea."
Acronym fu. Indiana University's School of Library and Information Sciences has an information visualization group, that's put together the XML Toolkit, software to do, you guessed it, a wide variety of visualizations, including graph visualizations!
No, the name is not a typo or a joke.
Earlier, I mentioned prefuse, a recently released Java graph visualization toolkit. Goonoodling around (I just made it up) following up a reference in the prefuse source I came across Carlos Perez' collection of Java toolkits for graph display. prefuse looks slick but it's alpha. These other kits are probably a bit more mature.
Flickr has added image annotation, akin to Fotonotes, to the toolbox. Assuming some reasonable APIs/protocols to get at the photo annotation data, some social network wonk could have a field day investigating the Flickr community graph.
Tip of the hat to Boing Boing.
Update: More details on the Flickr update, from Corante Many2Many.
Continuing in the slick hacks mode, memecodes is a wicked cool idea. Mutating, in the genetic algorithm sense, web pages based upon search engine referrals. Don't know where it's going or how it will work out in the end, but seems to be stirring things up.
Tip of the hat to Alex Halavais.
I can't crack info regarding the sect of homokaasu to give proper credit, but the rasterbator looks damn cool. Apparently it takes a digital image and generates a wall sized version rastered out of smaller images. No indication of what the smaller images are though.
Now that's a Web service!
Apparently, Google is actually working on making Blogger better. Biz Stone documents the major changes in the improved Blogger, most of which will be ho-hum to prosumer webloggers. Still, this raises the LCD for web authoring.
Someone should do a study of hosted weblogging services. There are probably about three to three and a half eras: EditThisPage, Blogger/Pitas/Manila, Radio Userland / MovableType / Trellix, TypePad and Blogger Redux.
Hmmmm, needs more thought.
Manufacturing reliable and fast communications of purely local, epidemic style, information transmission, is a forte of Cornell CS research. Werner Vogels recaps some of the epidemic communication work done/going on there.
Now if Werner and crew can get around some legal issues and unleash Newswire on the world, we might see some thing new and interesting in RSS land.
Link parkin': unmediated, a group blog about tools and technologies that support decentralized media, might be worth subscribing to.
More grist for the mill, 12 news feeds from Reuters.
For all of us wannabe social network analysts/software builders Jeffrey Heer's network visualization toolkit, prefuse, looks mighty good. The demo applets actually run halfway decently, which is more than can be said for a lot of other gui toolkits.
Tip of the hat to Corante's Many2Many.
Link parkin: A couple of folks at Georgia Tech are working on a P2P based Web crawler.
I keep reading and hearing about Google's "secret" and "propietary" PageRank algorithm. Hey folks, the damn original PageRank paper is on the Web! So's the original Google paper, illustrating that Python is actually decent for something ;-)
Hamish Sanderson (interpreting from the URL) is providing a nice collection of Python utilities. I originally stumbled onto HTMLTemplate but HTMLCalendar looks mighty tempting.
Soumen Chakrabarti chanced upon an old weblog post and shot me a quick e-mail. Focused crawlers are hitting their 3rd generation and he's participating in an effort to build an open source framework for creating intelligent crawlers, iVia, an Infomine creation. Sounds like RSS is on their radar.
Okay PL stuff has been pretty non-existent here, but IBM's Jikes Research Virtual Machine looks cool. A language has reached the "serious" (in an academic sense) when it can self-host, which Jikes RVM does in spades. Not only is it written in Java, the RVM can run the code for RVM on itself!!
As part of the behind the scenes team supporting the project, I'll send what little PageRank this site has to goskokie.com. Put together as a project for this year's Medill New Media Capstone, the site is a living laboratory on Hyperlocal Citizen's Media. Part weblog, part forum, part photo gallery, we're still feeling our way around, but if you have any investment in Skokie, IL, take a gander, contribute, and/or spread the word.
The fine folks at Intel's Berkeley Research Lab are deploying a mobile phone application, Jabberwocky, to investigate "familiar strangers", as part of the Urban Atmospheres project.
I haven't quite parsed what the purpose of the project is, other than to study the relationship we have with folks we regularly encounter, but never interact with. But Eric Paulos, a former neighbor in Soda Hall, is part of the team, and that guy has some chops.
Link parkin': Anand Pillai, howdy neighbor, is working on a new revision of HarvestMan, a modular Web crawler written in Python.
ACM Queue has an interview with Matt Wells, the one man propietor of the Gigablast search engine. Lots of nice morsels in here, including the observation that PageRank is a bit of a red herring when it comes to Google's success. And if carefully engineered, you can do a heck of a lot with only 8 commodity class computers.
Thanks to John Battelle (two ts, two ls, two es) for the pointer.
Have there been any major blogging technology related advances in the last year? Feedster is "enh" for me. Bittorrent and RSS feels like a solution in search of a problem. User interfaces and authoring tools are moribund. Atom is seeping into the ground, but there hasn't been a breakout application.
I know I've mentioned this before, but a desktop aggregator, extensible using a major scripting language, might spark things. The key leverage I'm looking for is hooking into the user's interactions with the aggregator, not synthetic feeds, or remote control. I wanna know when the user reads an item, when they add a new feed, when they delete a feed, when they mark an item, etc. And I want to be able to add new toolbars, menus, visualizations, etc. Yeah, I know Awasu and Radiou Userland can do some of that, but I found the extension mechanisms in both to be a bit clunky.
Still pining for an aggregator platform on par with Gnus/Emacs, or Mozilla.
Tim Bray relates some insider information on serious usage of RSS within organizations that don't give a hoot about the "blogosphere". Key points: RSS saves on bandwidth, and it's useful for monitoring information flows independent of content.
Cameron Marlowe suspects that Blogdex is being gamed. The jury is still out but there's a bigger issue. These ad hoc ranking services, which are often touted as reliable meausrements of blogosphere popularity, have the same problems as Google, with 1/1000th the staff.
The genie's out of the bottle now. How far can Blogdex, Technorati, Daypop, et.al. be trusted?
Link parkin': kernel machines are the generalization of SVMs. kernel-machines.org seems to be authoratative on the subject.
I don't know how well txtkit works, but the screenshots are damn cool looking!
The Schoenerwissen Office for Computational Design has put together a toolbox for doing visual text mining. Not only are there pretty pictures, but a command line interface. I'd grab it in a hearbeat if it wasn't Mac OS X only.
There's probably an interesting little domain specific/scripting language hidden in there.
Alex Hung's ecto for Windows looks like a nice weblog authoring tool. Of course my copy of Frequency is rolling over even as I post this.
Link parkin': PulpFiction is the MacOS X answer to NewsGator. Looks sweet.
And the AppleScript support seems mighty juicy.
Pining for the Mac.
Speaking of iRank, I actually read the paper closely, as opposed to most folks in the blogosphere. Many (or at least the ones I could find) drew their conclusions based upon the Wired News article. A fine piece, but it misses the true import of iRank.
A component of iRank is based upon how a weblog behaves over time. A literal reading of PageRank depends only upon what a site explicitly is, e.g. content and links. In iRank, behavior is used to generate implicit links between weblogs. Implicit links between posts, pages, and sites will be the next wave of crawling, indexing, searching. They're more adaptive and harder to spam.
Besides, it gives all of those machine learning folks something to do.
I don't know if it will lead to a more "egalitarian" blogosphere, but there can be as many implicit measures as opinions, and ergo, many more ranking systems.
Am I brain damaged or is Technorati? There doesn't seem to be any way to search for posts older than 7 days. Makes Technorati a bit less useful as a research tool. Other than pissing contests, what's the purpose of a search engine that goes only 7 days back?
For example, try digging around for information on iRank, the blog ranking algorithm devised to reflect information epidemic effects. You may have to use "infectious epidemic" to get anything useful. On Feedster.
Since I nailed the folks at Waypathbefore, I have to give them kudos for doing a pretty good job on this topic.
Just when I thought "no news is bad news", word comes in that my extended abstract for AOIR 5.0 was accepted. No good deed goes unpunished and now I have to write the damn thing!!
Sussex, here I come.
A free, as in at least beer, Flash image viewing application.
4 years ago, I had a student Dennis Killerich, work on an independent study to implement an MS Windows clipboard that could be shared across the network. We probably weren't the first and definitely not the last to pursue the notion.
However, Porchdog Software looks like they're seriously trying to perfect it with their crossplatform Spike application. Definitely worth a tire kick.
John Batelle has the first and best analysis of A9, Amazon's leap into search.
Jason Kottke opines that since the major usage of RSS and Atom is presentation and browsing in tools like FeedDemon, the term syndication is becoming less appropriate. I agree, but I'm not at a loss for what the right next term is: aggregation.
The unifying element amongst the current crop of microcontent browsers is that they collect from multiple sources. The only other unifying concept is the push, automated update, mechanisms they all have.
James Tauber's weblog is recommended by Tim Bray. Anybody interested in applying machine learning to weblogs gets a gold star in my book.
Link parkin': A tutorial from Microsoft Research on stastical learning methods including how Support Vector Machines work.
Link parkin': IronPython is a CLR Python implementation that allegedly runs faster than CPython.