Friday, December 23rd, 2005

AAAI-06 (February 16, 2006 / July 16-20, 2006)

The Twenty-First National Conference on Artificial Intelligence (AAAI-06) will be held in Boston.

AAAI is pleased to announce that the Twenty-first National Conference on Artificial Intelligence (AAAI-06) and the collocated Eighteenth Conference on Innovative Applications of Artificial Intelligence (IAAI-06) will be held in Boston, Massachusetts at the Seaport Hotel and World Trade Center, from July 16–20.

AAAI-06 welcomes submissions on traditional AI topics as well as novel cross-cutting work in related areas.Topics include but are not limited to the following:

Agents

Cognitive architectures

Cognitive modeling

Commonsense reasoning

Game playing and interactive entertainment

Information integration

Knowledge bases and expert systems

Knowledge representation and reasoning

Machine learning and data mining

Natural language processing

Neural networks

Planning and problem solving

Probabilistic reasoning

Robotics

Semantic web

Vision and perception

AAAI-06 specifically encourages submissions that combine results from multiple research areas within and outside AI, such as the following:

Autonomic computing

Autonomous behavior and learning

Grounded representations

Knowledge bases through text extraction

Large-scope and large-scale challenge problems

Learning in problem solving situations

Natural language interaction with intelligent systems

Perception and sensor networks

Robust and scalable architectures for distributed representation and reasoning

security and privacy

Comments (0)

Thursday, December 22nd, 2005

Try Ruby: Online Interactive Tutorial

Filed under: — Daniel Lemire @ 18:09

Standard Deviations suggests we try the Online Interactive Ruby Tutorial. If you don’t know the programming language Ruby and want to get a taste, this is really worth your time. Beautiful. We need to do the same thing for Python.

I’m not switching to Ruby any time soon. Python and Ruby are pretty much equivalent, but both of them are order of magnitudes more interesting than C++ and Java. It is a wonder that job postings never mention either Ruby or Python. Who is running these businesses? Even MBAs ought to know about the relative productivity of a programming language. In fact, they are the ones that should be most worried about all this Java and C++ work going on, just think about all these Enterprise Java Beans we will have to worry about in 10 years!

Disclaimer: I program in C++ and Java almost daily, by choice, and also, because most students only know Java. By using Java in my courses, I actually encourage students to learn to program using Java. No, I’m not contradicting myself.

Comments (0)

SSDBM 2006 (January 27, 2006 / July 3-5, 2006)

SSDBM 2006 will be held in Vienna.

The 18th SSDBM provides a forum for original research contributions and practical system design, implementation and evaluation. In keeping with the mission of the Faculty of Computer Science at the University Vienna, this year’s conference will focus on applications of scientific and statistical databases to Scientific and Statistical Computing, including:

Data and Computing on the Grid

Data Warehousing, Registries and Integration

Ontologies in Scientific Computing

Exploratory Analysis and Modeling of Scientific Data

In addition to this year’s focus areas, SSDBM 2006 invites original research contributions in any area related to scientific and statistical database management, including:

Conceptual models for scientific databases

Data modeling and representation standards

Experiment data management

Very large data bases

Performance evaluation

Scientific visualization

Indexing and query optimization

Managing scale and complexity in scientific data bases

Multimedia content retrieval

Data provenance

Metadata management

System architectures

Comments (0)

AJAX RSS Display

Filed under: — Daniel Lemire @ 1:36

Using JavaScript only without any server-side script, you want to be able to display a RSS feed? Easy! (But it took me 3 hours of hacking.) First setup a web page like this (write your own HTML, this is only an example):

<html> <head> <title>JavaScript RSS Reader</title> <script language="JavaScript" xsrc="rss.js" mce_src="rss.js" ></script> </head> <body> <h1>JavaScript RSS Reader</h1> <div id="rss"></div> <script language="JavaScript"> displayRSS("http://www.domain.com/rssfeed"); </script> </body>

You can download the rss.js script.

New: Download the AJAX RSS JavaScript modified by Theodor Foerster.

Voilà! No need for any server-side processing of the RSS feeds. You can magically display RSS feeds everywhere. Of course, this will only work with recent browsers.

Why is this useful? I found that the typical PHP-based solutions to parse RSS feeds were a pain to setup.

The downside? The RSS feed is read each time the page is displayed. However, the browser does a lot of the work (XML parsing) so I suspect this is an acceptable solution.

Any limitations? The only limitation is that the RSS feed must come from the same domain name as the HTML page because of the JavaScript security model. You can get around this limitation by making local copies of the RSS feeds using simple tools like wget and cron. I really wish the security model could be relaxed as it would make JavaScript much more potent.

Here’s an excellent reference on AJAX: using the XMLHttpRequest object. It is a bit technical, and it is not complete, but it is the most concise and useful reference I found.

Update: Stephen Downes reported this script in his newsletter.

Update 2: Please do not load the script from my server, make a copy! You are stealing my bandwidth! If so many people keep doing it, I will have to get nasty.

Update 3: No, this will not work with Atom. Nor will it work with all versions of RSS. You need to hack the JavaScript if you want to support anything but RSS 2.0. I have nothing against Atom, I just didn’t need to support it.

Subscribe to this blog
in a reader
or by Email.

Comments (18)

Wednesday, December 21st, 2005

Multisite lecturing on a budget using webhuddle

Filed under: Academia/Research — Daniel Lemire @ 23:02

Here is our current plan for this Spring’s CS6905 Advanced Technology on E-business course. The course will be offered in Fredericton and Saint John, and there are three lecturers: Yuhong Yan, Owen Kaser and myself (and I reside near Montreal). Also, while Owen and I are Linux addicts, Yuhong only uses Windows.

Our experiments show that gnomemeeting is quite capable, but unfortunately, it is point-to-point and doesn’t multicast. This means I can’t lecture to both Fredericton and Saint John simultaneously. I also do not know if gnomemeeting works perfectly with netmeeting, but I hope so.

Because I’m not so insanely pretty, we can do without the video stream. On the other hand, multisite teleconferencing is supported both by my employer and by NRC through the normal phone services.

What about slides? You need a tool that will project your slides to several sites. It needs to be cheap, simple and cross-platform. So far, the best tool in this domain is webhuddle. We tested it rather extensively and whereas the audio support is not very credible (the tool has no “settings” to specify which devices are used for audio input/output), it is a brilliant implementation as far as slides are concerned. In theory, you can install it on your own server, though I’m too lazy to do so.

Comments (1)

Wget doesn’t eat XML

Filed under: — Daniel Lemire @ 13:55

I wanted to retrieve a local copy of my online XML course. I instructed the technical staff to serve the XHTML files as application/xml. I believe this was to work around the limitations of Internet Explorer. In any case, I stumbled upon a wget bug! Wget won’t process XHTML with the mime-type application/xml as an XHTML file, and hence, it won’t follow the links inside it.

A deeper limitation is that wget doesn’t know XML. This means that it will not follow stylesheets. Wget also doesn’t know about javascript.

This meant I had to write my own scripts to recover the course. First, a bash script:

You see that the last line is repeated twice. Don’t do this type of scripting at home. Bad design!

Next I need a python script to extract the URLs I need (Perl or Ruby would also do):

#!/bin/env python import re,sys for filename in sys.argv[1:]: file=open(filename) #print “from “, file for line in file: # better hope that we don’t have repeated spaces! for m in re.findall( “(?< = re.findall( "(?<= re.findall("(?<=).*(?=)”,line)+\ re.findall(”(?< =openwindow\(').*?(?=')",line)+\ re.findall("(?< =stylesheet href=["']).*?(?=["'])",line): print "http://"+re.search("www.*/",filename).group()+m

This is a pretty awful hack, but it works!

Here is a project for the tech savvy among you: extend wget so that it can parse XML!

Comments (1)

Metaclass programming in Python

Filed under: — Daniel Lemire @ 10:58

I’m a few years behind, but while I knew you could modify classes in Python, it turns out you can do it conveniently as of Python 2.2.

Here’s how you can dynamically create a class:
>>> from new import classobj >>> Foo2 = classobj('Foo2',(),{"mymethod":lambda self,x: x}) >>> Foo2().mymethod(2) 2

Once the class has been created, you can easily modify it as well:

>>> setattr(Foo2,'anothermethod',lambda self,x: 2*x) >>> Foo2().anothermethod(2) 4

I recall that I found ways to achieve the same result by manipulating directly class objects, but this is much neater.

You can even dynamically change the class of an object:
>>> t.__class__=classobj('newFoo2', (Foo2,), {"supermethod":lambda self,x: 5*x}) >>> t.supermethod(2) 10

That’s pretty satisfying.

Comments (0)

« Previous Page — Next Page »