george tziralis thought log

a Course by Blog

I’m teaching the course ‘Information Extraction Algorithms’ (well, pure data mining in practice) at the post-graduate program  ’Applied Mathematical Sciences’ of NTUA’s School of Applied Mathematics and Physics.

I’m strongly considering the idea of employing a blog to serve at the core of the learning process. The class is held at a pc-lab, so it came somehow natural to me to create a blog where I admin and all students author.

The target? Except from posting lecture notes and publicizing assignments or announcements, I feel that a blog could evolve as the greatest of tools to foster conversation among students and finally enable a really educative experience, powered by the students themselves. The plan is to accept assignments by public posts, instead of filing them out of sight, and motivate each student to comment on the works of others, while learning by the comments received by herself. The concept looks simple and clear, however I have yet to find any similar references (any links would be really appreciated!). Let’s see how it all goes.

I also plan to stream live the whole course (it’s all in greek, though, apologies), while I’m looking for a wiki wordpress plugin to enable collaborative notes’ keeping during the lectures (any ideas?).

The feedback during the introductory lecture today, while we were creating the wordpress blog on the fly and assigning author rights to all students, was really hopeful: “It’s cool, it looks like facebook…” 

cross-posted from a Course by Blog 

CSS Rounded Corners 101

Rounded corners are often a designers’ holy grail. They seem to provide a concrete user experience and a feeling of smoothness in navigation, while they are being selected recently for more and more web 2.0 designs.

Well, I’m not a designer, but need forced me to explore and implement some rounded magic. There exists a lot of available techniques, most of them claiming to be ‘purely css’. But, I finished up disappointed by nearly all of them.

What I actually needed was light, css-powered rounded corners, giving the ability to switch colors easily (aka create customizable by user web pages). Nifty corners arose as a great option, so I gave them a try, however some problems occurred with the javascript calls. For example, you need to put the nifty corners call last, which is not a rather likable feature, and I had some loading errors on my browser.

So, I ended up creating my own technique, which I’m posting here just in case that proves to be helpful for anyone, anytime. I need to declare also that Photoshop is one of few software applications that make me feel really uncomfortable, so I’m putting the guidelines here assuming that you also feel so and have no specific Photoshop skills. Enough, let’s move to the example.

Say you have a div of 600x50 pixels, a header for example, and you need to put some rounded corners on it.

1. Create a new Photoshop file, selecting ‘custom’ as preset and ‘transparent’ in background contents. You’ d better here create an image of size 3x or 4x the size of div (to make the final corners to be created smoother)
2. Select the rounded rectangle tool, define the radius and create a rectangle of size equal to your image (edges of rectangle to coincide with image’s limits
3. Right click on the rectangle and rasterize its layer
4. Now, use the magic wand tool to select the rectangle, then right click and ‘select inverse’
5. Pick up the paint bucket tool, define as color the site’s original background color and fill the corners (they are already selected by the magic wand)
6. After that, select the magic wand tool again and right click on the areas just filled with color to select ‘layer via cut’.
7. Then, delete the old layer (layers window, select ‘shape 1’ and right click to delete)
8. Finally select ‘save for web and devices’ from the file menu, scale down the image through image size selection box and save it as .png (transparency option ticked, file created is about 4KB in size)

That’s it! Now you have a transparent image to put it as background in your div, while you can define and change on demand the div’s color by css. For example, you could define in your css:  

#this_is_my_div { background-image: url(/images/div_background.png); background-position: bottom; background-repeat: no-repeat; background: #abcdef;  } 

So, you end up creating a ‘wild card image’, of tiny size, no need for buggy javascript and color just defined by your css… Hope to be useful! :)

GIGO and prophets, tears and markets

Prediction markets failed to accurately predict the unexpected effect a few tears had on the New Hampshire primaries; and some analysts rushed to blame the tool and undermine its reliability and applicability. Let me restate some fundamentals and my view, in a snapshot:

• Markets are not prophets, prophets do not exist.
• A mechanism’s forecastability should not be judged against a virtual fool-proof prophet; we’d better compare it with other existing or widely-used mechanisms and -to my partial and context-bound knowledge- markets outperform all those.
• Markets are the only tool that intrinsically suggests their probability of failure. If Obama’s stock is traded at 70 cents, this suggests that there is a 30% probability of Obama losing; I’d say markets are by character modest and no fanfare has any place in describing their suggestions.
• Markets are primarily an aggregation/meta mechanism; as such, garbage-in-garbage-out effects are expected to happen, so we’d need to keep focus on minimizing garbage rather than blaming the market/compiler.
• Maturity of the mechanism and its use, as long as trading volume (in real-money intrade for example), have not yet reached a fully efficient level (more on this to come soon), but these result in significant profit opportunities, so I expect things to just keep getting better.

For 25 posts/day coverage and links to relevant articles (including WSJ, ABC, NYT, FT, NBC, etc), check Midas Oracle. 

A really hot year for prediction markets has just started

After the launch of Wall Street Journal’s Political Market (powered by intrade, coverage here and here) and CNN’s Political Market (powered by inkling), McCluskey and Hanson also debuted Presidential Decision Markets.

Today, Cowgill, Wolfers and Zitzewitz just released a full-fledged draft studying markets at Google, which is absolutely exceptional, definitely the most massive enterprise prediction markets application and study ever conducted! (coverage by the NY Times here.) I may return to comment further on this soon.

All of these in just 7 days. Who’s next? ;-)