Peter is getting one of these new Hitachi 1 TB drives. He remarked:

At $450 each, 1 million dollars will buy you 2000 terabytes!

Together with Gilbert Paquette and Petko Valtchev, I received a $1 million grant from the Canadian Foundation for Innovation to build a Research-oriented Data Warehouse called ERASME. The Data Warehouse will include terabytes of storage and a cluster of powerful machines for high performance computing.

Other researchers involved include Martin Brooks (NRC), Abdulmotaleb El Saddik (U. Ottawa), Étienne Gagnon (UQAM), Robert Godin (UQAM), Owen Kaser (UNB), Hakim Lounis (UQAM), François Magnan (UQAM), Olga Marino (UQAM), Hafedh Mili (UQAM), and Guy Tremblay (UQAM).

I am not entirely sure yet what the XXL library does and why it is important, but it felt important enough to blog about it. It seems to be a library allowing one to build custom database engines. Here’s the summary from their web site:

XXL is a Java library that contains a rich infrastructure for implementing advanced query processing functionality. The library offers low-level components like access to raw disks as well as high-level ones like a query optimizer. On the intermediate levels, XXL provides a demand-driven cursor algebra, a framework for indexing and a powerful package for supporting aggregation. The XXL project provides various packages. See the longer introduction to XXL for an explanation of the packages. The library is publicly available under GNU LGPL and comes with a full documentation.

Unacomputer. Computer that can run in a shack using mostly solar power or other off-the-grid power.

(A reference to the fact that the unabomber lived in a shack.)

Source: Owen.

My friend Dan recently asked me what philosophical questions interested me. Here is my answer.

What is knowledge, what is information? Can knowledge be captured and represented formally? Why is English (or other languages) not evolving
toward greater formalism, but rather evolving toward greater flexibility, fluidity?

Are we living inside a Turing machine?

How do you know that a duck is a duck? Do you have mentally a formal definition that you apply? Do you check successively for some characteristics, applying some kind of algorithm? If you do apply an algorithm, does it terminate? That is, do you ever know for sure it is a duck?

It looks like Value Investing News has turned to Slope One for its recommender system. Here are the features they support:

• You can spot the recommendations engine in action when you click on the Recommendations tab at the top of the page.
• You can also find it in action when you click on other user’s names, such as George. The top of the right hand column will show you a list of Recommended Users.
• Finally, you can find a list of users that liked a particular article by clicking on the Full story links. At the bottom of the each story link discussion you will find a Those that like this also liked section.

eWeek has an audio podcast on the Challenges of the Semantic Web. There is quite a lot of content. Stephen Downes is also part of the podcast and he makes an interesting point: proprietary data and its protection will prevent the Semantic Web from taking off. I am not sure I agree with Stephen on this one. After all, companies like Amazon already give you access to all of their metadata for free. However, I agree with Stephen on the fact that Tim Berners-Lee comes across as a bit hopeful. Proponents of the Semantic Web have a strong tendency to answer objections with the help of technobabble. “We wil XSLTize the problem away!” No, you won’t. You can’t resort to technobabble for ten years. Not unless you are a Star Trek writer.

I find it amazing how we have exactly the same problem at two ends. From SOAP to Semantic Web. Both are plagued by fundamental overestimates of the power of knowledge representation in IT.