Two new books

Two books I just finished reading:
* The Search by John Battelle;
* The Long Tail by Chris Anderson;

Impressions of SOLI 2007

The 2007 IEEE/INFORMS International Conference on Service Operations and Logistics, and Informatics was held in Philadelphia. This conference is clearly more industry-oriented, although it still follows the classical academic format (and as the proceedings are published by IEEE and indexed in IEEE Xplore, there's even quite strict technical format to follow).

Some highlights from the conference:

• Keynote "Service Supply Chain Customer Value Delivery: Optimization, Performance Incentives and Implementation" by prof. Morris A. Cohen (Service supply chain strategy & solutions; supply chain coordination; global operations strategy; product design management; benchmarking of manufacturing/logistics systems; manufacturing/marketing interfaces; procurement/supplier management);
• NSF Workshop on Service Science;
• Papers on applying data mining, customer behaviour modelling and forecasting on real-world data;
  

In my free day after the conference, I took a rental car and headed towards Lancaster county, with a plan to see a bit of Pennsylvania countryside and Amishes.

Impressions of WORLDCOMP 2007 / DMIN 2007

The 2007 World Congress in Computer Science, Computer Engineering, and Applied Computing (WORLDCOMP'07) was held in Las Vegas (June 25-28). It is an annual conference with about ten years of history, organized by Prof. Hamid Arabnia from University of Georgia.

Events that made the conference worth visiting:

• 3-hour tutorial "Data Mining in Time Series and Multimedia Databases" by Dr. Eamonn Keogh - besides my own presentation, Eamonn's tutorial was basically the main purpose of travelling from Tallinn to Las Vegas. Therefore I clearly had high hopes and I was not dissapointed - haven't seen so much "power" in a tutorial for a long time. I made myself a promise that I will look more into time series (especially dynamic time warping) after defending my PhD (hopefully some time in next May/June);
• Ad Hoc gap-fillings in DMIN conference by conference co-chair Sven F. Crone were also enjoyable, touching equally academic and industrial insights of the problems.
• Keynote "Innovation in Complex Adaptive Systems" by Professor John H. Holland from University of Michigan;
• Keynote "Challenges in Consumer Electronics for 21st Century" by Steve Leibson from Tensilica Inc - nice non-academic presentation which was really inspiring and full of interesting thoughts;
• Banquet keynote "Data Mining and Knowledge Discovery of Human Genome and Competitive Genomes, Past, Future, and the Impact on the Science and Medicine" by Dr. Jack Y. Yang from Harvard University;
  

After conference I mainly participated the DMIN social events with my colleague Toomas Kirt, people who know me can find me from about 50% of the social event pictures of DMIN.

Impressions of ICAIL 2007

ACM/AAAI 11th International Conference on Artificial Intelligence and Law (ICAIL 2007) was held in Stanford Law School, it had around 110 participants and surprisingly there where zero (!) gaps in the programm, that is - even if a participant could not make it, some substitude speaker was sent from the department. Another thing was having a single session (no parallel sessions!) so the conference had no conveyer-taste attached to it. Everybody got perfect attention and a decent discussion after a presentation.

Some highlights:

• Tutorial "Advanced Legal Technology in Practice: An Overview" by Marc Lauritsen;
• Workshop "DESI: Supporting search and sensemaking for electronically stored information in discovery proceedings"
• Dinner with an invited lecture from philosopher John Perry, who besides touching several interesting daily politics questions from untypical angles also laid out several tips for AI researchers for the next 15 years: after already "spoiling" departments of philosophy and psychology, AI researchers should next attack departments of sociology and history, forcing them to use some formalisms, objective measurements, mathematical models, relationship references etc. His main wish was to force historians to stop writing "soft science" essays with some formal frameworks to describe facts and events.*
• Workshop on Semantic Web, that couldn't escape from the classical ontology vs taxonomy discussion ;-)
  

Some thoughts/remarks I made (in general):

• Expert systems (!) should have a new breathing in public administration sector;
• NLP (natural language processing) was a popular topic - people were working with finding arguments from (natural language) text, visualizing, concept search and domain-specific (e.g. medicine) automatic tagging in the texts.
• Sentiment analysis seems promising;
• Legal ontologies are and will remain a hard topic for a long time;
• Even bigger datasets than in classical data mining conferences (1 terabyte should be processed (clustered, deduplicated and automatically tagged) in 5..8 hours);
• In my opinion all countries in EU should transform their public legal text/search systems to MetaLex. Efforts for reinventing the same wheel are pointless.
  

* I would personally suggest everyone interesting in this idea to read the following book: Mathematics in the Archaeological and Historical Sciences by F. R. Hodson, D. G. Kendall, P. Tautu - reading it at the moment myself and I can definitely recommend it.

Conferences this summer & reading

This summer has been quite busy, list of recent/upcoming conferences:

• IEEE/INFORMS SOLI 2007, Philadelphia (August 26-29)
• ICAI 2007 / DMIN 2007, Las Vegas, USA (June 25-28)
• ACM ICAIL 2007, Stanford, USA (June 4-8)

I hope to get some time to write a short summary about those conferences after returning from Philly.

Meanwhile, I will be reading a good old classic by Norbert Wiener - Human Use of Human Beings: Cybernetics and Society - which I bought few weeks ago from the street market in Haapsalu for less than 2 euros.

Beer and diapers

Who hasn't heard about the beer and diapers data mining story, which is frequently labeled as urban legend. I picked up an interesting book (Wal-Mart: The Face of Twenty-First-Century Capitalism) from Borders in San Mateo and on p.91 there's a nice related paragraph:

"Wal-Mart managers skillfully mine computerized data on sales for exploitable patterns of customer bevior. A famous example was the company's discovery that sales of beer and paper diapers rose in tandem on Fridays. By stocking the two items near each other, Wal-Mart made it easier for a parent picking up diapers on the way home from work to celebrate the weekend's arrival with a six-pack.

I can recommend this book without a doubt to anyone interested in retail industry. Regardless of what is said about it in the reviews, I enjoyed reading it - it wasn't a typical "easy reading about economy" - all the important facts/statements were backed up with references.

Welcome message

Following entries in this blog will mainly reflect my interests (but not limited to):
data mining, business intelligence, predictive analytics, knowledge discovery and management, visualization, operations research and all other sorts of intelligent(?) applications. You can blame me for being "a bit" biased towards industrial and corporate applications.