Some time ago, I received the following email from one of my academic colleagues: I’d like to alter the query engine to use a new scheduling algorithm, and I’d like to alter the way transactional conflicts are dealt with (...). Honestly, I’m very interested in transactional models, particularly in the context of modern data warehousing. However, transactional model in ICE is very limited (there was discussion about it at http://www.infobright.org/Forums/viewthread/294/). Moreover, it is richer in IEE but even if someone purchases an IEE license, it’s impossible to alter the IEE’s codebase. It’s possible only with ICE.

This story shows that, unfortunately, not every academic project can be based on ICE/IEE. Nevertheless, I strongly encourage everyone to consider ICE as one of possible research platforms, as far as its functionality enables it. You can download complete ICE’s codebase from http://www.infobright.org/Downloads/ICE/ and you can count on the Infobright’s developers’ support on the forums whenever you consider altering it in order to obtain, e.g., better performance of some specific types of queries. Perhaps new compression algorithms? Perhaps new kinds of knowledge nodes? We will gladly help with such ideas.

Let’s use this thread for discussing academic projects.

Also, speaking of transactional models, I’d be very grateful for links to good websites about transactional models in data warehousing.

Many thanks and best greetings,

Dominik

Last week I visited Professor Hiroshi Sakai at Kyushu Institute of Technology. We’ve been meeting regularly at rough set conferences for many years. Recently, we realized that we share common interests in SQL-based analysis of non-deterministic information systems, i.e., data tables where values of rows for some of attributes may take a form of sets instead of single elements. Professor Sakai and his team showed how to reconsider various algorithms originally designed for mining standard data to deal with such non-deterministic information systems. Related mathematical foundations can be found in the following papers:

[1] H. Sakai and A. Okuma: Basic Algorithms and Tools for Rough Non-deterministic Information Analysis, Transactions on Rough Sets, Springer-Verlag, Vol. 1 (LNCS Vol. 3100), pp. 209-231, 2004.
[2] H. Sakai: Possible Equivalence Relations and Their Application to Hypothesis Generation in Non-deterministic Information Systems, Transactions on Rough Sets, Springer-Verlag, Vol. 2 (LNCS Vol. 3135), pp. 86-112, 2004.
[3] M. Nakata and H. Sakai: Lower and Upper Approximations in Data Tables Containing Possibilistic Information, Transactions on Rough Sets, Springer-Verlag, Vol. 7 (LNCS Vol. 4400), pp. 170-189, 2007.
[4] H. Sakai, R. Ishibasi and M. Nakata: Rules and Apriori Algorithm in Non-deterministic Information Systems, Transactions on Rough Sets, Springer-Verlag, Vol. 9 (LNCS Vol. 5390), pp. 328-350, 2008.

On the other hand, SQL-based mining of large data may be a good application of ICE. When redesigning classical data mining algorithms to let them work with SQL instead of direct data access, one can expect highly complex analytical query workload (see e.g. our data mining forum). SQL statements generated automatically during execution of data mining algorithms will be even more complex in the case of non-deterministic information systems. Investigation of such algorithms may lead to new data mining applications. It may also enable us to test ICE within a completely new framework.

Taking it all into account, we decided to proceed with closer cooperation aimed at discovery and optimization of new methods of SQL-based analysis of non-deterministic information systems. We will keep everyone informed about intermediate results and articles published. We will be also happy to hear from other researchers with analogous interests.

Hello everyone,

I am Hiroshi Sakai at Kyushu Institute of Technology. I would thank the founder of Infobright.org, Dominik Slezak, for showing our papers in this thread. I and Professor Michinori Nakata are interested in non-deterministic information and incomplete information, and we have proven some theoretical results. We have tried to extend rough sets based concepts in a typical data table to new concepts in a data table with non-deterministic information.  This may be seen as an extension of rough sets.

I am truly grateful to Dominik for proceeding with closer cooperation on SQL-based analysis of non-deterministic information systems. I am very excited for our new cooperation, the realization of our framework over ICE and our next fruitful work.

This thread seems too general or wide for the details of non-deterministic information systems, I think.  As for the background of non-deterministic information systems, I will sometimes post small surveys to another proper thread, like a thread 228: data mining forum. Dominik, do you generate a thread on non-deterministic information systems?
Thank you.

Yours Sincerely,
Hiroshi Sakai

Dear Hiroshi,

Many thanks for following up!

I’m truly excited about our cooperation.

I also thought about how to organize it on forums. Let me suggest using the following thread for details:

http://www.infobright.org/Forums/viewthread/621/

Please let me know whether it’s a good idea.

Best greetings,

Dominik

Dear Dominik and everyone,

I know the tread http://www.infobright.org/Forums/viewthread/621/.

In this thread, various types of inexact data will be discussed, therefore we may add a subtopic like

“Subtopic: Non-deterministic information”  and “Subtopic: Interval information”

as the first line. In any way, I move to the new thread 621. Thank you.

Regards,
Hiroshi Sakai

Dear Hiroshi,

This is a very good idea! Actually, I’ll mention about it in thread 621.

Best greetings,

Dominik

nice discussion and thanx for the posting the information
Data Entry India