ARtool Frequently Asked Questions written by Laurentiu Cristofor version 1.0 Last updated on: 01/12/2002 _____________________________________________________________________ Q: What is this file? A: It's a Frequently Asked Questions for ARtool, or for short, a FAQ, but don't ask me how to pronounce that acronym because I never figured it out ;) _____________________________________________________________________ Q: What is ARtool? A: ARtool is a Java application for mining association rules in binary databases. It is distributed under the GNU General Public License and has been implemented by Laurentiu Cristofor, that is me. I implemented the algorithms in ARtool for the research I did toward my PhD Thesis at University of Massachusetts at Boston. I have designed or participated in the design of some of these algorithms, while the others have been implemented for reference or out of curiosity. ARtool comprises three components: a set of Java packages, a set of command line tools, and a graphical user interface (GUI). For more information on using ARtool see the various text documents that came with it, as well as the online help topics accessible while running the ARtool GUI. _____________________________________________________________________ Q: What is the relation between ARtool and ARMiner? Q: What is ARMiner? A: ARMiner is another data mining project that was implemented at UMass/Boston as a software engineering project. I was the project leader of the team that implemented ARMiner and I implemented the core mining classes, which now, after several revisions, are part of the laur.dm.ar package. One of ARMiner's goals was to be an educational tool that an instructor could use in a course to teach students about data mining algorithms. In this scenario students could experiment by adding their own algorithms, executing them, and seeing the results. This lead to the client-server design of ARMiner, to the group and user management feature, and to the capability of dynamically adding new algorithms to a running server. Recently, I wanted to have the ability of selecting among different algorithms for generating association rules out of frequent itemsets, the second stage of association rule mining. I also wanted the interface to display more information about the mining results. Modifying ARMiner to obtain this would have taken too much time, so instead I decided to start a new project whose goal would be to provide a single user mining application with more interface features. ARtool was the result and it only took me a week to build the interface for this new application. _____________________________________________________________________ Q: What are the differences between ARMiner and ARtool? A: The features of ARMiner that are not present in ARtool are: - the client-server design - the group and user management - the addition of algorithms at runtime - the benchmarking I designed ARtool to be a single-user application so I dropped the first two features mentioned above. Since I don't add new algorithms every day, I also dropped the dynamic addition of algorithms. Now you need to add your new algorithm to ARtool by editing the source code and recompiling the application. The benchmarking is not present in the current release of ARtool, although it might show up in a future version. I prefer to use the command line tools for comparing the behavior of the algorithms. The new features of ARtool are: - a set of command line tools that allow mining, synthetic database generation, operations on databases, etc - the GUI gives more information about a selected database - the GUI gives more information about the frequent itemsets - the GUI has a log window that keeps track of all operations performed - algorithm execution and database generation can now be interrupted - lengthy tasks are executed in threads and do not freeze the interface - an online help system - provides a quick introduction to association rule mining and to using ARtool - the GUI is easier to navigate since I use dialogs sparingly - there are plenty of tooltips to help the novice user Also, all the classes in the laur.dm.ar package have been revised. I have rewrote some of them to make the code easier to understand. If you have used the classes in ARMiner for your own projects, then you should check the javadoc sources to see what methods have disappeared, what method names have been changed, and what new methods and classes have appeared. All mining algorithms have been retrofitted to be subclasses of laur.tools.AbortableThread, so that they could be executed as threads, and their execution could be interrupted upon a user request. _____________________________________________________________________ EOF==================================================================