SlattFrontPage

Introduction

Slatt is a project on which some researchers will implement a number of algorithmic ideas regarding closure-based association mining. The purpose, as of now, is research: empirical investigations complementing the mathematical advances that could be put forward for the task.

Downloads

Most recent version is Slatt 0.2.2 alpha, which includes confidence boost bounds for both representative rules and the B** basis. **

Early: Public download offered nov 21, 2009, Slatt 0.2.0 alpha, first public download ever available - although sources can be checked out or browsed without limit. Followed shortly by a bugfix version, Slatt 0.2.1 alpha. Further, there remains a tiny trivial bug in rerulattice.py, corrected in the sources since dec 19th but which did not warrant a new download. Get that file by browsing through the trunk sources if you wish.

Details

As of the creation of this page, the code was able to find association rules but through a somewhat chaotic structure. There were a number of desirable variants: what sort of rules are output among various sets or bases, what info is necessary for it, maybe just closures, maybe minimal generators; how to output rules and compare the output rule sets; and maybe we can also support a simulator of the Angluin-Frazier-Pitt query learner for implications. This is likely to lead to several branches, and the project creator has decided to open a project in Google Code for this.

The project is slowly developing since. See the page AboutTheTrunk.

Add your content here. Format your content with:

• Text in bold or italic
• Headings, paragraphs, and lists
• Automatic links to other wiki pages