ARVAL (Association Rule's
Validation) PROJECT
SUMMARY:
- Overview
- Purposes
- Downloading
Arval
- Installation
- History
- Screenshots
Overview:
Arval
is a
datamining project and a workbench specialized in validating of
association rules extracted in very large databases. Arval acronym
come
from Association Rul'e's VALidation. Arval is 100% Java application
using the Swing API, and was developped for extracting, validating and
exporting association rule's.
The core functions of the first version are:
- Importing a association rule's mining model in PMML format,
- Importing a association rule's dataset in Weka format,
- Importing a association
rule's dataset in SAS format,
- Exporting association rule's mining model in PMML format,
- Calulating quality objectives measures.
Last Arval version is 1.0 (15/09/03).
Purposes:
Association
rule mining whithin a transaction database is a classical data mining
topic, whereby the most challenging problem is the detection of
frequently occuring patterns in the transaction sets (frequents
itemsets). A major difficulty with association rules is the prohibitive
number of frequent itemsets (and hence association rules) that results
from even a reasonably dataset.
One
important problem that the data mining expert must
confront with is the efficient
discovery of previously
unknown, valid, novel, potentially useful, and understandable pattern
in large databases. In
order to help the expert to validate the discovered knowledge, it is
important
to measure the quality of a rule. To
carry out the task, an important number
of interestgness measures
have been developed in order to filter and sort automatically a set of
rules
with respect to given goals.
Arval goals
are to calculate the various quality measures, on
specific inputted sets of rules, to evaluate the behaviour of the
quality
measures on the data sets, and to evaluate the links between indices in
order
to help the user in choosing adapted quality measures for his specific
needs.
This
project has been designed at IRIN (Institut de Recherche en
Informatique de Nantes) http://www.sciences.univ-nantes.fr/info/recherche/
as a software engineering project, in the team
"Connaissances, Informations et Données (CID)" .
Arval
's features are:
- Plug Arval with
applications KDD
software and extracting association rules,
-
Supporting The Predictive
Model Markup Language (PMML V2.0) for association rules models, and
others KDD files format,
-
Supporting the user when browsing the rule
set, items and itemset in a table,
-
Calculating many efficient
objectives quality measures.
Downloading Arval
You can download Arval application from this web site. Click on Arvalx.x.x.zip link to download it.
| FILE |
SIZE
|
CONDITIONS |
VERSION |
COMMENTS
|
 Arval1.0.0.zip
|
6352ko |
JRE
>=1.4 |
1.0.0
|
Stable
release.
|
| Arval1.1.1.zip |
|
JRE
>=1.4 |
1.1.1
|
Developpment release. Internal use only
|
Installation
Requirements:
Everything
needed to run Arval is included in the release except the Java run
time. Java
1.4 is required to get all the funtionality of Arval:
- You
need Java 1.4 or
better to run Arval application properly. Please make sure you have
installed Java 1.4 on your system before attempting to run Arval. If
you don't have a JRE, you can get it at : http://java.sun.com,
- Download
zip archive Arvalx.x.zip at our web
site (see Download section),
- Open an MS-DOS window and use the
script Run.bat or run.sh depending of your platform
and the current working directory.
History
This is a history of changes
of Arval software.
December 30th, 2003 - Version 1.0.0
Initial release
Contibutors:
the contributions that have been made by the developers listed below:
Laurent Jaffrelot
Jean-gabriel Limbourg
Fabien Trouvé
Eugen popovici
Management:
Fabrice Guillet
Screenshots
ManageData view:
Rule's view:
Measure's view:
For problems or question
regarding this web site contact
Fabrice Guillet
(fabrice POINT guillet AT univ-nantes.fr)
.
