This repository provides the ability to transform the data provided by the qual-o-mat-data repository into scaled contexts for use with the conexp-clj tool.

The qual-o-mat-data repository contains archived data on the questions from the Wahl-O-Mat election quiz.

conexp-clj provides a variety of algorithms and methods for Formal Concept Analysis. The fundamental data structure underlying Formal Concept Analysis is the formal context. This repository provides the means to transform the set of quiz questions and answers of each election in the qual-o-mat-data dataset into such a formal context.

To make use of this repository, you need to execute setup.sh by running:

bash setup.sh

This will download the qual-o-mat-data repository, and create all required files. To create the context files from the dataset, run:

bash update.sh

Since the Wahl-O-Mat permits "agree", "disagree" and "neutral" ()as answers, these contexts need to be created from many valued contexts through scaling.

The two scalings offered by this repository are the nominal scale, and an ordinal scale:

Nominal Scale:

                  |Stimme zu   Stimme nicht zu  Neutral 
------------------+-------------------------------------
Stimme zu         |x                .              .   
Stimme nicht zu   |.                x              .   
Neutral           |.                .              x  

Ordinal Scale:

                  |Stimme zu   Stimme nicht zu  Neutral 
------------------+-------------------------------------
Stimme zu         |x                x              x   
Stimme nicht zu   |.                x              .   
Neutral           |.                x              x  

The ordinal scale can be interpreted as a ranking of "disagree -> neutral -> agree".

The script saves the contexts in the Burmeister format. conexp-clj can then be used to perform further operations on them.

This project was created for the Knowledge & Data Engineering Group (KDE), University of Kassel.