||-ROSETTA is a continuation of ROSETTA, a classification program developed in 1997 by Aleksander Öhrn. It features a modular framework for algorithms and datastructures intended for classification and prediction. Over the years the pipeline has been extended with new algorithms and methods. ||-ROSETTA contains all the methods and also allows for parallel execution, taking advantage of all the advances made in Computer Science over the last 20 years. ||-ROSETTA is open to anyone that wishes to add further methods or algorithms to the framework. Together we can accomplish more. In short, ||-ROSETTA offers multiple algorithms for handling:
- Completion
- Discretization
- Reduct computation
- Rule Generation
- Classification
- And many more
If you want transparent classification and prediction, ||-ROSETTA is for you. Unlike current Machine Learning trends, Rough Set classification is completely open to interpretation and process tracing. The classification of every object can be traced back to give the complete reasoning, and the classification foundation can also be exported from one classifier and used within another.
Alongside the source code you will find a perl script that offers a simple way to run ||-ROSETTA. Simply give the name of the dataset and it will be classified using the default settings. Going into the script you will a command list for algorithms that you can change around according to your needs. For all the commands available, see the manual.