al-che-my : A medieval chemical philosophy having as its asserted aims the transmutation of base metals into gold, the discovery of the panacea, and the preparation of the elixir of longevity.
Alkemy is a decision-tree learning system designed to learn comprehensible theories from structured data. The logical foundations for Alkemy is provided in the book `Logic for Learning' by John Lloyd. Algorithmic and learning-theoretic aspects of the system are discussed in the thesis Learning Comprehensible Theories from Structured Data.

Alkemy is implemented in Noweb-C++, with fairly extensive documentation. It is being actively supported. The current version is Release 1.3. (Escher is now implemented.) The program and its source code is protected by copyright, but is available for research and educational purposes. If you wish to use this code in any other way, please contact me for more information.

Kindly email a short note to let me know you have downloaded the system. Also let me know if you wish to be added to an Alkemy mailing list. The list will inform people of updates to the system. Obviously, I am also interested in any feedback and bug reports for Alkemy.

An older (but more efficient) version that does not make use of Escher for computation is available here.

Last Updated: 29 September, 2009
Kee Siong Ng:
Copyright 2009, All Rights Reserved, Kee Siong Ng, The Australian National University