|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||||||
 |
|
| © Simon Colton 2002 Originally developed at the Universities
of Edinburgh
and York. in
The Computational Bioinformatics Group
|
|
|
|
|
The HR system performs automated theory formation. Given background information about a domain, HR invents concepts, calculates examples, makes hypotheses and seeks explanations of the hypotheses. The introduction page contains information about the past, present and future of the project, including the aims of the project, the people working on it and the implementation of HR. |
The theory of automated theory formation upon which HR is implemented has four modules: (i) data generation (ii) concept generation (iii) hypothesis generation and (iv) explanation generation. The theory page contains information about the production rules, measures of interestingness, search strategies and theory formation cycles that make up the seperate processes in the modules. |
|
|
|
HR is a machine learning program which can be applied to the tasks of concept identification, prediction, scientific discovery and puzzle generation. The applications have so far been to discovery tasks in mathematics. The applications page contains details of applications to number theory, algebra, puzzle generation, constraint satisfaction and setting mathematical exercises. |
HR can be used as a Java library for theory formation. The manual page contains the API of this library. It also points to technical notes about the workings of HR, and has tutorials for understanding HR and a screen-by-screen manual. |
|
|
|
There are several large projects under way with HR, and some which have been completed. The projects page contains details about these projects, including their aims, who is working on them, how they are funded and results to date. |
The latest release of HR is available from the download page. Also available for download are papers about HR. |
|
|
|