ILP is a set of techniques to perform Machine Learning using a high level, relational representation language: Datalog (function free Horn clauses).

More references on the theory behind ILP can be found here:
http://en.wikipedia.org/wiki/Inductive_logic_programming

Planned features:
<ul>
<li>Top-down and bottom-up refinement operators;</li>
<li>[long term] Meta-heuristic search strategies: memetic algorithm combining Genetic Logic Programming and local refinement operators;</li>
<li>CSP-based subsumption tests (cf <a href="http://www.lri.fr/~malobert/Django">Django</a> by Maloberti and Sebag);</li>
<li>SQL DB or datalog text files as background knowledge datasources;
<li>Use of <a href="http://mrdm.dantec.nl/mrdm/siebesknobbe.pdf">UML/CDBL (Common Declarative Bias Language)</a> to specify the bias;</li>
<li>[long term] parallelization using the "Island Model" for distributed evolutionary computation.</li>
<li>[very long term] add probabilistic reasoning capabilities to the representation language (such as BLP, SLP or other Probabilistic-Logical Models).</li>
</ul>
LibILP is Free Software and is distributed under the terms of the GNU GPL license.

Registration Date: Sat 10 Jul 2004 03:28:23 AM UTC
License: GNU General Public License V2 or later
Development Status: 1 - Planning