ChangeLog

Build 2010–08–07

• support for OWL API 3
• ORE tool based on DL-Learner algorithms (soon to be migrated to an own project)
• implemented several new heuristics, e.g. generalised F-Measure
• stochastic approximation of computing F-Measure
• learning algorithms for the EL description logic
• support for hasValue construct in combination with string datatype
• support for refining existing definitions (instead of learning from scratch) for CELOE ontology engineering algorithm
• increased number of unit tests (now 40)
• support for direct Pellet 2 integration and reasoners connected via OWLLink
• 24 bugs fixed and 12 feature requests implemented at sourceforge.net bug tracker

Build 2009–05–06

• new algorithm: CELOE (class expression learning for ontology engineering)
• Protégé Plugin based on CELOE
• wrote a manual for DL-Learner
• an efficient refinement operator for the EL description logic
• fast stochastic class expression coverage estimation included
• reasoner component design and learning problem structure improved
• more learning examples provided and unit tests for ensuring code quality extended
• 6 bugs and feature requests reported at the sourceforge.net tracker fixed

Build 2008–10–13

• improved refinement operator based learning approach taking domain/range of properties, property hierarchies, disjoint classes into account to structure search space more efficiently
• DL-Learner GUI for loading, saving, and modifying configuration files
• fast instance checking algorithm reduces the time to test example coverage of class descriptions significantly
• Carcinogenesis Benchmark
• extraction component: more flexible structure, SPARQL results are converted to OWL on the fly, correct blank node handling Poster Abstract
• more learning examples provided in release
• 12 bugs and 10 feature requests reported at the sourceforge.net tracker fixed

Build 2008–02–18

• Flexible new component based structure:
  • 4 types of components: knowledge sources, reasoners, learning problems, learning algorithms
  • easily extensible: to implement a new component of one of the above types you only have to extend the corresponding class in org.dllearner.core and add the name of your class to the components.ini file
  • each component can maintain and easily extend its own configuration options
• Support for using SPARQL endpoints as background knowledge, including mechanisms for knowledge fragment selection. This feature enables DL-Learner to use DBpedia as background knowledge.
• Preliminary support for learning from only positive examples and learning of inclusion axioms instead of definitions.
• Support for N-Triple files.
• Support for using role hierarchies in the refinement operator based algorithm.
• Much more powerful web service interface allowing to access and modify all DL-Learner components.
• Reasoners:
  • preliminary OWL API reasoner interface support: Pellet, FaCT++
  • KAON2 dropped, such that DL-Learner now depends solely on open source libraries
• A Prolog parser, which can help in converting Prolog files to OWL (thereby transfering ILP problems into OWL learning problems).
• More examples added:
  • complete Moral Reasoner Benchmarks
  • more SPARQL benchmarks
  • all examples now also available in OWL

Build 2007–08–31

Initial release.