- All Implemented Interfaces:
- Direct Known Subclasses:
- ParameterLearner, StructureLearner
public abstract class Learner
- extends java.lang.Object
- implements ModelComponent
This abstract class is the common ancestor for all specific model-learners. A learner's
main purpose is to provide the method "learnModel" which takes a list of atom expressions
(the data) and learns a knowledge base of the given type out of them. For each model type
that provides at least one learning algorithm a specific learner has to be implemented.
The task of learning a knowledge base can be divided into structure learning and
parameter learning. For both of these tasks abstract learners are available that
can be employed for implementing a specific approach.
- Matthias Thimm
|Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
public java.util.List<AtomExpression> incorporateClosedWorldAssumption(java.util.Collection<? extends AtomExpression> samples)
- This method incorporates a closed world assumption into the given set of samples. This means,
for every boolean predicate "p" of arity "n" appearing in "samples" and any n-vector
"[c1,...,cn]" of constants appearing in "samples": if "p(c1,...,cn)=true" is not in "samples"
then "p(c1,...,cn)=false" is added.
samples - a collection of atom expression.
- a set of atom expressions that consists of all atom expressions in "samples" and every
"p(c1,...,cn)=false" such that "p(c1,...,cn)=true" is not in "samples".