java.lang.Object
de.uni_mannheim.informatik.dws.melt.matching_jena_matchers.metalevel.TrainingAlignmentGenerator
All Implemented Interfaces:
IMatcher<org.apache.jena.ontology.OntModel,​Alignment,​Properties>, IMatcherCaller

public class TrainingAlignmentGenerator
extends Object
implements IMatcherCaller, IMatcher<org.apache.jena.ontology.OntModel,​Alignment,​Properties>
Deprecated.
replacement is to use AddNegativesViaMatcher which has the same functionality (but it really uses only correspondences with equivalence relation as positive correspondences) .
This matcher assumes that the input alignment is a kind of reference alignment. After applying the recallMatcher given in the constructor, a new alignment is returned which contains positive (equivalence relation) and negative(incompat relation) correspondences. With the help of this alignment, supervised matchers can be trained.
  • Field Details

    • recallMatcher

      private final Object recallMatcher
      Deprecated.
  • Constructor Details

    • TrainingAlignmentGenerator

      public TrainingAlignmentGenerator​(Object recallMatcher)
      Deprecated.
  • Method Details

    • match

      public AlignmentAndParameters match​(Set<Object> sourceRepresentations, Set<Object> targetRepresentations, Object inputAlignment, Object parameters) throws Exception
      Deprecated.
      Description copied from interface: IMatcherCaller
      Aligns two ontologies/knowledge graphs given as the first and second parameter. An additional input alignment can be given as well as parameters which further define how and what to match. In case the matcher needs the objects in a specific type, call
       
       TODO
       
       
      Specified by:
      match in interface IMatcherCaller
      Parameters:
      sourceRepresentations - a set of objects which all represents the source ontology/knowledge graph
      targetRepresentations - a set of objects which all represents the target ontology/knowledge graph
      inputAlignment - this object represents the input alignment
      parameters - object representing additional parameters. Sensible classes are Properties, Map<String, Object> or any similar data structure. Some already specified keys (strings) can be found at ParameterConfigKeys.
      Returns:
      the resulting alignment and updated parameters of the matching process.
      Throws:
      Exception - any exception which occurs during matching
    • match

      public Alignment match​(org.apache.jena.ontology.OntModel source, org.apache.jena.ontology.OntModel target, Alignment inputAlignment, Properties parameters) throws Exception
      Deprecated.
      Description copied from interface: IMatcher
      Aligns two ontologies/knowledge graphs given as the first and second parameter. An additional input alignment can be given as well as parameters which further define how and what to match. In case inputAlignment or parameters is not used, making them to the Object class, will required no additional transformations of these objects (no overhead).
      Specified by:
      match in interface IMatcher<org.apache.jena.ontology.OntModel,​Alignment,​Properties>
      Parameters:
      source - this object represents the source ontology/knowledge graph
      target - this object represents the target ontology/knowledge graph
      inputAlignment - this object represents the input alignment.
      parameters - object representing additional parameters. Only add to this object and do not create a new Object like parameters= new ...() because otherwise the parameters are lost (java ist call by value). Sensible classes are Properties (preferred), Map<String, Object> or any similar data structure. Some already specified keys (strings) can be found at ParameterConfigKeys.
      Returns:
      the resulting alignment of the matching process.
      Throws:
      Exception - any exception which occurs during matching
    • getTrainingAlignment

      public static Alignment getTrainingAlignment​(Alignment recallAlignment, Alignment referenceAlignment)
      Deprecated.
      This method returns a training alignment based on a recall alignment and a reference alignment. The training alignment is generated by getting all correspondences in the recall alignment where at least one part is also contained in the reference alignment. If the correspondences is directly in the reference alignment, then it is assumed to be a positive example. If only one part of the correspondence is in the reference alignment, then is is assumed to be a negative correspondence. The result contains the correspondences from the recall alignment, where positive examples have the equivalence relation and negative examples have INCOMPAT relation.
      Parameters:
      recallAlignment - recall alignment
      referenceAlignment - reference alignment which does not need to really be the reference alignment of a track.
      Returns:
      the correspondences from the recall alignment, where positive examples have the equivalence relation and negative examples have INCOMPAT relation