edu.iu.pcl.absurdist.absurdist
Class FeatureVectorConcept

java.lang.Object
  |
  +--edu.iu.pcl.absurdist.absurdist.Concept
        |
        +--edu.iu.pcl.absurdist.absurdist.FeatureVectorConcept

All Implemented Interfaces:

java.io.Serializable

public class FeatureVectorConcept

extends Concept

This is a special subclass of Concept, in which each instance contains an internal feature vector used to compute external similarity.

See Also:

Serialized Form

Nested Class Summary

static class

FeatureVectorConcept.IVPair An index-value pair; used as a component of fv.

Field Summary

java.util.Vector

fv The fv vector contained in each FeatureVectorConcept is used to determine "distance" between concepts, instead of the (d1,d2) pair in the parent Concept class.

Fields inherited from class edu.iu.pcl.absurdist.absurdist.Concept

category, d1, d2, name, rank

Constructor Summary

FeatureVectorConcept(java.lang.String name, java.lang.String category, int rank, java.util.Vector fv)

Method Summary

double

distance(Concept concept) Computes the normalized semantic distance between this concept and the specified one.

Methods inherited from class edu.iu.pcl.absurdist.absurdist.Concept

addNoise, isRanked, isValued, toString

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

fv

public java.util.Vector fv

The fv vector contained in each FeatureVectorConcept is used to determine "distance" between concepts, instead of the (d1,d2) pair in the parent Concept class. Such a vector can be used to describe, for example, the frequency of a term's occurrences in the paragraphs of a document, or in the documents of a corpus.

Constructor Detail

FeatureVectorConcept

public FeatureVectorConcept(java.lang.String name,
                            java.lang.String category,
                            int rank,
                            java.util.Vector fv)

Method Detail

distance

public double distance(Concept concept)

Computes the normalized semantic distance between this concept and the specified one. The distance is computed as

     distance(a,b) = 1 - similarity(a,b),
     

where similarity(a,b) is a Tanimoto coefficient f(a&b)/f(a|b)

Overrides:

distance in class Concept