com.sun.labs.minion.classification
Class Rocchio.FQR

java.lang.Object
  com.sun.labs.minion.classification.Rocchio.FQR

Enclosing class:

Rocchio

protected class Rocchio.FQR
extends java.lang.Object

A class to collate and hold the results of a feedback query.

Field Summary

protected float[]	f1pts The f1 measure at our precision points.
protected float	maxF1 The maximum f1 encountered.
protected float	maxF1Sim The similarity associated with the maximum f1 measure.
protected int	nonrel The number of non-relevant docs encountered.
protected float	p The last calculated precision.
protected int	pos The current position in our arrays.
protected float[]	ppts 11 precision points.
protected float	r The last calculated recall.
protected int	R The total number of relevant docs.
protected float	rd The delta for the recall levels
protected int	rel The number of relevant docs encountered.
protected int[]	rl The number of retrieved documents at which we'll measure our levels.
protected float[]	sim The similarities corresponding to our precision points.

Constructor Summary

Rocchio.FQR(int R)

Method Summary

void	addNonRel(float s) Adds a new relevant document.
void	addRel(float s) Adds a new relevant document.
float	avgP() Gets the average precision.
boolean	done() Returns true if we've hit recall of 1.0.
float	f1()
void	finish() Fills out the last entry in our precision points.
float	getThreshold() Gets the threshold that one would want to use for this feedback query.
float	precision()
float	recall()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

rl

protected int[] rl

The number of retrieved documents at which we'll measure our levels.

ppts

protected float[] ppts

11 precision points.

f1pts

protected float[] f1pts

The f1 measure at our precision points.

sim

protected float[] sim

The similarities corresponding to our precision points.

rd

protected float rd

The delta for the recall levels

p

protected float p

The last calculated precision.

r

protected float r

The last calculated recall.

pos

protected int pos

The current position in our arrays.

maxF1

protected float maxF1

The maximum f1 encountered.

maxF1Sim

protected float maxF1Sim

The similarity associated with the maximum f1 measure.

rel

protected int rel

The number of relevant docs encountered.

nonrel

protected int nonrel

The number of non-relevant docs encountered.

R

protected int R

The total number of relevant docs.

Constructor Detail

Rocchio.FQR

public Rocchio.FQR(int R)

Method Detail

addRel

public void addRel(float s)

Adds a new relevant document. This method also keeps track of the precision at each of our

Parameters:

s - the similarity associated with the document.

addNonRel

public void addNonRel(float s)

Adds a new relevant document.

Parameters:

s - the similarity associated with the document.

finish

public void finish()

Fills out the last entry in our precision points.

avgP

public float avgP()

Gets the average precision.

getThreshold

public float getThreshold()

Gets the threshold that one would want to use for this feedback query.

done

public boolean done()

Returns true if we've hit recall of 1.0.

recall

public float recall()

precision

public float precision()

f1

public float f1()