Package ciir.umass.edu.learning

Class SparseDataPoint

java.lang.Object

ciir.umass.edu.learning.DataPoint

ciir.umass.edu.learning.SparseDataPoint

public class SparseDataPoint
extends DataPoint

Implements a sparse data point using a compressed sparse row data structure

Author:

Siddhartha Bagaria

Field Summary

Fields inherited from class ciir.umass.edu.learning.DataPoint

cached, description, FEATURE_INCREASE, fVals, id, knownFeatures, label, MAX_FEATURE, missingZero, UNKNOWN

Constructor Summary

Constructors

Constructor	Description
SparseDataPoint(float[] fVals, int[] fIds, java.lang.String queryId, float relevanceLabel)	Allows the direct allocation of a SparseDataPoint instead of having it parse text.
SparseDataPoint(SparseDataPoint dp)
SparseDataPoint(java.lang.String text)

Method Summary

Modifier and Type	Method	Description
float	getFeatureValue(int fid)	Get the value of the feature with the given feature ID
float[]	getFeatureVector()	Gets the value of all features as a dense array of feature values.
boolean	hasFeature(int fid)
void	setFeatureValue(int fid, float fval)	Set the value of the feature with the given feature ID
void	setFeatureVector(float[] dfVals)	Sets the value of all features with the provided dense array of feature values

Methods inherited from class ciir.umass.edu.learning.DataPoint

getCached, getDescription, getID, getKey, getLabel, getNumberOfKnownFeatures, getValue, isUnknown, parse, resetCached, setCached, setDescription, setID, setLabel, toString

Methods inherited from class java.lang.Object

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

Constructor Detail

SparseDataPoint

public SparseDataPoint(java.lang.String text)

SparseDataPoint

public SparseDataPoint(float[] fVals,
                       int[] fIds,
                       java.lang.String queryId,
                       float relevanceLabel)

Allows the direct allocation of a SparseDataPoint instead of having it parse text.

The fVals array represents the sparse feature values. Only non-null values must be specified.

The fIds array has equal length to fVals and specifies for each position the feature ID that this position in both array stands for. Thus, if fIds[1] == '5' then fVals[5] is the value of this data point for feature number 5. The values in fIds must be sorted ascendingly.

Parameters:

fVals - The feature values, starting at index 0.

fIds - The feature IDs, starting at index 0.

queryId - The ID of the query this datapoint is associated with.

relevanceLabel - The relevance label of this datapoint with respect to the query.

SparseDataPoint

public SparseDataPoint(SparseDataPoint dp)

Method Detail

hasFeature

public boolean hasFeature(int fid)

getFeatureValue

public float getFeatureValue(int fid)

Description copied from class: DataPoint

Get the value of the feature with the given feature ID

Specified by:

getFeatureValue in class DataPoint

Returns:

setFeatureValue

public void setFeatureValue(int fid,
                            float fval)

Description copied from class: DataPoint

Set the value of the feature with the given feature ID

Specified by:

setFeatureValue in class DataPoint

getFeatureVector

public float[] getFeatureVector()

Description copied from class: DataPoint

Gets the value of all features as a dense array of feature values.

Specified by:

getFeatureVector in class DataPoint

setFeatureVector

public void setFeatureVector(float[] dfVals)

Description copied from class: DataPoint

Sets the value of all features with the provided dense array of feature values

Specified by:

setFeatureVector in class DataPoint