com.sun.labs.minion.classification
Class QueryZone

java.lang.Object
  com.sun.labs.minion.classification.QueryZone

public class QueryZone
extends java.lang.Object

A query zone is a set of documents that are centered around a set of feature clusters. The features in the clusters are organized into a big query, and results are returned in score order. The results will contain a mix of documents in and out of the training set.

Field Summary

protected java.lang.String	fromField The field we're building classifiers from.
protected java.util.PriorityQueue<com.sun.labs.minion.classification.QueryZone.HE>	h The contents of the query zone
protected java.util.List<com.sun.labs.minion.classification.QueryZone.HE>	heapElements A list of all the heap elements
protected PartitionManager	manager The partition manager for the documents in the collection
protected java.util.Map<DiskPartition,TermCache>	termCaches A term cache to use when building classifiers.
protected ResultSetImpl	training The training set for the class under construction
protected WeightingComponents	wc The weighting components to use
protected WeightingFunction	wf The weighting function to use
protected FeatureClusterSet	wfClusters The features that we will use for our model.

Constructor Summary

QueryZone(ResultSetImpl training, java.lang.String fromField, FeatureClusterSet featureClusters, java.util.Map<java.lang.String,TermStatsImpl> termStats, java.util.Map<DiskPartition,TermCache> termCaches, PartitionManager manager)

Method Summary

java.util.PriorityQueue<com.sun.labs.minion.classification.QueryZone.HE>	computeZone()
java.util.PriorityQueue<com.sun.labs.minion.classification.QueryZone.HE>	getHeap() Gets a heap that represents the members of the query zone.
FeatureClusterSet	getWeightedClusters() Returns the feature clusters used to determine this query zone with collection level weights assigned to them.

Methods inherited from class java.lang.Object

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

Field Detail

fromField

protected java.lang.String fromField

The field we're building classifiers from.

termCaches

protected java.util.Map<DiskPartition,TermCache> termCaches

A term cache to use when building classifiers.

wfClusters

protected FeatureClusterSet wfClusters

The features that we will use for our model.

manager

protected PartitionManager manager

The partition manager for the documents in the collection

training

protected ResultSetImpl training

The training set for the class under construction

wf

protected WeightingFunction wf

The weighting function to use

wc

protected WeightingComponents wc

The weighting components to use

h

protected java.util.PriorityQueue<com.sun.labs.minion.classification.QueryZone.HE> h

The contents of the query zone

heapElements

protected java.util.List<com.sun.labs.minion.classification.QueryZone.HE> heapElements

A list of all the heap elements

Constructor Detail

QueryZone

public QueryZone(ResultSetImpl training,
                 java.lang.String fromField,
                 FeatureClusterSet featureClusters,
                 java.util.Map<java.lang.String,TermStatsImpl> termStats,
                 java.util.Map<DiskPartition,TermCache> termCaches,
                 PartitionManager manager)

Method Detail

computeZone

public java.util.PriorityQueue<com.sun.labs.minion.classification.QueryZone.HE> computeZone()

getHeap

public java.util.PriorityQueue<com.sun.labs.minion.classification.QueryZone.HE> getHeap()

Gets a heap that represents the members of the query zone. Each element of the heap contains a big query and an array group that represent all the results from a partition. The heap elements provide method to facilitate access to the current/top document from the partition that the element represents.

Returns:

the heap of all documents in the query zone

getWeightedClusters

public FeatureClusterSet getWeightedClusters()

Returns the feature clusters used to determine this query zone with collection level weights assigned to them. Note that once computeZone is called, the feature clusters returned will be normalized. Before computeZone is called they are not. A copy of the feature clusters should be made if the clusters will be modified.

Returns:

the internal set of feature clusters