net.obsearch.query
Class OBQueryLong<O extends OBLong>

java.lang.Object
  net.obsearch.query.AbstractOBQuery<O>
      net.obsearch.query.OBQueryLong<O>

public final class OBQueryLong<O extends OBLong>
extends net.obsearch.query.AbstractOBQuery<O>

Object used to store a query request.

Since:

0.7

Author:

Arnoldo Jose Muller Molina

Field Summary

protected long[]	max Maximum part of the rectangle of the query.
protected long[]	min Minimum part of the rectangle of the query.
protected OBPriorityQueueLong<O>	result Holds the results for the query.
protected long[]	smap SMAPed vector

Constructor Summary

OBQueryLong()
Constructor.

OBQueryLong(O object, long range, OBPriorityQueueLong<O> result)
Creates a new OBQueryLong object.

OBQueryLong(O object, long range, OBPriorityQueueLong<O> result, long[] smap)
Creates a new OBQueryLong object.

OBQueryLong(O object, OBPriorityQueueLong<O> result)

Method Summary

boolean	add(long id, O object)
boolean	add(long id, O obj, long d) Add the given object, object id and distance of type float to the queue.
double	approx(long[] dbin) Calculate the 1 + E or (c for Adonis et al) for approx.
boolean	collides(long[][] rectangle) Returns true if the given rectangle collides with this query.
double	compound(long[] dbin) Calculates ep without multiplying by 1/k and
double	ep(List<net.obsearch.AbstractOBResult<O>> dbin)
double	ep(long[] dbin) Calculate the EP value for a sorted list of distances.
long	getDistance()
long[]	getHigh() Return low of the query rectangle.
long[]	getLow() Return low of the query rectangle.
O	getObject()
OBPriorityQueueLong<O>	getResult()
List<net.obsearch.AbstractOBResult<O>>	getSortedElements()
boolean	isCandidate(long smapDistance) Returns true if we should calculate the real distance.
boolean	isFull()
long	peek() peek to the largest value if the queue is full.
double	precision(long[] dbin) Calculates the precision
double	rde(long[] dbin)
double	recall(List<net.obsearch.AbstractOBResult<O>> perfectQuery)
void	setResult(OBPriorityQueueLong<O> result) Set the results of the matching to a new object.
double	tDR(long[] dbin) Total distance ratio
boolean	updatedRange(long originalRange) Returns true if the originalRange has been modified.

Methods inherited from class java.lang.Object

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

Field Detail

result

protected OBPriorityQueueLong<O extends OBLong> result

Holds the results for the query.

min

protected long[] min

Minimum part of the rectangle of the query.

max

protected long[] max

Maximum part of the rectangle of the query.

smap

protected long[] smap

SMAPed vector

Constructor Detail

OBQueryLong

public OBQueryLong()

Constructor.

OBQueryLong

public OBQueryLong(O object,
                   long range,
                   OBPriorityQueueLong<O> result)

Creates a new OBQueryLong object.

Parameters:

object - The object that will be matched.

range - The range to be used for the match.

result - The priority queue were the results will be stored.

OBQueryLong

public OBQueryLong(O object,
                   OBPriorityQueueLong<O> result)

OBQueryLong

public OBQueryLong(O object,
                   long range,
                   OBPriorityQueueLong<O> result,
                   long[] smap)

Creates a new OBQueryLong object.

Parameters:

object - The object that will be matched.

range - The range to be used for the match.

result - The priority queue were the results will be stored.

smap - SMAP vector representation of the given object.

Method Detail

getObject

public O getObject()

collides

public boolean collides(long[][] rectangle)

Returns true if the given rectangle collides with this query.

Parameters:

rectangle - The rectangle to search.

getLow

public long[] getLow()

Return low of the query rectangle.

getHigh

public long[] getHigh()

Return low of the query rectangle.

getResult

public OBPriorityQueueLong<O> getResult()

Returns:

The current results of the matching.

setResult

public void setResult(OBPriorityQueueLong<O> result)

Set the results of the matching to a new object.

Parameters:

result - The new result.

isCandidate

public boolean isCandidate(long smapDistance)

Returns true if we should calculate the real distance.

Parameters:

smapDistance - The lower-resolution distance calculated with SMAP.

add

public boolean add(long id,
                   O object)
            throws InstantiationException,
                   IllegalAccessException,
                   net.obsearch.exception.OBException

Specified by:

add in class net.obsearch.query.AbstractOBQuery<O extends OBLong>

Throws:

InstantiationException

IllegalAccessException

net.obsearch.exception.OBException

getDistance

public long getDistance()

getSortedElements

public List<net.obsearch.AbstractOBResult<O>> getSortedElements()

Specified by:

getSortedElements in class net.obsearch.query.AbstractOBQuery<O extends OBLong>

add

public boolean add(long id,
                   O obj,
                   long d)
            throws InstantiationException,
                   IllegalAccessException

Add the given object, object id and distance of type float to the queue. Updates the range of the query as needed if the range shrinks after this insertion.

Parameters:

id - The id of the object to be used

obj - The object to be added

d - The distance to be added

Throws:

IllegalAccessException - If there is a problem when instantiating objects O

InstantiationException - If there is a problem when instantiating objects O

updatedRange

public boolean updatedRange(long originalRange)

Returns true if the originalRange has been modified.

Returns:

true If the current range (getDistance()) is different than originalRange.

isFull

public boolean isFull()

Specified by:

isFull in class net.obsearch.query.AbstractOBQuery<O extends OBLong>

Returns:

true if the underlying priority queue's size is equal to k

recall

public double recall(List<net.obsearch.AbstractOBResult<O>> perfectQuery)

Specified by:

recall in class net.obsearch.query.AbstractOBQuery<O extends OBLong>

ep

public double ep(List<net.obsearch.AbstractOBResult<O>> dbin)

Specified by:

ep in class net.obsearch.query.AbstractOBQuery<O extends OBLong>

ep

public double ep(long[] dbin)

Calculate the EP value for a sorted list of distances.

approx

public double approx(long[] dbin)
              throws net.obsearch.exception.OBException

Calculate the 1 + E or (c for Adonis et al) for approx. nearest neighbor This is the approximation and "real" is the real result.

Parameters:

q -

Returns:

Throws:

RAException

net.obsearch.exception.OBException

compound

public double compound(long[] dbin)

Calculates ep without multiplying by 1/k and

rde

public double rde(long[] dbin)

tDR

public double tDR(long[] dbin)

Total distance ratio

precision

public double precision(long[] dbin)

Calculates the precision

peek

public long peek()
          throws net.obsearch.exception.OBException

peek to the largest value if the queue is full.

Throws:

net.obsearch.exception.OBException