net.siefkes.nlstego.predict
Class PPMFetchStats

java.lang.Object
  net.siefkes.nlstego.predict.FetchStats
      net.siefkes.nlstego.predict.PPMFetchStats

All Implemented Interfaces:

Serializable

public final class PPMFetchStats
extends FetchStats

Implementation of the FetchStats class using a Prediction-by-Partial-Match strategy. This class collects statistics about the item access history that can be used for prediction.

Created (04.10.2001 11:16:19).

Version:

$Revision: 1.14 $, $Date: 2005/07/28 15:03:09 $, $Author: siefkes $

Author:

Christian Siefkes

See Also:

Serialized Form

Field Summary

Fields inherited from class net.siefkes.nlstego.predict.FetchStats

END_OF_SESSION, START_OF_SESSION

Constructor Summary

PPMFetchStats(Configuration conf, StatelessFilter filter)
Creates a new instance.

Method Summary

void	add(String key, int sessionID, boolean train) Adds a request of an item.
int	getBlendingNum() A standard get*().
double	getDecFactor() A standard get*().
int	getInnerCacheSize() A standard get*().
int	getLargestOuterCacheSize() A standard get*().
int	getLengthOfHistory() A standard get*().
int	getMinAccesses() A standard get*().
SortedMap	predict(int number, int sessionID) Predicts the items that will probably be requested next.
String	printModelStats() Prints statistical information about this model.
void	sessionEnded(int sessionID) Marks a session as ended.
String	toString() Returns a String that represents the value of this object.

Methods inherited from class net.siefkes.nlstego.predict.FetchStats

add, getID, getIDInteger, getStatelessFilter, mapKey, shutdown, unmapKey

Methods inherited from class java.lang.Object

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

Constructor Detail

PPMFetchStats

public PPMFetchStats(Configuration conf,
                     StatelessFilter filter)

Creates a new instance.

Created (11.09.2001 16:35:54) by Christian Siefkes.

Parameters:

conf - the Configuration to use

filter - optional filter that decides which predictions should be stored in the stateless (global) model; if null, all predictions are stored

Method Detail

add

public void add(String key,
                int sessionID,
                boolean train)

Adds a request of an item.

Created (27.09.2001 12:16:43) by Christian Siefkes.

Specified by:

add in class FetchStats

Parameters:

key - unique ID of the item currently requested fetching the last document in this session

sessionID - the ID of the current session (must be unique among all sessions parallel in progress)

train - true iff the model should be trained, i.e. this request should be incorporated into the prediction model; otherwise only the internal state will be changed so the next call to predict(int, int) will find suitable predictions (this happens in any case)

getBlendingNum

public int getBlendingNum()

A standard get*().

Created (01.10.2001 11:35:12) by Christian Siefkes.

Returns:

the value of the attribute

getDecFactor

public double getDecFactor()

A standard get*().

Created (01.10.2001 11:16:32) by Christian Siefkes.

Returns:

the value of the attribute

getInnerCacheSize

public int getInnerCacheSize()

A standard get*().

Created (01.10.2001 11:16:23) by Christian Siefkes.

Returns:

the value of the attribute

getLargestOuterCacheSize

public int getLargestOuterCacheSize()

A standard get*().

Created (01.10.2001 11:16:13) by Christian Siefkes.

Returns:

the value of the attribute

getLengthOfHistory

public int getLengthOfHistory()

A standard get*().

Created (26.09.2001 17:52:52) by Christian Siefkes.

Returns:

the value of the attribute

getMinAccesses

public int getMinAccesses()

A standard get*().

Created (01.10.2001 11:34:51) by Christian Siefkes.

Returns:

the value of the attribute

predict

public SortedMap predict(int number,
                         int sessionID)

Predicts the items that will probably be requested next.

Created (27.09.2001 12:16:43) by Christian Siefkes.

Specified by:

predict in class FetchStats

Parameters:

number - the number of results to return, >0

sessionID - the ID of the current session (must be unique among all sessions parallel in progress)

Returns:

a SortedMap containing mapped keys as keys and FetchStatsResult's as values. The length of the map is max (number, all matching documents that had at least minAccesses)

printModelStats

public String printModelStats()

Prints statistical information about this model. This method might take some time while the data is gathered.

Specified by:

printModelStats in class FetchStats

Returns:

a string listing statistical details about this model

sessionEnded

public void sessionEnded(int sessionID)

Marks a session as ended.

Created (27.09.2001 12:16:43) by Christian Siefkes.

Specified by:

sessionEnded in class FetchStats

Parameters:

sessionID - the ID of the current session (must be unique among all sessions parallel in progress)

toString

public String toString()

Returns a String that represents the value of this object.

Created (04.10.2001 17:42:12) by Christian Siefkes.

Overrides:

toString in class Object

Returns:

a string representation of the receiver