Reasoner
Class Distiller

java.lang.Object
  |
  +--Reasoner.Distiller

public class Distiller

extends java.lang.Object

Distills a set of MonteCarloRunSets created by the Reasoner into a set of slightly different interpetations, each of which will be rendered as its own Sandbox. Distillation serves two purposes:

1. It clusters different instances of the same object depending on the similarity of their properties.
2. It averages the contents of each cluster to present a single, unified intepretation.

Any number of Distillers can be run on the same aggregrate data for experimentation and comparison purposes. Distillation does not affect the raw source data.

Since distillation eliminates some interpretations by combining them if they are similar, there can never be more interpretations (or Sandboxes) than there are MonteCarloRunSets.

Version:

Created 14 March 2003

Author:

Dan Tappan

See Also:

MonteCarloRunSet, Reasoner, Sandbox

Field Summary

private float	COMPOSITE_OBJECT_TRANSPARENCY
private boolean	ENABLE_VERBOSE_LOGGING
private java.lang.String	m_depiction_name
private java.util.ArrayList	m_monte_carlo_run_sets
private SemanticNetwork	m_semantic_network
private double	SIGMA_SCALE_FACTOR

Constructor Summary

Distiller(java.lang.String depiction_name, SemanticNetwork semantic_network)
Creates a new Distiller.

Method Summary

void	addMonteCarloRunSet(MonteCarloRunSet monte_carlo_run_set) Adds a MonteCarloRunSet to the collection to be distilled.
void	addMonteCarloRunSets(java.util.ArrayList monte_carlo_run_sets) Adds a set of MonteCarloRunSets to the collection to be distilled.
java.util.LinkedList	distillDepiction(boolean return_all_otherwise_one, boolean return_composite) Performs a predefined distillation.
java.util.LinkedList	distillDepiction(boolean distill_by_position, float max_position_sigma, boolean distill_by_attitude, float max_attitude_sigma, boolean distill_by_dimensions, float max_dimensions_sigma, boolean return_all_otherwise_one, boolean return_composite) Distills the MonteCarloRunSets that have been added into a set of Sandboxes for rendering.
void	export(java.lang.String output_filename, java.util.LinkedList object_set) Exports to a file a text representation of the object inferences in the Monte Carlo run sets of this distiller.
int	getMonteCarloRunSetCount() Returns the number of MonteCarloRunSets that have been added.
java.util.ArrayList	getMonteCarloRunSets() Returns a list of the MonteCarloRunSets that have been added.
java.lang.String	toString() Returns a string representation of this object.

Methods inherited from class java.lang.Object

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

Field Detail

m_semantic_network

private SemanticNetwork m_semantic_network

m_monte_carlo_run_sets

private java.util.ArrayList m_monte_carlo_run_sets

m_depiction_name

private java.lang.String m_depiction_name

COMPOSITE_OBJECT_TRANSPARENCY

private final float COMPOSITE_OBJECT_TRANSPARENCY

See Also:

Constant Field Values

ENABLE_VERBOSE_LOGGING

private final boolean ENABLE_VERBOSE_LOGGING

See Also:

Constant Field Values

SIGMA_SCALE_FACTOR

private final double SIGMA_SCALE_FACTOR

See Also:

Constant Field Values

Constructor Detail

Distiller

public Distiller(java.lang.String depiction_name,
                 SemanticNetwork semantic_network)

Creates a new Distiller.

Parameters:

depiction_name - - the depiction name

semantic_network - - the semantic network defining the source description

Method Detail

addMonteCarloRunSet

public void addMonteCarloRunSet(MonteCarloRunSet monte_carlo_run_set)

Adds a MonteCarloRunSet to the collection to be distilled.

Parameters:

monte_carlo_run_set - - the set to add

addMonteCarloRunSets

public void addMonteCarloRunSets(java.util.ArrayList monte_carlo_run_sets)

Adds a set of MonteCarloRunSets to the collection to be distilled.

Parameters:

monte_carlo_run_sets - - the set to add

distillDepiction

public java.util.LinkedList distillDepiction(boolean return_all_otherwise_one,
                                             boolean return_composite)
                                      throws java.lang.Exception

Performs a predefined distillation. This is equivalent to calling distillDepiction() with all the averaging parameters at zero or infinity. At zero all interpretations are returned; at infinity one averaged interpretation is.

Parameters:

return_all_otherwise_one - - whether to return all interpretations or one

return_composite - - whether the composite interpretation of the aggregrate sets should be returned

Returns:

a list of sandboxes defines as Monte Carlo run sets

Throws:

java.lang.Exception - - if distillation fails for any reason

See Also:

Sandbox

distillDepiction

public java.util.LinkedList distillDepiction(boolean distill_by_position,
                                             float max_position_sigma,
                                             boolean distill_by_attitude,
                                             float max_attitude_sigma,
                                             boolean distill_by_dimensions,
                                             float max_dimensions_sigma,
                                             boolean return_all_otherwise_one,
                                             boolean return_composite)
                                      throws java.lang.Exception

Distills the MonteCarloRunSets that have been added into a set of Sandboxes for rendering. Each Sandbox portrays a slight different interpretation of the aggregrate sets as specified in the parameters.

Parameters:

distill_by_position - - whether position deviation contributes to distillation

max_position_sigma - - the minimum sigma difference between object positions to spawn a new interpretation. This applies only when distill_by_position is enabled

distill_by_attitude - - whether attitude deviation contributes to distillation

max_attitude_sigma - - the minimum sigma difference between object attitudes to spawn a new interpretation. This applies only when distill_by_attitude is enabled

distill_by_dimensions - - whether dimensions deviation contribute to distillation

max_dimensions_sigma - - the minimum sigma difference between object dimensions to spawn a new interpretation. This applies only when max_dimensions_sigma is enabled

return_all_otherwise_one - - whether to return all interpretations or one

return_composite - - whether the composite interpretation of the aggregrate sets should be returned

Returns:

a list of sandboxes defines as Monte Carlo run sets

Throws:

java.lang.Exception - - if distillation fails

See Also:

Sandbox

export

public void export(java.lang.String output_filename,
                   java.util.LinkedList object_set)
            throws java.lang.Exception

Exports to a file a text representation of the object inferences in the Monte Carlo run sets of this distiller. This is used by Visio to render a two-dimensional top-view figure.

Parameters:

output_filename - - the fully qualified output filename

object_set - - the distilled object set to export, or null to use the Monte Carlo run sets stored in this distiller, which is equivalent to this.export(output_filename, this.getMonteCarloRunSets())

Throws:

java.lang.Exception - - if the export fails

getMonteCarloRunSetCount

public int getMonteCarloRunSetCount()

Returns the number of MonteCarloRunSets that have been added.

Returns:

the count

getMonteCarloRunSets

public java.util.ArrayList getMonteCarloRunSets()

Returns a list of the MonteCarloRunSets that have been added.

Returns:

the run sets

See Also:

MonteCarloRunSet

toString

public java.lang.String toString()

Returns a string representation of this object.

Overrides:

toString in class java.lang.Object

Returns:

the String representation

See Also:

Support.toPrettyString(java.lang.String)