org.fit.layout.classify.articles

Class ArticleFeatureExtractor

java.lang.Object

org.fit.layout.classify.DefaultFeatureExtractor

org.fit.layout.classify.articles.ArticleFeatureExtractor

All Implemented Interfaces:

FeatureExtractor

public class ArticleFeatureExtractor
extends DefaultFeatureExtractor

A feature extractor for generic article processing.

Author:

burgetr

Field Summary

Fields

Modifier and Type	Field and Description
static double	CENTERING_THRESHOLD Maximal difference between left and right margin to consider the area to be centered (percentage of the parent area width)
static double[]	DEFAULT_WEIGHTS
static double	MIN_MARKEDNESS_DIFFERENCE Minimal difference in the markedness that should be interpreted as a difference between the meaning of the areas.

Constructor Summary

Constructors

Constructor and Description
ArticleFeatureExtractor()

Method Summary

Methods

Modifier and Type	Method and Description
static double	colorLuminosity(Color c)
weka.core.Instances	createEmptyDataset() Creates an empty data set containing the header appropriate to this extractor.
protected Set<org.fit.layout.model.Tag>	getAllTags(org.fit.layout.model.Area area) Obtains all the tags assigned to this area and its child areas (not all descendant areas).
weka.core.Instance	getAreaFeatures(org.fit.layout.model.Area node, weka.core.Instances dataset) Creates a classification data instance from the given area.
double	getAverageBoxColorLuminosity(org.fit.layout.model.Area area)
FeatureVector	getFeatureVector(org.fit.layout.model.Area node)
double	getIndentation(org.fit.layout.model.Area node) Computes the indentation metric.
int	getLineCount(org.fit.layout.model.Area a)
double	getMarkedness(org.fit.layout.model.Area node) Computes the markedness of the area.
double	getRelY(org.fit.layout.model.Area a)
org.fit.layout.model.Area	getTreeRoot() Obtains the current tree root.
double[]	getWeights()
boolean	isCentered(org.fit.layout.model.Area area) Checks whether the area is horizontally centered within its parent area
void	setTree(org.fit.layout.model.Area rootNode) Initializes the extractor to use the tree with the given root node.
void	setWeights(double[] weights)

Methods inherited from class org.fit.layout.classify.DefaultFeatureExtractor

loadArffDatasetResource

Methods inherited from class java.lang.Object

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

Field Detail

MIN_MARKEDNESS_DIFFERENCE

public static final double MIN_MARKEDNESS_DIFFERENCE

Minimal difference in the markedness that should be interpreted as a difference between the meaning of the areas.

See Also:

Constant Field Values

DEFAULT_WEIGHTS

public static final double[] DEFAULT_WEIGHTS

CENTERING_THRESHOLD

public static final double CENTERING_THRESHOLD

Maximal difference between left and right margin to consider the area to be centered (percentage of the parent area width)

See Also:

Constant Field Values

Constructor Detail

ArticleFeatureExtractor

public ArticleFeatureExtractor()

Method Detail

setTree

public void setTree(org.fit.layout.model.Area rootNode)

Description copied from interface: FeatureExtractor

Initializes the extractor to use the tree with the given root node.

Parameters:

rootNode - the new area tree root node

getTreeRoot

public org.fit.layout.model.Area getTreeRoot()

Description copied from interface: FeatureExtractor

Obtains the current tree root.

Returns:

the current tree root area or null when the tree was not set before.

createEmptyDataset

public weka.core.Instances createEmptyDataset()

Description copied from interface: FeatureExtractor

Creates an empty data set containing the header appropriate to this extractor.

Returns:

an empty data set

getAreaFeatures

public weka.core.Instance getAreaFeatures(org.fit.layout.model.Area node,
                                 weka.core.Instances dataset)

Description copied from interface: FeatureExtractor

Creates a classification data instance from the given area.

Parameters:

node - the area whose features should be computed

dataset - the data set the created instance should belong to

Returns:

the classification data instance

setWeights

public void setWeights(double[] weights)

getWeights

public double[] getWeights()

getMarkedness

public double getMarkedness(org.fit.layout.model.Area node)

Computes the markedness of the area. The markedness generally describes the visual importance of the area based on different criteria.

Returns:

the computed expressiveness

getFeatureVector

public FeatureVector getFeatureVector(org.fit.layout.model.Area node)

isCentered

public boolean isCentered(org.fit.layout.model.Area area)

Checks whether the area is horizontally centered within its parent area

Returns:

true if the area is centered

getIndentation

public double getIndentation(org.fit.layout.model.Area node)

Computes the indentation metric.

Returns:

the indentation metric (0..1) where 1 is for the non-indented areas, 0 for the most indented areas.

getAverageBoxColorLuminosity

public double getAverageBoxColorLuminosity(org.fit.layout.model.Area area)

colorLuminosity

public static double colorLuminosity(Color c)

getRelY

public double getRelY(org.fit.layout.model.Area a)

getLineCount

public int getLineCount(org.fit.layout.model.Area a)

getAllTags

protected Set<org.fit.layout.model.Tag> getAllTags(org.fit.layout.model.Area area)

Obtains all the tags assigned to this area and its child areas (not all descendant areas).

Returns:

a set of tags