gr.demokritos.iit.jinsect.documentModel.representations
Class DocumentNGramGraph

java.lang.Object
  gr.demokritos.iit.jinsect.documentModel.representations.DocumentNGramGraph

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable

Direct Known Subclasses:

DocumentNGramDistroGraph, DocumentNGramGaussNormGraph, DocumentNGramSymWinGraph, DocumentWordGraph

public class DocumentNGramGraph
extends java.lang.Object
implements java.io.Serializable, java.lang.Cloneable

Represents the graph of a document, with vertices n-grams of the document and edges the number of the n-grams' co-occurences within a given window.

See Also:

Serialized Form

Field Summary

protected int	CorrelationWindow
protected java.lang.String	DataString
protected java.util.HashMap	DegradedEdges
protected int	MaxSize
protected int	MinSize
protected Graph[]	NGramGraphArray
NormalizerListener	Normalizer
TextPreprocessorListener	TextPreprocessor
WordEvaluatorListener	WordEvaluator

Constructor Summary

DocumentNGramGraph()
Creates a new instance of INSECTDocumentGraph

DocumentNGramGraph(int iMinSize, int iMaxSize, int iCorrelationWindow)
Creates a new instance of INSECTDocumentGraph

Method Summary

double	calcCoexistenceImportance(java.lang.String sNode) Returns a functions of [element graph edges max],[number of neighbours], where [element graph edges max] refers to the maximum weight of the edges including [sNode], and [number of neightbours] is its number of neighbours in the graph.
double	calcCoexistenceImportance(salvo.jesus.graph.Vertex vNode)
java.lang.Object	clone()
void	createEdgesConnecting(Graph gGraph, java.lang.String sStartNode, java.util.List lOtherNodes, java.util.HashMap hAppearenceHistogram) Creates an edge in [gGraph] connecting [sBaseNode] to each node in the [lOtherNodes] list of nodes.
void	createGraphs() Creates the graph of n-grams, for all the levels specified in the MinSize, MaxSize range.
void	createWeightedEdgesConnecting(Graph gGraph, java.lang.String sStartNode, java.util.List lOtherNodes, double dStartWeight, double dNewWeight, double dDataImportance) Creates an edge in [gGraph] connecting [sBaseNode] to each node in the [lOtherNodes] list of nodes.
void	degrade(DocumentNGramGraph dgOtherGraph)
double	degredationDegree(salvo.jesus.graph.Edge e)
void	deleteItem(java.lang.String sItem) Removes an item (node) from all graphs.
java.util.HashSet	getAllNodes()
java.lang.String	getDataString()
Graph	getGraphLevel(int iIndex) Returns graph with M-based index
Graph	getGraphLevelByNGramSize(int iNGramSize) Returns graph with n-gram-size-based index
int	getMaxSize()
int	getMinSize()
int	getWindowSize()
protected void	InitGraphs()
DocumentNGramGraph	intersectGraph(DocumentNGramGraph dgOtherGraph)
DocumentNGramGraph	inverseIntersectGraph(DocumentNGramGraph dgOtherGraph) Returns the difference (inverse of the intersection) graph between the current graph and a given graph.
boolean	isEmpty()
int	length()
void	loadDataStringFromFile(java.lang.String sFilename) Creates the graph based on a data string loaded from a given file.
static void	main(java.lang.String[] args)
void	mergeGraph(DocumentNGramGraph dgOtherGraph, double fWeightPercent) Merges the data of [dgOtherGraph] document graph to the data of this graph, by adding all existing edges and moving the values of those existing in both graphs towards the new graph values based on a tendency modifier.
void	nullify() Sets all weights in all graphs to zero
void	prune(double dMinCoexistenceImportance)
void	setDataString(java.lang.String sDataString)
java.lang.String	toCooccurenceText(java.util.Map mCooccurenceMap)

Methods inherited from class java.lang.Object

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

Field Detail

MinSize

protected int MinSize

MaxSize

protected int MaxSize

CorrelationWindow

protected int CorrelationWindow

DataString

protected java.lang.String DataString

DegradedEdges

protected java.util.HashMap DegradedEdges

Normalizer

public NormalizerListener Normalizer

WordEvaluator

public WordEvaluatorListener WordEvaluator

TextPreprocessor

public TextPreprocessorListener TextPreprocessor

NGramGraphArray

protected Graph[] NGramGraphArray

Constructor Detail

DocumentNGramGraph

public DocumentNGramGraph()

Creates a new instance of INSECTDocumentGraph

DocumentNGramGraph

public DocumentNGramGraph(int iMinSize,
                          int iMaxSize,
                          int iCorrelationWindow)

Creates a new instance of INSECTDocumentGraph

Parameters:

iMinSize - The minimum n-gram size

iMaxSize - The maximum n-gram size

iCorrelationWindow - The maximum distance of terms to be considered as correlated.

Method Detail

InitGraphs

protected void InitGraphs()

length

public int length()

isEmpty

public boolean isEmpty()

loadDataStringFromFile

public void loadDataStringFromFile(java.lang.String sFilename)
                            throws java.io.IOException,
                                   java.io.FileNotFoundException

Creates the graph based on a data string loaded from a given file.

Parameters:

sFilename - The filename of the file containing the data string.

Throws:

java.io.IOException

java.io.FileNotFoundException

getGraphLevel

public Graph getGraphLevel(int iIndex)

Returns graph with M-based index

Parameters:

iIndex - The index of the graph. Zero (0) equals to the graph for level MinSize n-grams.

Returns:

The Graph of the corresponding level.

getGraphLevelByNGramSize

public Graph getGraphLevelByNGramSize(int iNGramSize)

Returns graph with n-gram-size-based index

Parameters:

iNGramSize - The n-gram size of the graph.

Returns:

The Graph of the corresponding level.

getAllNodes

public java.util.HashSet getAllNodes()

createEdgesConnecting

public void createEdgesConnecting(Graph gGraph,
                                  java.lang.String sStartNode,
                                  java.util.List lOtherNodes,
                                  java.util.HashMap hAppearenceHistogram)

Creates an edge in [gGraph] connecting [sBaseNode] to each node in the [lOtherNodes] list of nodes. If an edge exists, its weight is increased by [iIncreaseWeight], else its weight is set to [iStartWeight]

Parameters:

gGraph - The graph to use

sStartNode - The node from which all edges begin

lOtherNodes - The list of nodes to which sBaseNode is connected

hAppearenceHistogram - The histogram of appearences of the terms

createWeightedEdgesConnecting

public void createWeightedEdgesConnecting(Graph gGraph,
                                          java.lang.String sStartNode,
                                          java.util.List lOtherNodes,
                                          double dStartWeight,
                                          double dNewWeight,
                                          double dDataImportance)

Creates an edge in [gGraph] connecting [sBaseNode] to each node in the [lOtherNodes] list of nodes. If an edge exists, its weight is increased by [iIncreaseWeight], else its weight is set to [iStartWeight]

Parameters:

gGraph - The graph to use

sStartNode - The node from which all edges begin

lOtherNodes - The list of nodes to which sBaseNode is connected

dStartWeight - The initial weight for first-occuring nodes

dNewWeight - The new weight

dDataImportance - The tendency towards the new value. 0.0 means no change to the current value. 1.0 means the old value is completely replaced by the new. 0.5 means the final value is the average of the old and the new.

createGraphs

public void createGraphs()

Creates the graph of n-grams, for all the levels specified in the MinSize, MaxSize range.

mergeGraph

public void mergeGraph(DocumentNGramGraph dgOtherGraph,
                       double fWeightPercent)

Merges the data of [dgOtherGraph] document graph to the data of this graph, by adding all existing edges and moving the values of those existing in both graphs towards the new graph values based on a tendency modifier. The convergence tendency towards the starting value or the new value is determined by [fWeightPercent].

Parameters:

dgOtherGraph - The second graph used for the merging

fWeightPercent - The convergence tendency parameter. A value of 0.0 means no change to existing value, 1.0 means new value is the same as that of the new graph. A value of 0.5 means new value is exactly between the old and new value (average).

intersectGraph

public DocumentNGramGraph intersectGraph(DocumentNGramGraph dgOtherGraph)

inverseIntersectGraph

public DocumentNGramGraph inverseIntersectGraph(DocumentNGramGraph dgOtherGraph)

Returns the difference (inverse of the intersection) graph between the current graph and a given graph.

Parameters:

dgOtherGraph - The graph to compare to.

Returns:

A DocumentNGramGraph that is the difference between the current graph and the given graph.

getMinSize

public int getMinSize()

getMaxSize

public int getMaxSize()

getWindowSize

public int getWindowSize()

calcCoexistenceImportance

public double calcCoexistenceImportance(java.lang.String sNode)

Returns a functions of [element graph edges max],[number of neighbours], where [element graph edges max] refers to the maximum weight of the edges including [sNode], and [number of neightbours] is its number of neighbours in the graph.

Parameters:

sNode - The node object the Coexistence Importance of which we calculate

calcCoexistenceImportance

public double calcCoexistenceImportance(salvo.jesus.graph.Vertex vNode)

prune

public void prune(double dMinCoexistenceImportance)

deleteItem

public void deleteItem(java.lang.String sItem)

Removes an item (node) from all graphs.

Parameters:

sItem - The item to remove.

nullify

public void nullify()

Sets all weights in all graphs to zero

setDataString

public void setDataString(java.lang.String sDataString)

getDataString

public java.lang.String getDataString()

degrade

public void degrade(DocumentNGramGraph dgOtherGraph)

degredationDegree

public double degredationDegree(salvo.jesus.graph.Edge e)

toCooccurenceText

public java.lang.String toCooccurenceText(java.util.Map mCooccurenceMap)

main

public static void main(java.lang.String[] args)

clone

public java.lang.Object clone()

Overrides:

clone in class java.lang.Object