de.linguatools.disco

Class SparseVector

java.lang.Object

de.linguatools.disco.SparseVector

public class SparseVector
extends java.lang.Object

Methods for sparse vectors. A sparse vector is stored in an object of type Map.

Constructor Summary

Constructors

Constructor and Description
SparseVector()

Method Summary

Modifier and Type	Method and Description
static java.util.Map<java.lang.String,java.lang.Float>	add(java.util.Map<java.lang.String,java.lang.Float> v1, java.util.Map<java.lang.String,java.lang.Float> v2) Adds the two argument map vectors.
static java.util.Map<java.lang.String,java.lang.Float>	average(java.util.List<java.util.Map<java.lang.String,java.lang.Float>> vectorList) Compute the average vector of all vectors in the list.
static float	dotProduct(java.util.Map<java.lang.String,java.lang.Float> v1, java.util.Map<java.lang.String,java.lang.Float> v2) Compute the dot product (inner product, scalar product) of v1 and v2.
static java.util.Map<java.lang.String,java.lang.Float>	getMapVector(org.apache.lucene.document.Document doc) Convert a Document from a DISCOLuceneIndex (which stores a word's data) to a map vector.
static java.util.Map<java.lang.String,java.lang.Float>	getNormalizedVector(java.util.Map<java.lang.String,java.lang.Float> mapVector)
static java.util.Map<java.lang.String,java.lang.Float>	getSecondOrderMapVector(org.apache.lucene.document.Document doc) Get the second order word vector from doc as a sparse vector. Important: this only works with documents from a DISCOLuceneIndex of type WordspaceType.SIM.
static java.util.Map<java.lang.String,java.lang.Float>	mul(java.util.Map<java.lang.String,java.lang.Float> v, float scalarValue) Multiply vector with scalar.
static java.util.Map<java.lang.String,java.lang.Float>	mul(java.util.Map<java.lang.String,java.lang.Float> v1, java.util.Map<java.lang.String,java.lang.Float> v2) Element-wise vector multiplication.
static float	norm(java.util.Map<java.lang.String,java.lang.Float> mapVector)
static java.util.Map<java.lang.String,java.lang.Float>	sub(java.util.Map<java.lang.String,java.lang.Float> v1, java.util.Map<java.lang.String,java.lang.Float> v2) Subtract the second word vector from the first.
static java.util.Map<java.lang.String,java.lang.Float>	vectorExtrema(java.util.Map<java.lang.String,java.lang.Float> v1, java.util.Map<java.lang.String,java.lang.Float> v2) Choose for each dimension the highest absolute value.

Methods inherited from class java.lang.Object

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

Constructor Detail

SparseVector

public SparseVector()

Method Detail

getMapVector

public static java.util.Map<java.lang.String,java.lang.Float> getMapVector(org.apache.lucene.document.Document doc)

Convert a Document from a DISCOLuceneIndex (which stores a word's data) to a map vector.

Parameters:

doc -

Returns:

map vector representation of word vector from doc.

getSecondOrderMapVector

public static java.util.Map<java.lang.String,java.lang.Float> getSecondOrderMapVector(org.apache.lucene.document.Document doc)

Get the second order word vector from doc as a sparse vector.
Important: this only works with documents from a DISCOLuceneIndex of type WordspaceType.SIM. Documents from a word space of type COL always return null.

Parameters:

doc -

Returns:

map vector representation of the second order word vector from doc. The second order word vector are the distributionally most similar words for a word with their similarity scores. If doc is from a word space of type COL then the return value is null.

norm

public static float norm(java.util.Map<java.lang.String,java.lang.Float> mapVector)

Parameters:

mapVector -

Returns:

the L2-norm of mapVector.

getNormalizedVector

public static java.util.Map<java.lang.String,java.lang.Float> getNormalizedVector(java.util.Map<java.lang.String,java.lang.Float> mapVector)

Parameters:

mapVector -

Returns:

mapVector converted to unit length.

add

public static java.util.Map<java.lang.String,java.lang.Float> add(java.util.Map<java.lang.String,java.lang.Float> v1,
                                                                  java.util.Map<java.lang.String,java.lang.Float> v2)

Adds the two argument map vectors.

Parameters:

v1 -

v2 -

Returns:

v1 + v2

sub

public static java.util.Map<java.lang.String,java.lang.Float> sub(java.util.Map<java.lang.String,java.lang.Float> v1,
                                                                  java.util.Map<java.lang.String,java.lang.Float> v2)

Subtract the second word vector from the first.

Parameters:

v1 -

v2 -

Returns:

v1 - v2

mul

public static java.util.Map<java.lang.String,java.lang.Float> mul(java.util.Map<java.lang.String,java.lang.Float> v1,
                                                                  java.util.Map<java.lang.String,java.lang.Float> v2)

Element-wise vector multiplication.

Parameters:

v1 -

v2 -

Returns:

mul

public static java.util.Map<java.lang.String,java.lang.Float> mul(java.util.Map<java.lang.String,java.lang.Float> v,
                                                                  float scalarValue)

Multiply vector with scalar.

Parameters:

v -

scalarValue -

Returns:

dotProduct

public static float dotProduct(java.util.Map<java.lang.String,java.lang.Float> v1,
                               java.util.Map<java.lang.String,java.lang.Float> v2)

Compute the dot product (inner product, scalar product) of v1 and v2.

Parameters:

v1 -

v2 -

Returns:

result (a scalar, not a vector)

average

public static java.util.Map<java.lang.String,java.lang.Float> average(java.util.List<java.util.Map<java.lang.String,java.lang.Float>> vectorList)

Compute the average vector of all vectors in the list.

Parameters:

vectorList -

Returns:

average vector

Since:

3.0

vectorExtrema

public static java.util.Map<java.lang.String,java.lang.Float> vectorExtrema(java.util.Map<java.lang.String,java.lang.Float> v1,
                                                                            java.util.Map<java.lang.String,java.lang.Float> v2)

Choose for each dimension the highest absolute value.

Parameters:

v1 -

v2 -

Returns:

extrema vector.