datafu.pig.hash.lsh.metric

Class Cosine

java.lang.Object

org.apache.pig.EvalFunc<org.apache.pig.data.Tuple>

datafu.pig.hash.lsh.metric.MetricUDF

datafu.pig.hash.lsh.metric.Cosine

public class Cosine
extends MetricUDF

A UDF used to find a vector v in a bag such that for query point q, metric m and threshold t m(v,q) < t. In other words, find the first vector in the bag within a threshold distance away. It returns one of the tuples of the bag of vectors. The metric used is Cosine Similarity, which technically does not form a metric, but I'm stretching the definition here.

See Also:

CosineDistanceHash for an example

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.pig.EvalFunc

org.apache.pig.EvalFunc.SchemaType

Field Summary

Fields inherited from class datafu.pig.hash.lsh.metric.MetricUDF

dim

Fields inherited from class org.apache.pig.EvalFunc

log, pigLogger, reporter, returnType

Constructor Summary

Constructors

Constructor and Description
Cosine(java.lang.String sDim) Create a new Cosine Metric UDF with a given dimension.

Method Summary

Methods

Modifier and Type	Method and Description
protected double	dist(org.apache.commons.math.linear.RealVector v1, org.apache.commons.math.linear.RealVector v2) Cosine similarity.
static double	distance(org.apache.commons.math.linear.RealVector v1, org.apache.commons.math.linear.RealVector v2) Cosine similarity.

Methods inherited from class datafu.pig.hash.lsh.metric.MetricUDF

exec, outputSchema

Methods inherited from class org.apache.pig.EvalFunc

allowCompileTimeCalculation, finish, getArgToFuncMapping, getCacheFiles, getInputSchema, getLogger, getPigLogger, getReporter, getReturnType, getSchemaName, getSchemaType, getShipFiles, isAsynchronous, progress, setInputSchema, setPigLogger, setReporter, setUDFContextSignature, warn

Methods inherited from class java.lang.Object

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

Constructor Detail

Cosine

public Cosine(java.lang.String sDim)

Create a new Cosine Metric UDF with a given dimension.

Parameters:

sDim - dimension

Method Detail

distance

public static double distance(org.apache.commons.math.linear.RealVector v1,
              org.apache.commons.math.linear.RealVector v2)

Cosine similarity.

Parameters:

v1 - first vector

v2 - second vector

Returns:

The cosine of the angle between the vectors

dist

protected double dist(org.apache.commons.math.linear.RealVector v1,
          org.apache.commons.math.linear.RealVector v2)

Cosine similarity.

Specified by:

dist in class MetricUDF

Parameters:

v1 - first vector

v2 - second vector

Returns:

Roughly the cosine of the angle between the vectors