Class SnowballStemmer

Namespace

NOpenNLP.Tools.Stemmer.Snowball

Assembly

NOpenNLP.Tools.dll

A IStemmer backed by one of the Snowball stemming algorithms.

public class SnowballStemmer : IStemmer

Inheritance

object

SnowballStemmer

Implements

IStemmer

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

The stemmers behind this class are generated by the Snowball compiler from the same algorithm sources Apache OpenNLP generates its Java ones from, pinned to Snowball 2.0.0. Seventeen of the twenty-one produce output identical to OpenNLP 1.9.4 over the full Snowball vocabularies.

FINNISH, HUNGARIAN, FRENCH and INDONESIAN differ, on 0.4% to 6% of words, because OpenNLP ships generated code predating several upstream Snowball fixes and the sources behind it survive at no revision that can be pinned. In each case the difference is the fix: OpenNLP's Finnish gives one noun four different stems, its Hungarian leaves the ablative and delative case suffixes unstripped, and its Indonesian strips a final -i that belongs to the root of loanwords such as demokrasi. See build/generate-snowball-stemmers.ps1 for the measurements and the reasoning.

Constructors

SnowballStemmer(ALGORITHM)

Creates a stemmer for the given algorithm, applying it once.

public SnowballStemmer(ALGORITHM algorithm)

Parameters

algorithm ALGORITHM

SnowballStemmer(ALGORITHM, int)

Creates a stemmer for the given algorithm, applying it repeat times.

public SnowballStemmer(ALGORITHM algorithm, int repeat)

Parameters

algorithm ALGORITHM

repeat int

Exceptions

ArgumentException

Thrown when algorithm is not a known algorithm.

Methods

Stem(string)

public string Stem(string word)

Parameters

word string

Returns

string