Lodestar.Text 0.6.0. This page is frozen at that release. Read the current documentation for what main says now. A link to a decision or a migration page follows main, and leaves the archive.

ArabicSnowballStemmer.Stem

The Snowball stem of one Arabic word.

public static string Stem(string word)

Parameters — word is a single Arabic word. Vocalisation marks, the kasheeda and the lam-alef ligatures are resolved before the rules run, so a fully vocalised spelling and a bare one reach the same stem.

Returns — string, the stem. Text the algorithm has no rule for — Latin letters, ASCII digits — comes back unchanged.

Exceptions — ArgumentNullException when word is null. An empty string, or a word of one character, is returned unchanged.

Example — the definite article and a preposition in front of it, and a feminine noun.

using Lodestar.Text.Stemming;

string article = ArabicSnowballStemmer.Stem("بالكتاب");  // => كتاب
string feminine = ArabicSnowballStemmer.Stem("المدرسة");  // => مدرس
string vocalised = ArabicSnowballStemmer.Stem("مُحَمَّد");  // => محمد

Remarks — المدرسة → مدرس shows both ends working at once: the feminine ة goes with the suffix steps and the article ال with the prefix steps, and the same مدرس is what مدرسة and مدرستها reach. A stemmer produces keys, not words.

The two results below are the ones most likely to look wrong. معلمات keeps its ا, because the rule that removes a plural ات needs a longer word than this one and only the ت is taken; and ١٢٣ becomes 123, because folding the Arabic-Indic digits onto the ASCII ones is part of the published normalisation rather than an accident.

using Lodestar.Text.Stemming;

string plural = ArabicSnowballStemmer.Stem("معلمات");  // => معلما
string digits = ArabicSnowballStemmer.Stem("١٢٣");  // => 123
string verb = ArabicSnowballStemmer.Stem("يستخدم");  // => استخدم

Applies to — net10.0, netstandard2.0.

See also — ArabicSnowballStemmer, the stemming index.