Text 0.4.0 portuguesesnowballstemmer

Lodestar.Text 0.4.0. 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.

PortugueseSnowballStemmer

Portuguese stemming by the Snowball algorithm.

public static class PortugueseSnowballStemmer

Example — a noun's four forms on one key.

using Lodestar.Text.Stemming;

string masculine = PortugueseSnowballStemmer.Stem("musico");  // => music
string feminine = PortugueseSnowballStemmer.Stem("musica");  // => music
string plural = PortugueseSnowballStemmer.Stem("musicas");  // => music

Remarks — Portuguese and Spanish look alike and their stemmers are not interchangeable. Two differences decide it.

Nasal vowels are expanded before the rules run: ã becomes a~ and õ becomes o~, two characters where there was one. This keeps the nasal from being read as a plain vowel by a suffix rule, and it is undone before the stem is returned.

Accents are kept, where the Spanish algorithm strips them at the end — 19 of the 105 pinned words stem to something still carrying one. So a corpus that drops accents will not index to the same keys as one that keeps them, the opposite of the Spanish situation and the thing to check before mixing sources. It also means a nasal can outlive the rules that would have merged two forms: Stem has the geração/gerações case, where a singular and its plural end on different keys.

Reference behaviour is nltk.stem.snowball.SnowballStemmer("portuguese"), matched over 105 words.

Applies to — net10.0, netstandard2.0.

See also — SpanishSnowballStemmer, the neighbour it is easy to reach for by mistake, and the stemming index.

Members

Member	What it does
PortugueseSnowballStemmer.Stem	The Snowball stem of one Portuguese word.