Development build. This page describes main, not a released package. The latest published Lodestar.Text is 0.6.0 — read its documentation.

Home › Text › Stemming

RomanianSnowballStemmer

Romanian stemming by the Snowball algorithm.

public static class RomanianSnowballStemmer

Example — a noun with its enclitic article, and the same noun in the genitive.

using Lodestar.Text.Stemming;

string bare = RomanianSnowballStemmer.Stem("copil");  // => copil
string articled = RomanianSnowballStemmer.Stem("copilul");  // => copil
string genitive = RomanianSnowballStemmer.Stem("copilului");  // => copil

Remarks — Romanian is the one Romance language among the languages added after the first six, so it is built on the same RV region as Spanish, Portuguese and Italian rather than on R1 alone. Four things about it are worth knowing before reading a stem.

It has five steps, and one of them loops. Step 0 takes the enclitic article and the plural it attaches to; step 1 reduces the combining derivational endings repeatedly, so activitate loses ivitate for iv and stops at activ; step 2 takes the standard endings against R2; step 3 takes the verb endings, but only if neither step 1 nor step 2 removed anything; step 4 takes a final vowel.

Only step 3's region qualifies the search. Everywhere else the longest ending wins outright, and one that reaches past its region stops the step. In step 3 a rejected ending falls through to a shorter one — casem ends with asem, which reaches past RV, and with em, which does not, so em is what goes.

ş and ș are two different letters here. The published description and nltk both write the cedilla forms — ş U+015F and ţ U+0163 — in every suffix table, and Romanian has been correctly written with the comma-below ș U+0219 and ț U+021B since 1993. A word typed the modern way matches only the endings that contain neither letter, so informaţie stems to inform while informație stems to informaț. Nothing is folded, deliberately — decision 0006 has why, and what a caller does about it.

i and u between two vowels are marked before the regions are measured, so they count as consonants and do not close a region early. They come back as themselves.

Reference behaviour is nltk.stem.snowball.SnowballStemmer("romanian"), matched over 349 words.

Applies to — net10.0, netstandard2.0.

See also — the stemming index.

Members

Member	What it does
RomanianSnowballStemmer.Stem	The Snowball stem of one Romanian word.