public void ExtensiveTest()
{
// Create an array with words to test
string[] words = new string[] { "consign", "consigned", "consigning", "consignment", "consist", "consisted", "consistency", "consistent", "consistently",
"consisting", "consists", "consolation", "consolations", "consolatory", "console", "consoled", "consoles", "consolidate", "consolidated", "consolidating",
"consoling", "consolingly", "consols", "consonant", "consort", "consorted", "consorting", "conspicuous", "conspicuously",
"conspiracy", "conspirator", "conspirators", "conspire", "conspired", "conspiring", "constable", "constables", "constance", "constancy", "constant",
"knack", "knackeries", "knacks", "knag", "knave", "knaves", "knavish", "kneaded", "kneading", "knee", "kneel", "kneeled", "kneeling", "kneels",
"knees", "knell", "knelt", "templates", "cry", "Sky", "absolutely" };
// Create an array with correct steams
string[] steams = new string[] { "consign", "consigned", "consigning", "consignment", "consist", "consisted", "consistency", "consistent", "consistently",
"consisting", "consists", "consolation", "consolations", "consolatory", "console", "consoled", "consoles", "consolidate", "consolidated", "consolidating",
"consoling", "consolingly", "consols", "consonant", "consort", "consorted", "consorting", "conspicuous", "conspicuously",
"conspiracy", "conspirator", "conspirators", "conspire", "conspired", "conspiring", "constable", "constables", "constance", "constancy", "constant",
"knack", "knackeries", "knacks", "knag", "knave", "knaves", "knavish", "kneaded", "kneading", "knee", "kneel", "kneeled", "kneeling", "kneels",
"knees", "knell", "knelt", "templates", "cry", "Sky", "absolutely" };
// Create a default stemmer
Stemmer stemmer = new DefaultStemmer();
// Test the stemmer
for (int i = 0; i < words.Length; i++)
{
Assert.AreEqual(steams[i], stemmer.GetSteamWord(words[i]));
}
} // End of the ExtensiveTest method
public void StemTokens()
{
var Result = new DefaultStemmer(new IStemmerLanguage[] { new EnglishLanguage() }).Stem(new DefaultTokenizer(new ITokenizerLanguage[] { new Enlighten.Tokenizer.Languages.English.EnglishLanguage(new IEnglishTokenFinder[] { new Word(), new Whitespace(), new Symbol() }) }).Tokenize("This is a test.", TokenizerLanguage.EnglishRuleBased), StemmerLanguage.EnglishPorter2);
Assert.Equal("this", Result[0].Value);
Assert.Equal(" ", Result[1].Value);
Assert.Equal("is", Result[2].Value);
Assert.Equal(" ", Result[3].Value);
Assert.Equal("a", Result[4].Value);
Assert.Equal(" ", Result[5].Value);
Assert.Equal("test", Result[6].Value);
Assert.Equal(".", Result[7].Value);
Result = new DefaultStemmer(new IStemmerLanguage[] { new EnglishLanguage() }).Stem(new DefaultTokenizer(new ITokenizerLanguage[] { new Enlighten.Tokenizer.Languages.English.EnglishLanguage(new IEnglishTokenFinder[] { new Word(), new Whitespace(), new Symbol() }) }).Tokenize("These are some more tests.", TokenizerLanguage.EnglishRuleBased), StemmerLanguage.EnglishPorter2);
Assert.Equal("these", Result[0].Value);
Assert.Equal(" ", Result[1].Value);
Assert.Equal("are", Result[2].Value);
Assert.Equal(" ", Result[3].Value);
Assert.Equal("some", Result[4].Value);
Assert.Equal(" ", Result[5].Value);
Assert.Equal("more", Result[6].Value);
Assert.Equal(" ", Result[7].Value);
Assert.Equal("test", Result[8].Value);
Assert.Equal(".", Result[9].Value);
}
public void ExtensiveTest()
{
// Create an array with words to test
string[] words = new string[] { "consign", "consigned", "consigning", "consignment", "consist", "consisted", "consistency", "consistent", "consistently",
"consisting", "consists", "consolation", "consolations", "consolatory", "console", "consoled", "consoles", "consolidate", "consolidated", "consolidating",
"consoling", "consolingly", "consols", "consonant", "consort", "consorted", "consorting", "conspicuous", "conspicuously",
"conspiracy", "conspirator", "conspirators", "conspire", "conspired", "conspiring", "constable", "constables", "constance", "constancy", "constant",
"knack", "knackeries", "knacks", "knag", "knave", "knaves", "knavish", "kneaded", "kneading", "knee", "kneel", "kneeled", "kneeling", "kneels",
"knees", "knell", "knelt", "templates", "cry", "Sky", "absolutely" };
// Create an array with correct steams
string[] steams = new string[] { "consign", "consigned", "consigning", "consignment", "consist", "consisted", "consistency", "consistent", "consistently",
"consisting", "consists", "consolation", "consolations", "consolatory", "console", "consoled", "consoles", "consolidate", "consolidated", "consolidating",
"consoling", "consolingly", "consols", "consonant", "consort", "consorted", "consorting", "conspicuous", "conspicuously",
"conspiracy", "conspirator", "conspirators", "conspire", "conspired", "conspiring", "constable", "constables", "constance", "constancy", "constant",
"knack", "knackeries", "knacks", "knag", "knave", "knaves", "knavish", "kneaded", "kneading", "knee", "kneel", "kneeled", "kneeling", "kneels",
"knees", "knell", "knelt", "templates", "cry", "Sky", "absolutely" };
// Create a default stemmer
Stemmer stemmer = new DefaultStemmer();
// Test the stemmer
for (int i = 0; i < words.Length; i++)
{
Assert.AreEqual(steams[i], stemmer.GetSteamWord(words[i]));
}
}
public void Stem()
{
var Result = new DefaultStemmer(new IStemmerLanguage[] { new EnglishLanguage() }).Stem(new string[] { "This", "is", "a", "test" }, StemmerLanguage.EnglishPorter2);
Assert.Equal("this", Result[0]);
Assert.Equal("is", Result[1]);
Assert.Equal("a", Result[2]);
Assert.Equal("test", Result[3]);
Result = new DefaultStemmer(new IStemmerLanguage[] { new EnglishLanguage() }).Stem(new string[] { "These", "are", "some", "more", "tests" }, StemmerLanguage.EnglishPorter2);
Assert.Equal("these", Result[0]);
Assert.Equal("are", Result[1]);
Assert.Equal("some", Result[2]);
Assert.Equal("more", Result[3]);
Assert.Equal("test", Result[4]);
}
} // End of the GetCurrentDomain method
/// <summary>
/// Get the stemmer based on the language
/// </summary>
/// <param name="language">A reference to the language</param>
/// <returns>A reference to a Stemmer</returns>
public static Stemmer GetStemmer(Language language)
{
// Create a default stemmer
Stemmer stemmer = new DefaultStemmer();
// Get the language code in lower case
string language_code = language.language_code.ToLower();
// Get a stemmer depending on the language
if (language_code == "da")
{
stemmer = new DanishStemmer();
}
else if (language_code == "nl")
{
stemmer = new DutchStemmer();
}
else if (language_code == "en")
{
stemmer = new EnglishStemmer();
}
else if (language_code == "fi")
{
stemmer = new FinnishStemmer();
}
else if (language_code == "fr")
{
stemmer = new FrenchStemmer();
}
else if (language_code == "de")
{
stemmer = new GermanStemmer();
}
else if (language_code == "it")
{
stemmer = new ItalianStemmer();
}
else if (language_code == "no")
{
stemmer = new NorwegianStemmer();
}
else if (language_code == "pt")
{
stemmer = new PortugueseStemmer();
}
else if (language_code == "ro")
{
stemmer = new RomanianStemmer();
}
else if (language_code == "es")
{
stemmer = new SpanishStemmer();
}
else if (language_code == "sv")
{
stemmer = new SwedishStemmer();
}
// Return the stemmer
return stemmer;
} // End of the GetStemmer method
