/// <summary>
/// Stems the a string containing ngrams. Only returns the stemmed version
/// of words that are actually in the dictionary.
/// </summary>
/// <param name="ngram">The ngram.</param>
/// <param name="dictionary">The dictionary.</param>
/// <returns></returns>
public static String StemNgram(String ngram, IDictionary <String, String> dictionary)
{
var newNgram = new StringBuilder();
string[] ngramsArray = ngram.Split(' ');
foreach (string currentWord in ngramsArray)
{
var stemmer = new EnglishWordStemmer(currentWord);
string stemmedNgram = stemmer.Stem;
// only use the stemmed version if it is an actual word in our dictionary
if (dictionary.ContainsKey(stemmedNgram))
{
newNgram.Append(stemmedNgram);
}
else
{
newNgram.Append(currentWord);
}
// space out our words
newNgram.Append(" ");
}
return(newNgram.ToString().Trim());
}
public void BasicStemmerTest()
{
var stemmer = new EnglishWordStemmer("");
Assert.AreEqual(stemmer.Stem, "");
stemmer = new EnglishWordStemmer("exceedingly");
Assert.AreEqual(stemmer.Stem, "exceed");
stemmer = new EnglishWordStemmer("running");
Assert.AreEqual(stemmer.Stem, "run");
// TODO: add dictionary check to results, return orig if not valid
//stemmer = new EnglishWordStemmer("happier");
//Assert.AreEqual(stemmer.Stem, "happy");
stemmer = new EnglishWordStemmer("slim");
Assert.AreEqual(stemmer.Stem, "slim");
}
