public StemPair(string Stem, KeywordPair KeywordRef, Artifact ArtifactRef)
{
this.stem = Stem;
this.keywordRef = KeywordRef;
this.artifactRef = ArtifactRef;
}
private void AddKeywordsStems(string unparsed)
{
string[] chunks = unparsed.Split(new string[2] { "\r\n", "\n" }, StringSplitOptions.None);
foreach (string chunk in chunks)
{
// add english keywords to the list first
string longKeywordEnglish = ExtractLanguage(LANGUAGE_ENGLISH, chunk);
string longKeywordFrench = ExtractLanguage(LANGUAGE_FRENCH, chunk);
KeywordPair keywordPair = new KeywordPair(longKeywordEnglish, longKeywordFrench, this);
if (longKeywordEnglish != "" && longKeywordFrench != "")
{
mKeywords.Add(keywordPair);
}
// now get the stems for each keyword, using the english keywords since that is what the stemmer works for
// split each english keyword into separate words
string[] preStemWordList = longKeywordEnglish.Split(new string[4] { " ", "-", "&", "," }, StringSplitOptions.None);
foreach (string word in preStemWordList)
{
// and get the stem for each word
string stem = Stemmer.Stem(word.ToLower());
StemPair stemPair = new StemPair(stem, keywordPair, this);
if (stem != "" && Stopwords.Contains(stem) == false)
{
mStems.Add(stemPair);
}
}
}
}
