public static IEnumerable <POS> Extract(string text, ref NLPCount count)
{
string normalized = TwitterKoreanProcessor.normalize(text).toString();
var tokenized = TwitterKoreanProcessor.tokenize(normalized);
var stemmed = TwitterKoreanProcessor.stem(tokenized);
return(stemmed.ToPosListFromTokens().Where(s => s.PosTag != "Punctuation" && s.PosTag != "Space"));
}
public static IEnumerable <string> TokensToStrings(this IEnumerable <KoreanToken> tokens)
{
var scalaTokenSeq = Utils.ScalaCSHelper.ReverseScalaSeqConverter <KoreanTokenizer.KoreanToken, KoreanToken>(tokens, t => t.ToScalaToken());
var scalaResults = TwitterKoreanProcessor.tokensToStrings(scalaTokenSeq);
var results = Utils.ScalaCSHelper.ScalaSeqStringConverter(scalaResults);
return(results);
}
public static IEnumerable <KoreanToken> Tokenize(this string text)
{
var scalaResults = TwitterKoreanProcessor.tokenize(text);
List <KoreanToken> results = Utils.ScalaCSHelper.ScalaSeqConverter <KoreanToken, KoreanTokenizer.KoreanToken>(
scalaResults, (scalaResult) => { return(new KoreanToken(scalaResult)); });
return(results);
}
public static IEnumerable <KoreanToken> Stem(this IEnumerable <KoreanToken> tokens)
{
var scalaTokenSeq = Utils.ScalaCSHelper.ReverseScalaSeqConverter <KoreanTokenizer.KoreanToken, KoreanToken>(tokens, t => t.ToScalaToken());
var scalaResults = TwitterKoreanProcessor.stem(scalaTokenSeq);
var results = Utils.ScalaCSHelper.ScalaSeqConverter <KoreanToken, KoreanTokenizer.KoreanToken>(scalaResults, t => new KoreanToken(t));
return(results);
}
public static List <KoreanPhrase> ExtractPhrases(this IEnumerable <KoreanToken> tokens, bool filterSpam = false, bool enableHashTags = true)
{
var scalaTokenSeq = Utils.ScalaCSHelper.ReverseScalaSeqConverter <KoreanTokenizer.KoreanToken, KoreanToken>(tokens, t => t.ToScalaToken());
// returns: Seq[KoreanPhrase]
var scalaResults = TwitterKoreanProcessor.extractPhrases(scalaTokenSeq, filterSpam, enableHashTags);
var results = Utils.ScalaCSHelper.ScalaSeqConverter <KoreanPhrase, ScalaKoreanPhrase>(scalaResults, p => new KoreanPhrase(p));
return(results);
}
public static string Normalize(string text)
{
string result = TwitterKoreanProcessor.normalize(text).toString();
return(result);
}
