/// <summary>
/// Tags the specified tokens.
/// </summary>
/// <param name="tokens">The tokens.</param>
/// <returns>The tokens tagged.</returns>
public Token[] Tag(Token[] tokens)
{
Token PreviousToken = null;
for (int i = 0, tokensLength = tokens.Length; i < tokensLength; i++)
{
var Token = tokens[i];
if (Token.TokenType == TokenType.Word)
{
if (Lexicon.TryGetValue(Token.Value, out IEnumerable <string> POS) || Lexicon.TryGetValue(Token.Value.ToLowerInvariant(), out POS))
{
}
else if (Token.Value.Length == 1)
{
POS = new string[] { Token.Value + "^" };
}
else
{
POS = new string[] { "NN" };
}
string Word = Token.PartOfSpeech = POS.First();
if (PreviousToken?.PartOfSpeech == "DT")
{
if (Word == "VBD" || Word == "VBP" || Word == "VB")
{
Token.PartOfSpeech = "NN";
}
}
if (Word.StartsWith("N", StringComparison.OrdinalIgnoreCase) && Token.Value.EndsWith("ED", StringComparison.OrdinalIgnoreCase))
{
Token.PartOfSpeech = "VBN";
}
if (Token.Value.EndsWith("LY", StringComparison.OrdinalIgnoreCase))
{
Token.PartOfSpeech = "RB";
}
if (Token.PartOfSpeech.StartsWith("NN", StringComparison.OrdinalIgnoreCase) && Token.Value.EndsWith("AL", StringComparison.OrdinalIgnoreCase))
{
Token.PartOfSpeech = "JJ";
}
if (Token.PartOfSpeech.StartsWith("NN", StringComparison.OrdinalIgnoreCase) && string.Equals(PreviousToken?.Value, "WOULD", StringComparison.OrdinalIgnoreCase))
{
Token.PartOfSpeech = "VB";
}
if (Token.PartOfSpeech == "NN" && Token.Value.EndsWith("S", StringComparison.OrdinalIgnoreCase))
{
Token.PartOfSpeech = "NNS";
}
if (Token.PartOfSpeech.StartsWith("NN", StringComparison.OrdinalIgnoreCase) && Token.Value.EndsWith("ING", StringComparison.OrdinalIgnoreCase))
{
Token.PartOfSpeech = "VBG";
}
PreviousToken = Token;
}
else if (Token.TokenType == TokenType.Number)
{
Token.PartOfSpeech = "CD";
}
else if (Token.TokenType == TokenType.Email || Token.TokenType == TokenType.HashTag || Token.TokenType == TokenType.Username)
{
Token.PartOfSpeech = "NN";
}
else if (Token.TokenType == TokenType.Emoji)
{
Token.PartOfSpeech = "EM";
}
}
return(tokens);
}
