/*Stem the given word with, return the stemmed word
*/
public List <string> Stem(string word)
{
if (pipeline == null)
{
Initial();
}
var lemmas = new List <String>();
// create an empty Annotation just with the given text
var document = new Annotation(word);
// run all Annotators on this text
try
{
pipeline.annotate(document);
}
catch (Exception)
{
return(null);
}
// Iterate over all of the sentences found
var senObj = new edu.stanford.nlp.ling.CoreAnnotations.SentencesAnnotation();
var obj = document.get(senObj.getClass());
var tokenObj = new edu.stanford.nlp.ling.CoreAnnotations.TokensAnnotation();
var lemmaObj = new edu.stanford.nlp.ling.CoreAnnotations.LemmaAnnotation();
var sentences = (ArrayList)obj;
foreach (CoreMap sentence in sentences)
{
// Iterate over all tokens in a sentence
lemmas.AddRange(from CoreLabel token in (ArrayList)sentence.get(tokenObj.getClass())
select(string) token.get(lemmaObj.getClass()));
}
return(lemmas);
}
public List <string> Tokenize(string sequence)
{
if (sequence == null)
{
throw new Exception("Sequence should not be null for tokenizer.");
}
if (pipeline == null)
{
Initial();
}
var document = new Annotation(sequence);
pipeline.annotate(document);
var tokens = (ArrayList)document.get(tokenObj.getClass());
return((from CoreMap token in tokens select token.ToString()).ToList());
}
public void Parse(string sentence)
{
if (pipeline == null)
{
Initial();
}
Annotation context = new Annotation(sentence);
pipeline.annotate(context);
this.tokens = (ArrayList)context.get(tokenObj.getClass());
var sentences = (ArrayList)context.get(senObj.getClass());
foreach (CoreMap sen in sentences)
{
this.dependencies = (SemanticGraph)sen.get(depObj.getClass());
break;
}
}
/*Stem the given word with, return the stemmed word
*/
public void FindNer(string context)
{
if (context == null)
{
throw new Exception("Input should not be null for finding NER!");
}
if (pipeline == null)
{
Initial();
}
// create an empty Annotation just with the given text
var annotation = new Annotation(context);
// run all Annotators on this text
pipeline.annotate(annotation);
var ners = (edu.stanford.nlp.util.CoreMap)annotation.get(nerObj.getClass());
var sentences = (ArrayList)annotation.get(new edu.stanford.nlp.ling.CoreAnnotations.SentencesAnnotation().getClass());
nerPairs = new List <Pair <string, string> >();
foreach (CoreMap sentence in sentences)
{
var prevNerToken = "O";
var currNerToken = "O";
bool newToken = true;
var buffer = new StringBuilder();
foreach (CoreLabel token in (ArrayList)sentence.get(tokenObj.getClass()))
{
currNerToken = (string)token.get(nerObj.getClass());
var word = (string)token.get(textObj.getClass());
// Strip out "O"s completely, makes code below easier to understand
if (currNerToken.Equals("O"))
{
if (!prevNerToken.Equals("O") && (buffer.Length > 0))
{
if (prevNerToken.Equals("LOCATION") || prevNerToken.Equals("PERSON") || prevNerToken.Equals("ORGANIZATION"))
{
nerPairs.Add(new Pair <string, string>(buffer.ToString(), prevNerToken));
}
buffer.Clear();
newToken = true;
prevNerToken = "O";
}
continue;
}
if (newToken)
{
prevNerToken = currNerToken;
newToken = false;
buffer.Append(word);
continue;
}
if (currNerToken.Equals(prevNerToken))
{
buffer.Append(" " + word);
}
else
{
// We're done with the current entity - print it out and reset
// TODO save this token into an appropriate ADT to return for useful processing..
if (prevNerToken.Equals("LOCATION") || prevNerToken.Equals("PERSON") || prevNerToken.Equals("ORGANIZATION"))
{
nerPairs.Add(new Pair <string, string>(buffer.ToString(), prevNerToken));
}
buffer.Clear();
buffer.Append(word);
newToken = true;
}
prevNerToken = currNerToken;
}
if (!prevNerToken.Equals("O") && buffer.Length > 0)
{
if (prevNerToken.Equals("LOCATION") || prevNerToken.Equals("PERSON") || prevNerToken.Equals("ORGANIZATION"))
{
nerPairs.Add(new Pair <string, string>(buffer.ToString(), prevNerToken));
}
}
}
}