number possible part of speech of the word - stanford-nlp

Question

I know how to get pos of a word in the text but I need to know what would be the possible pos of a word in a sentence for example "like" can have 4 part of speechs: verb noun preposition ....
Is it possible to get that from Stanford library?

Answer 1

Stanford CoreNLP doesn't seem to have an interface to WordNet, but it's pretty easy to do this with one of the other small Java WordNet libraries. For this example, I used JWI 2.3.3.
Besides JWI, you'll need to download a copy of the WordNet database. For example, you can download WordNet-3.0.tar.gz from Princeton. Untar the dictionary.
The following code includes a function that returns a list of the possible parts of speech for a word:
import java.io.File;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Collection;
import edu.mit.jwi.Dictionary;
import edu.mit.jwi.item.POS;
import edu.mit.jwi.item.IIndexWord;
import edu.mit.jwi.morph.WordnetStemmer;
public class WNDemo {
/**
* Given a dictionary and a word, find all the parts of speech the
* word can be.
*/
public static Collection getPartsOfSpeech(Dictionary dict, String word) {
ArrayList<POS> parts = new ArrayList<POS>();
WordnetStemmer stemmer = new WordnetStemmer(dict);
// Check every part of speech.
for (POS pos : POS.values()) {
// Check every stem, because WordNet doesn't have every surface
// form in its database.
for (String stem : stemmer.findStems(word, pos)) {
IIndexWord iw = dict.getIndexWord(stem, pos);
if (iw != null) {
parts.add(pos);
}
}
}
return parts;
}
public static void main(String[] args) {
try {
Dictionary dict = new Dictionary(new File("WordNet-3.0/dict"));
dict.open();
System.out.println("'like' is a " + getPartsOfSpeech(dict, "like"));
} catch (IOException e) {
System.err.println("Error: " + e);
}
}
}
And the output:
'like' is a [noun, verb, adjective]