[DL-Learner SVN] SF.net SVN: dl-learner:[4061] trunk/components-core/src/main/java/org/ dllearner/algorithms/isle

[<dfl...@us...>]

Revision: 4061
          http://sourceforge.net/p/dl-learner/code/4061
Author:   dfleischhacker
Date:     2013-09-04 14:28:12 +0000 (Wed, 04 Sep 2013)
Log Message:
-----------
Add class providing wrapper for common linguistic operations

Modified Paths:
--------------
    trunk/components-core/src/main/java/org/dllearner/algorithms/isle/WordNet.java

Added Paths:
-----------
    trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/LinguisticUtil.java

Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/WordNet.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/WordNet.java	2013-09-04 14:26:47 UTC (rev 4060)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/WordNet.java	2013-09-04 14:28:12 UTC (rev 4061)
@@ -4,6 +4,7 @@
 import java.util.ArrayList;
 import java.util.Iterator;
 import java.util.List;
+import java.util.Set;
 
 import net.didion.jwnl.JWNL;
 import net.didion.jwnl.JWNLException;
@@ -71,8 +72,31 @@
 		}
 		return synonyms;
 	}
-	
-	public List<String> getSisterTerms(POS pos, String s){
+
+    public List<String> getAllSynonyms(POS pos, String s) {
+        List<String> synonyms = new ArrayList<String>();
+        try {
+            IndexWord iw = dict.getIndexWord(pos, s);
+            if (iw != null) {
+                Synset[] synsets = iw.getSenses();
+                for (Synset synset : synsets) {
+                    for (Word w : synset.getWords()) {
+                        String lemma = w.getLemma();
+                        if (!lemma.equals(s) && !lemma.contains(" ")) {
+                            synonyms.add(lemma);
+                        }
+                    }
+                }
+            }
+        }
+        catch (JWNLException e) {
+            e.printStackTrace();
+        }
+
+        return synonyms;
+    }
+
+    public List<String> getSisterTerms(POS pos, String s){
 		List<String> sisterTerms = new ArrayList<String>();
 		
 		try {

Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/LinguisticUtil.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/LinguisticUtil.java	                        (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/LinguisticUtil.java	2013-09-04 14:28:12 UTC (rev 4061)
@@ -0,0 +1,78 @@
+package org.dllearner.algorithms.isle.index;
+
+import net.didion.jwnl.data.POS;
+import org.dllearner.algorithms.isle.WordNet;
+
+import java.util.ArrayList;
+import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+/**
+ * Provides shortcuts to
+ * @author Daniel Fleischhacker
+ */
+public class LinguisticUtil {
+    private static final WordNet wn = new WordNet();
+    private static POS[] RELEVANT_POS = new POS[]{POS.NOUN, POS.VERB};
+
+    /**
+     * Processes the given string and puts camelCased words into single words.
+     * @param camelCase    the word containing camelcase to split
+     * @return all words as camelcase contained in the given word
+     */
+    public static String[] getWordsFromCamelCase(String camelCase) {
+        ArrayList<String> resultingWords = new ArrayList<String>();
+        StringBuilder sb = new StringBuilder();
+        for (int i = 0; i < camelCase.length(); i++) {
+            // we just ignore characters not matching the defined pattern
+            char curChar = camelCase.charAt(i);
+            if (!Character.isLetter(curChar)) {
+                continue;
+            }
+            if (Character.isUpperCase(curChar)) { // found a new upper case letter
+                resultingWords.add(sb.toString());
+                sb = new StringBuilder();
+                sb.append(Character.toLowerCase(curChar));
+            }
+            else { // lower case letter
+                sb.append(curChar);
+            }
+        }
+
+        if (sb.length() > 0) {
+            resultingWords.add(sb.toString());
+        }
+
+        return resultingWords.toArray(new String[resultingWords.size()]);
+    }
+
+    /**
+     * Split word into words it contains divided by underscores.
+     *
+     * @param underScored    word to split at underscores
+     * @return words contained in given word
+     */
+    public static String[] getWordsFromUnderscored(String underScored) {
+        return underScored.split("_");
+    }
+
+    // get synonyms
+    public static String[] getSynonymsForWord(String word) {
+        ArrayList<String> synonyms = new ArrayList<String>();
+
+        for (POS pos : RELEVANT_POS) {
+            synonyms.addAll(wn.getAllSynonyms(pos, word));
+        }
+        return synonyms.toArray(new String[synonyms.size()]);
+    }
+
+    public static void main(String[] args) {
+        for (String s : getWordsFromCamelCase("thisIsAClassWith1Name123")) {
+            System.out.println(s);
+            for (String w : getSynonymsForWord(s)) {
+                System.out.println(" --> " + w);
+            }
+        }
+    }
+}

This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.