dl-learner-svn

[DL-Learner SVN] SF.net SVN: dl-learner:[4177] trunk/components-core/src/main/java/org/ dllearner/algorithms/isle/index/TokenTree.java

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

Revision: 4177
          http://sourceforge.net/p/dl-learner/code/4177
Author:   dfleischhacker
Date:     2013-12-02 12:51:30 +0000 (Mon, 02 Dec 2013)
Log Message:
-----------
Add TokenTree class

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

Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java	                        (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java	2013-12-02 12:51:30 UTC (rev 4177)
@@ -0,0 +1,111 @@
+package org.dllearner.algorithms.isle.index;
+
+import org.dllearner.core.owl.Entity;
+
+import java.util.*;
+
+/**
+ * Tree for finding longest matching Token sequence
+ *
+ * @author Daniel Fleischhacker
+ */
+public class TokenTree {
+    private HashMap<Token, TokenTree> children;
+    private Set<Entity> entities;
+
+    public TokenTree() {
+        this.children = new HashMap<>();
+        this.entities = new HashSet<>();
+    }
+
+    /**
+     * Adds all given entities to the end of the path resulting from the given tokens.
+     *
+     * @param tokens   tokens to locate insertion point for entities
+     * @param entities entities to add
+     */
+    public void add(List<Token> tokens, Set<Entity> entities) {
+        TokenTree curNode = this;
+        for (Token t : tokens) {
+            TokenTree nextNode = curNode.children.get(t);
+            if (nextNode == null) {
+                nextNode = new TokenTree();
+                curNode.children.put(t, nextNode);
+            }
+            curNode = nextNode;
+        }
+        curNode.entities.addAll(entities);
+    }
+
+    /**
+     * Adds the given entity to the tree.
+     *
+     * @param tokens tokens to locate insertion point for entities
+     * @param entity entity to add
+     */
+    public void add(List<Token> tokens, Entity entity) {
+        add(tokens, Collections.singleton(entity));
+    }
+
+    /**
+     * Returns the set of entities located by the given list of tokens.
+     *
+     * @param tokens tokens to locate the information to get
+     * @return located set of entities or null if token sequence not contained in tree
+     */
+    public Set<Entity> get(List<Token> tokens) {
+        TokenTree curNode = this;
+        for (Token t : tokens) {
+            TokenTree nextNode = curNode.children.get(t);
+            if (nextNode == null) {
+                return null;
+            }
+            curNode = nextNode;
+        }
+        return curNode.entities;
+    }
+
+    /**
+     * Returns the list of tokens which are the longest match with entities assigned in this tree.
+     *
+     * @param tokens list of tokens to check for longest match
+     * @return list of tokens being the longest match, sublist of {@code tokens} anchored at the first token
+     */
+    public List<Token> getLongestMatch(List<Token> tokens) {
+        List<Token> fallbackTokenList = new ArrayList<>();
+        TokenTree curNode = this;
+
+        for (Token t : tokens) {
+            TokenTree nextNode = curNode.children.get(t);
+            if (nextNode == null) {
+                return fallbackTokenList;
+            }
+            curNode = nextNode;
+            fallbackTokenList.add(t);
+        }
+        return fallbackTokenList;
+    }
+
+    /**
+     * Returns the set of entities assigned to the longest matching token subsequence of the given token sequence.
+     * @param tokens    token sequence to search for longest match
+     * @return set of entities assigned to the longest matching token subsequence of the given token sequence
+     */
+    public Set<Entity> getEntitiesForLongestMatch(List<Token> tokens) {
+        TokenTree fallback = this.entities.isEmpty() ? null : this;
+        TokenTree curNode = this;
+
+        for (Token t : tokens) {
+            TokenTree nextNode = curNode.children.get(t);
+            if (nextNode == null) {
+                return fallback == null ? null : fallback.entities;
+            }
+            curNode = nextNode;
+            if (!curNode.entities.isEmpty()) {
+                fallback = curNode;
+            }
+        }
+
+        return fallback.entities;
+    }
+}

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

[DL-Learner SVN] SF.net SVN: dl-learner:[4181] trunk/components-core/src/main/java/org/ dllearner/algorithms/isle/index/TokenTree.java

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

Revision: 4181
          http://sourceforge.net/p/dl-learner/code/4181
Author:   dfleischhacker
Date:     2013-12-02 14:34:42 +0000 (Mon, 02 Dec 2013)
Log Message:
-----------
Add toString to TokenTree

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

Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java	2013-12-02 14:30:18 UTC (rev 4180)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java	2013-12-02 14:34:42 UTC (rev 4181)
@@ -1,11 +1,10 @@
 package org.dllearner.algorithms.isle.index;
 
+import com.google.common.base.Splitter;
+import com.google.common.collect.Lists;
 import org.dllearner.core.owl.Entity;
 import org.dllearner.core.owl.NamedClass;
 
-import com.google.common.base.Splitter;
-import com.google.common.collect.Lists;
-
 import java.util.*;
 
 /**
@@ -112,7 +111,7 @@
 
         return fallback.entities;
     }
-    
+
     public static void main(String[] args) throws Exception {
     	List<Token> tokens1 = Lists.newLinkedList();
     	for (String s : Splitter.on(" ").split("this is a token tree")) {
@@ -127,5 +126,28 @@
 		TokenTree tree = new TokenTree();
 		tree.add(tokens1, new NamedClass("TokenTree"));
 		tree.add(tokens2, new NamedClass("TokenizedTree"));
-	}
+        System.out.println(tree);
+    }
+
+    
+    public String toString() {
+        return "TokenTree\n"+ toString(0);
+    }
+
+    public String toString(int indent) {
+        StringBuilder indentStringBuilder = new StringBuilder();
+        for (int i = 0; i < indent; i++) {
+            indentStringBuilder.append(" ");
+        }
+        String indentString = indentStringBuilder.toString();
+        StringBuilder sb = new StringBuilder();
+        for (Map.Entry<Token, TokenTree> e : children.entrySet()) {
+            sb.append(indentString).append(e.getKey().toString());
+            sb.append("\n");
+            sb.append(e.getValue().toString(indent + 1));
+        }
+        return sb.toString();
+    }
+
+    
 }

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

[DL-Learner SVN] SF.net SVN: dl-learner:[4182] trunk/components-core/src/main/java/org/ dllearner/algorithms/isle/index/TokenTree.java

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

Revision: 4182
          http://sourceforge.net/p/dl-learner/code/4182
Author:   dfleischhacker
Date:     2013-12-02 14:41:21 +0000 (Mon, 02 Dec 2013)
Log Message:
-----------
Prevent possible NPE

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

Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java	2013-12-02 14:34:42 UTC (rev 4181)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java	2013-12-02 14:41:21 UTC (rev 4182)
@@ -109,7 +109,7 @@
             }
         }
 
-        return fallback.entities;
+        return fallback == null ? Collections.<Entity>emptySet() : fallback.entities;
     }
 
     public static void main(String[] args) throws Exception {

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

[DL-Learner SVN] SF.net SVN: dl-learner:[4201] trunk/components-core/src/main/java/org/ dllearner/algorithms/isle/index/TokenTree.java

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

Revision: 4201
          http://sourceforge.net/p/dl-learner/code/4201
Author:   dfleischhacker
Date:     2013-12-10 09:56:55 +0000 (Tue, 10 Dec 2013)
Log Message:
-----------
Retrieve entities from all possible leaf nodes in the TokenTree

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

Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java	2013-12-09 15:38:09 UTC (rev 4200)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TokenTree.java	2013-12-10 09:56:55 UTC (rev 4201)
@@ -81,7 +81,7 @@
     public Set<Entity> get(List<Token> tokens) {
         TokenTree curNode = this;
         for (Token t : tokens) {
-            TokenTree nextNode = curNode.children.get(t);
+            TokenTree nextNode = getNextTokenTree(curNode, t);
             if (nextNode == null) {
                 return null;
             }
@@ -90,6 +90,25 @@
         return curNode.entities;
     }
 
+    public Set<Entity> getAllEntities(List<Token> tokens) {
+        HashSet<Entity> resEntities = new HashSet<>();
+        getAllEntitiesRec(tokens, 0, this, resEntities);
+        return resEntities;
+    }
+
+    public void getAllEntitiesRec(List<Token> tokens, int curPosition, TokenTree curTree, HashSet<Entity> resEntities) {
+        if (curPosition == tokens.size()) {
+            resEntities.addAll(curTree.entities);
+            return;
+        }
+        Token t = tokens.get(curPosition);
+        for (Map.Entry<Token, TokenTree> entry : curTree.children.entrySet()) {
+            if (t.equalsWithAlternativeForms(entry.getKey())) {
+                getAllEntitiesRec(tokens, curPosition + 1, entry.getValue(), resEntities);
+            }
+        }
+    }
+
     /**
      * Returns the list of tokens which are the longest match with entities assigned in this tree.
      *
@@ -148,7 +167,7 @@
     }
 
     /**
-     * Returns the original token for the longest match
+     * Returns the original ontology tokens for the longest match
      */
     public List<Token> getOriginalTokensForLongestMatch(List<Token> tokens) {
         TokenTree fallback = this.entities.isEmpty() ? null : this;

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