AssociationRuleRecommender time complexity reduced(by Mr No2)

core/src/main/java/net/librec/recommender/ext/AssociationRuleRecommender.java

@@ -17,17 +17,12 @@
  */
 package net.librec.recommender.ext;
 
-import com.google.common.cache.LoadingCache;
 import com.google.common.collect.HashBasedTable;
 import com.google.common.collect.Table;
 import net.librec.common.LibrecException;
-import net.librec.math.structure.SparseVector;
-import net.librec.math.structure.VectorEntry;
+import net.librec.math.structure.*;
 import net.librec.recommender.AbstractRecommender;
 
-import java.util.Map;
-import java.util.concurrent.ExecutionException;
-
 /**
  * Choonho Kim and Juntae Kim, <strong>A Recommendation Algorithm Using Multi-Level Association Rules</strong>, WI 2003.
  * <p>
@@ -42,42 +37,52 @@ public class AssociationRuleRecommender extends AbstractRecommender {
     /**
      * confidence matrix of association rules
      */
-    private Table<Integer, Integer, Double> associationTable;
+    private SparseMatrix associations;
 
-    /**
-     * user-vector cache, item-vector cache
-     */
-    protected LoadingCache<Integer, SparseVector> userCache;
+    private DenseMatrix userItemRates;
 
-    /**
-     * Guava cache configuration
-     */
-    protected static String cacheSpec;
+    // /**
+    // * user-vector cache, item-vector cache
+    // */
+    // protected LoadingCache<Integer, SparseVector> userCache;
+
+    // /**
+    // * Guava cache configuration
+    // */
+    // protected static String cacheSpec;
 
     /**
      * setup
      *
-     * @throws LibrecException if error occurs
+     * @throws LibrecException
+     *             if error occurs
      */
     @Override
     protected void setup() throws LibrecException {
         super.setup();
-        cacheSpec = conf.get("guava.cache.spec", "maximumSize=200,expireAfterAccess=2m");
+        // cacheSpec = conf.get("guava.cache.spec",
+        // "maximumSize=200,expireAfterAccess=2m");
 
-        associationTable = HashBasedTable.create(numItems, numItems);
-        userCache = trainMatrix.rowCache(cacheSpec);
+        userItemRates = new DenseMatrix(trainMatrix.numRows, trainMatrix.numColumns);
+        for (MatrixEntry entry : trainMatrix) {
+            userItemRates.set(entry.row(), entry.column(), entry.get());
+        }
+        // userCache = trainMatrix.rowCache(cacheSpec);
     }
 
     @Override
     protected void trainModel() throws LibrecException {
-        // simple rule: X => Y, given that each user vector is regarded as a transaction
+        // simple rule: X => Y, given that each user vector is regarded as a
+        // transaction
+        Table<Integer, Integer, Double> associationTable = HashBasedTable.create(numItems, numItems);
         for (int itemIdx = 0; itemIdx < numItems; itemIdx++) {
             // all transactions for item itemIdx
             SparseVector userRatingsVector = trainMatrix.column(itemIdx);
             int userCount = userRatingsVector.size();
 
             for (int assoItemIdx = 0; assoItemIdx < numItems; assoItemIdx++) {
-                // compute confidence where containing item assoItemIdx among userRatingsVector
+                // compute confidence where containing item assoItemIdx among
+                // userRatingsVector
                 int count = 0;
                 for (VectorEntry vectorEntry : userRatingsVector) {
                     int userIdx = vectorEntry.index();
@@ -92,30 +97,29 @@ protected void trainModel() throws LibrecException {
                 }
             }
         }
+        associations = new SparseMatrix(numItems, numItems, associationTable);
     }
 
     /**
      * predict a specific rating for user userIdx on item itemIdx.
      *
-     * @param userIdx user index
-     * @param itemIdx item index
+     * @param userIdx
+     *            user index
+     * @param itemIdx
+     *            item index
      * @return predictive rating for user userIdx on item itemIdx
-     * @throws LibrecException if error occurs
+     * @throws LibrecException
+     *             if error occurs
      */
     @Override
     protected double predict(int userIdx, int itemIdx) throws LibrecException {
-        SparseVector itemRatingsVector = null;
-        try {
-            itemRatingsVector = userCache.get(userIdx);
-        } catch (ExecutionException e) {
-            e.printStackTrace();
-        }
 
         double predictRatings = 0;
-        for (Map.Entry<Integer, Double> mapEntry : associationTable.column(itemIdx).entrySet()) {
-            int assoItemIdx = mapEntry.getKey();
-            double support = mapEntry.getValue();
-            predictRatings += itemRatingsVector.get(assoItemIdx) * support;
+        for (VectorEntry mapEntry : associations.column(itemIdx)) {
+            int assoItemIdx = mapEntry.index();
+            double support = mapEntry.get();
+            double value = userItemRates.get(userIdx, assoItemIdx);
+            predictRatings += value * support;
         }
 
         return predictRatings;