logging comparison vec size

dice-group · Oct 10, 2018 · 30fe9a0 · 30fe9a0
1 parent 5e04aac
commit 30fe9a0
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 4 deletions.
diff --git a/src/main/java/org/aksw/word2vecrestful/word2vec/W2VNrmlMemModelBinSrch.java b/src/main/java/org/aksw/word2vecrestful/word2vec/W2VNrmlMemModelBinSrch.java
@@ -193,7 +193,9 @@ private String getClosestEntry(float[] vector, String subKey) {
 			}
 			tl.printTime(1, "Setting Bits");
 			tl.logTime(1);
-			int[] nearbyIndexes = new int[finBitSet.cardinality()];
+			int nearbyWordsCount = finBitSet.cardinality();
+			LOG.info("Number of nearby words: "+nearbyWordsCount);
+			int[] nearbyIndexes = new int[nearbyWordsCount];
 			int j = 0;
 			for (int i = finBitSet.nextSetBit(0); i >= 0; i = finBitSet.nextSetBit(i + 1), j++) {
 				// operate on index i here

diff --git a/src/main/java/org/aksw/word2vecrestful/word2vec/W2VNrmlMemModelKMeans.java b/src/main/java/org/aksw/word2vecrestful/word2vec/W2VNrmlMemModelKMeans.java
@@ -198,7 +198,9 @@ private String getClosestEntry(float[] vector, String subKey) {
 			}
 			tl.printTime(1, "Setting Bits");
 			tl.logTime(1);
-			int[] nearbyIndexes = new int[finBitSet.cardinality()];
+			int nearbyWordsCount = finBitSet.cardinality();
+			LOG.info("Number of nearby words: "+nearbyWordsCount);
+			int[] nearbyIndexes = new int[nearbyWordsCount];
 			int j = 0;
 			for (int i = finBitSet.nextSetBit(0); i >= 0; i = finBitSet.nextSetBit(i + 1), j++) {
 				// operate on index i here

diff --git a/src/test/java/org/aksw/word2vecrestful/NrmlzdThetaMdlPrfmncTester.java b/src/test/java/org/aksw/word2vecrestful/NrmlzdThetaMdlPrfmncTester.java
@@ -5,6 +5,7 @@
 import java.util.List;
 
 import org.aksw.word2vecrestful.utils.Cfg;
+import org.aksw.word2vecrestful.utils.Word2VecMath;
 import org.aksw.word2vecrestful.word2vec.W2VNrmlMemModelKMeans;
 import org.aksw.word2vecrestful.word2vec.Word2VecFactory;
 import org.aksw.word2vecrestful.word2vec.Word2VecModel;
@@ -81,12 +82,18 @@ public void testNbmTime() throws IOException {
 			totTime += diff;
 			LOG.info("Query time recorded for Centroid " + (i + 1) + " is " + diff + " milliseconds.");
 		}
-		LOG.info(
-				"Average query time for W2VNrmlMemModelKMeans is : " + (totTime / centroids.length) + " milliseconds");
+		LOG.info("Average query time for W2VNrmlMemModelKMeans is : " + (totTime / centroids.length) + " milliseconds");
 		LOG.info("Predicted Words are :" + lrModelWords);
 		float percVal = NrmlzdMdlPrfmncTester.calcPercScore(correctWords, lrModelWords);
 		LOG.info("Score for Test is : " + percVal + "%");
 		lrModelWords.clear();
+
+		String word1 = "By_Jonas_Elmerraji";
+		String word2 = "%_#F########_3v.jsn";
+		float[] word1Vec = nbm.word2vec.get("By_Jonas_Elmerraji");
+		float[] word2Vec = nbm.word2vec.get("%_#F########_3v.jsn");
+		LOG.info("Cosine Similarity between " + word1 + " & " + word2 + " is : "
+				+ Word2VecMath.cosineSimilarity(word1Vec, word2Vec));
 	}
 
 	private static float[][] fetchWordsVec(String[] words, Word2VecModel nbm) {