32-bit version of minhash just in case someone cares

mrflip · May 8, 2009 · e358795 · e358795
1 parent ef05eb5
commit e358795
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 0 deletions.
diff --git a/lib/minhash.rb b/lib/minhash.rb
@@ -14,4 +14,13 @@ def self.minhash(history, seed)
     end
     val
   end
+
+  def self.minhash32(history, seed)
+    val = Infinity
+    Array(history).each do |item|
+      n = MurmurHash.murmur_hash(item, seed)
+      val = n if n < val
+    end
+    val
+  end
 end
diff --git a/test/test_minhash.rb b/test/test_minhash.rb
@@ -13,4 +13,15 @@ def test_minhash_works
     m = MinHash.minhash(["baz", "bar", "foo"], 23)
     assert_equal 1157430835185436796, m
   end
+
+  def test_minhash32_works
+    m = MinHash.minhash32(["foo", "bar", "baz"], 23)
+    assert_equal 568256043, m
+
+    m = MinHash.minhash32(["bar", "foo", "baz"], 23)
+    assert_equal 568256043, m
+
+    m = MinHash.minhash32(["baz", "bar", "foo"], 23)
+    assert_equal 568256043, m
+  end
 end