write computeEntropy

1 files changed, 20 insertions, 1 deletions

diff --git a/newword.py b/newword.py
index 434a27f..50ee8aa 100644
--- a/newword.py
+++ b/newword.py
@@ -1,8 +1,9 @@
-#!/usr/bin/sqlite3
+#!/usr/bin/python3
 import os
 import os.path
 import sqlite3
 from argparse import ArgumentParser
+from math import log
 import utils
 from myconfig import MyConfig
 from dirwalk import walkIndex
@@ -107,6 +108,24 @@ def populateBigramSqlite(workdir):
 
 
 ############################################################
+#             Information Entropy Model                    #
+############################################################
+
+def computeEntropy(freqs):
+    print(freqs)
+
+    totalfreq = sum(freqs)
+    freqs = [ freq / float(totalfreq) for freq in freqs ]
+    assert 1 == sum(freqs)
+
+    entropy = sum([ - freq * log(freq) for freq in freqs ])
+    print(entropy)
+    return entropy
+
+
+
+
+############################################################
 #                Get Threshold Pass                        #
 ############################################################