1 files changed, 20 insertions, 0 deletions
diff --git a/tf.py b/tf.py
new file mode 100755
index 0000000..fc57f8b
--- /dev/null
+++ b/tf.py
@@ -0,0 +1,20 @@
+#!/usr/bin/env python3
+
+import os
+import json
+from collections import Counter
+
+if __name__ == '__main__':
+    store = dict()
+    for filename in os.listdir('information-retrieval-data/'):
+        with open('information-retrieval-data/' + filename) as f:
+            entity = json.load(f)
+        for field, values in entity.items():
+            for value in values:
+                if field not in store:
+                    store[field] = []
+                store[field] += [v.lower() for v in value.split(" ")]
+    for field in store:
+        cnt = Counter(store[field])
+        for term in cnt.items():
+            print('{}\t{}\t{}'.format(field, term[0], term[1]))