add sentenceindex list

Senseval_Prep/senseval3_prep.py

@@ -128,3 +128,23 @@ def read_pkl(path = "d000.pkl"):
     with open (path, "rb") as f:
         li = pkl.load(f)
         return li
+
+def get_sent_list(key = "d000"):
+    counter = 0
+    sent_li = []
+    sent_start = 0
+    data = read_pkl(key + ".pkl")
+    sent = data[0]
+    sent_end_bool = False
+    for elm in sent:
+        if elm[1] == "PUNCT":
+            if elm[0] == "." or elm[0] == "!" or elm[0] == "?":
+                sent_li.append([sent_start, counter])
+                sent_end_bool = True
+        elif sent_end_bool == True:
+            sent_start = counter
+            sent_end_bool = False
+        counter += 1
+    with open(key + "_sent.pkl", "wb") as f:
+        pkl.dump(sent_li, f)
+        print ("Done")