Merge pull request #722 from swicaksono/master

added djaccard similiarity algorithm

Author: probot-auto-merge[bot]

Python/djaccard_simmiliarity.py

@@ -0,0 +1,37 @@
+'''
+Given a list of sentence with same topics (or from a news), this function will return 
+a text which have score similiar between sentences. It will be useful to find the main sentence.
+'''
+import numpy as np
+def jaccard_similarity(str_1, str_2):
+        """ compute of intersection to get similraity score between words """
+        str_1 = set(str_1.split())
+        str_2 = set(str_2.split())
+        intersect = str_1.intersection(str_2)
+        return float(len(intersect)) / (len(str_1) + len(str_2) - len(intersect))
+
+def max_avg_jaccard_sim(sentence_list, show_avg=False):
+    """ compute of intersection each sentence in cluster, and return sentence with maximum of average similarity score between sentence """
+    sim = []
+    text_avg_sim = {}
+    for idx in range(len(sentence_list)):
+        for text in sentence_list:
+            if len(text) < 2:
+                continue
+            similarity = jaccard_similarity(sentence_list[idx], text)
+            sim.append(similarity)
+            text_avg_sim[sentence_list[idx]] = sum(sim) / len(sim)
+
+    # key of max values
+    if show_avg:
+        return max(text_avg_sim, key=text_avg_sim.get), max(text_avg_sim.values())
+    else:
+        return max(text_avg_sim, key=text_avg_sim.get)
+
+
+if __name__ == "__main__":
+    sentences = ["Manchester United midfielder Scott McTominay secured the Man-of-the Match award after the Reds’ 3-1 win at Norwich City.",
+                "The 21-year-old took home 50 per cent of your vote after netting our 21st-minute opener – which was our 2,000th strike in Premier League history.",
+                "The Scotland international said he was delighted with the team's performance, as well as securing his place in the record books."]
+    main_sent = max_avg_jaccard_sim(sentence_list=sentences)
+    print (main_sent)