save

ex2/2.py

@@ -10,6 +10,21 @@
 exec(open('../functions.py').read())
 exec(open('priors.py').read())
 exec(open('weights.py').read())
+tt=0
+weights1 = 0
+
+def build_graph_matrix(sentences,weight_func,t):
+     if not callable(weight_func):
+         return 'Not functions!'
+     nsents = len(sentences)
+     weights = np.zeros([nsents,nsents])
+
+     cos_matrix = get_cosine_similarities_matrix(sentences)
+     #create weights
+     for i in range(len(sentences)):
+         for j in range(len(sentences)):
+             weights[i][j] = weight_func(i,j,sentences,cos_matrix,t)
+     return weights
 
 def build_priors(sentences,graph,prior_func):
 	if not callable(prior_func):
@@ -58,9 +73,10 @@ def get_top_n(array,n):
     return top
 
 def build_summary(sentences,prior_func,weight_func,t):
-
     weights = build_graph_matrix(sentences,weight_func,t)
+   # print(np.sum(weights))
     priors = build_priors(sentences,weights,prior_func)
+    #print(priors)
     ranks = rank(weights,priors,50,0.15)
     top = get_top_n(ranks,5)
     summary = ""
@@ -82,16 +98,22 @@ def build_summary(sentences,prior_func,weight_func,t):
 source_texts = os.listdir(source_path)
 
 for thresh in tvals:
+    tt=thresh
+    print('---',tt)
     MAP = 0
     for text_file in source_texts:
         with open(source_path + text_file,'r',encoding='Latin-1') as file: #source_path + text_file
             text = file.read()
         sentences = text_to_sentences(text)
-        summary = build_summary(sentences,degree_centrality_prior,uniform_weight,thresh)
-        with open(sums_path+ 'Ext-' + text_file,'r',encoding='Latin-1') as summary_file: #sums_path+ 'Ext-' + text_file
+        summary = build_summary(sentences,uniform_prior,cos_sim_weight,thresh)
+        with open(source_path + text_file,'r',encoding='Latin-1') as summary_file: #sums_path+ 'Ext-' + text_file    '../ex1/textsum.txt'
             MAP += AP(summary,summary_file.read())
-        MAP /= len(source_texts)
+            #print(MAP)
+    MAP /= len(source_texts)
+    print(MAP)
     results.append(MAP)
+   # if tt==0.05:
+   #     break
 print(results)
 max = np.argmax(results)
 print("Best summary with MAP =",results[max],' for threshold =',tvals[max])

ex2/priors.py

@@ -1,2 +1,8 @@
 def uniform_prior(sent_index,graph,sentences):
     return 1/len(sentences)
+
+#receives graph matrix for convenience...
+def degree_centrality_prior(sent_index,graph,sentences):
+	links = graph[sent_index]
+	nonzero = np.nonzero(links)[0]
+	return len(nonzero)/len(links)

ex2/weights.py

@@ -5,6 +5,7 @@
 
 exec(open('../functions.py').read())
 
+#----------------------weights--------------------------------#
 def is_edge(sent1_index,sent2_index,cosine_matrix,t):
 	return (cos_sim(sent1_index,sent2_index,cosine_matrix) >= t) and (sent1_index != sent2_index)
 
@@ -14,3 +15,9 @@ def uniform_weight(sent1_index,sent2_index,sentences,cosine_matrix,t):
 	else:
 		return 0
 
+def cos_sim_weight(sent1_index,sent2_index,sentences,cosine_matrix,t):
+	if is_edge(sent1_index,sent2_index,cosine_matrix,t):
+		return cos_sim(sent1_index,sent2_index,cosine_matrix)
+	else:
+		return 0
+

functions.py

@@ -42,45 +42,30 @@ def get_cosine_similarities_matrix(sentences):
 
 #receives cossim matrix for performance reasons
 #receives indexes for performance reasons aswell...
+#also doubles as a weight, ignore t and sentences
 def cos_sim(sent1_index,sent2_index,cosine_matrix):
     return cosine_matrix[sent1_index][sent2_index]
 
 def degree_centrality(sent_index,sentences,t=0.2):
-	graph = build_graph_matrix(sentences,t)
+	graph = build_graph_uniform(sentences,t)
 	return degree_centrality_prior(sent_index,graph,sentences)
 
 
 
-#----------------------weights--------------------------------#
-def is_edge(sent1_index,sent2_index,cosine_matrix,t):
-	return (cos_sim(sent1_index,sent2_index,cosine_matrix) >= t) and (sent1_index != sent2_index)
 
-def uniform_weight(sent1_index,sent2_index,sentences,cosine_matrix,t):
-	if is_edge(sent1_index,sent2_index,cosine_matrix,t):
-		return 1
-	else:
-		return 0
-
-#---------------------priors---------------------------------#
-#receives graph matrix for convenience...
-def degree_centrality_prior(sent_index,graph,sentences):
-	links = graph[sent_index]
-	nonzero = np.nonzero(links)[0]
-	return len(nonzero)/len(links)
 
 
 #--------------------------graph building stuff------------------------#
-def build_graph_matrix(sentences,weight_func=uniform_weight,t=0.2):
+def build_graph_uniform(sentences,t=0.2):
     if not callable(weight_func):
         return 'Not functions!'
     nsents = len(sentences)
     weights = np.zeros([nsents,nsents])
-
     cos_matrix = get_cosine_similarities_matrix(sentences)
     #create weights
     for i in range(len(sentences)):
         for j in range(len(sentences)):
-            weights[i][j] = weight_func(i,j,sentences,cos_matrix,t)
+            weights[i][j] = (cos_sim(sent1_index,sent2_index,cosine_matrix) >= t) and (sent1_index != sent2_index)
     return weights