17.

wesenberg/SmatchClass.py

@@ -32,7 +32,7 @@ def calc_smatch_to_r(name, source, summary, graph_source, graph_summary):
 
     smatch_pred = calc_one_smatch(graph_source=graph_source, graph_summary=graph_summary)
 
-    if smatch_pred[0] == 0:
+    if smatch_pred[0] == 0 and smatch_pred[1] != 0:
         safe_smatch_error(name=name + "_null", one=source, two=summary, recall=0, precision=None, smatch_pred=smatch_pred,
                           graph_source=graph_source, graph_summary=graph_summary)
 

wesenberg/util.py

@@ -20,7 +20,7 @@ from wesenberg.Konstanten import LIST_SYMBOLS_TO_DELETE, ROOT_PATH, AMRLIB_PATH,
 
 def remove_bugs(sentence):
     sentence = remove_all_hash_for_amr(sentence)
-    # replace with "  " to find and between two removed chars
+    # replace with "  " to find "  and  " between two removed chars
     sentence = remove_single_char_for_amr(sentence, "  ")
     sentence = remove_minus_for_amr(sentence, "  ")
     sentence = remove_and_for_amr(sentence)
@@ -38,7 +38,9 @@ def calc_one_amr(nlp, sentence, file_name):
         else:
             output = strip_amr_for_smatch(graphs[0])
     except IndexError:
-        return ""
+        output = ""
+        safe_amr_error(file_name + "_empty", "", output, sentence)
+
 
     if SAFE_AMR_ERROR:
         output_list = output.split("\n")
@@ -48,15 +50,21 @@ def calc_one_amr(nlp, sentence, file_name):
             if item not in search_list:
                 search_list.append(item)
             else:
-                path = os.path.join(AMR_ERROR_PATH)
-                if not os.path.exists(path):
-                    os.makedirs(path)
-                with open(path + file_name + ".txt", 'a') as txt:
-                    txt.write(sentence + "\n" + item + "\n" + output + "\n\n")
+                safe_amr_error(file_name, item, output, sentence)
                 break
     return output
 
 
+def safe_amr_error(file_name, item, output, sentence):
+    path = os.path.join(AMR_ERROR_PATH)
+    if not os.path.exists(path):
+        os.makedirs(path)
+    if item != "":
+        item = "\n" + item
+    with open(path + file_name + ".txt", 'a') as txt:
+        txt.write(sentence + item + "\n" + output + "\n\n")
+
+
 def init_amr(stog_name):
     amrlib.setup_spacy_extension()
     nlp = spacy.load(LOAD_SPARCY)
@@ -333,7 +341,8 @@ def calc_f_score_smatch(name, one, two, smatch_pred, graph_source, graph_summary
 def calc_precision_smatch(name, one, two, smatch_pred, graph_source, graph_summary):
     precision = smatch_pred[0] / smatch_pred[1]
     if 1 < precision:
-        safe_smatch_error(name, one, two, None, precision, smatch_pred, graph_source, graph_summary)
+        safe_smatch_error(name=name, one=one, two=two, result=precision, smatch_pred=smatch_pred,
+                          graph_source=graph_source, graph_summary=graph_summary)
         precision = 1
     return precision
 
@@ -341,27 +350,21 @@ def calc_precision_smatch(name, one, two, smatch_pred, graph_source, graph_summa
 def calc_recall_smatch(name, one, two, smatch_pred, graph_source, graph_summary):
     recall = smatch_pred[0] / smatch_pred[2]
     if 1 < recall:
-        safe_smatch_error(name=name, one=one, two=two, recall=recall, precision=None, smatch_pred=smatch_pred,
+        safe_smatch_error(name=name, one=one, two=two, result=recall, smatch_pred=smatch_pred,
                           graph_source=graph_source, graph_summary=graph_summary)
         recall = 1
     return recall
 
 
-def safe_smatch_error(name, one, two, recall, precision, smatch_pred, graph_source, graph_summary):
+def safe_smatch_error(name, one, two, result, smatch_pred, graph_source, graph_summary):
     if SAFE_SMATCH_ERROR:
         path = os.path.join(BACKUP_SMATCH_ERROR_PATH)
         if not os.path.exists(path):
             os.makedirs(path)
         with open(path + name + ".txt", 'a') as txt:
-            if recall is not None:
-                txt.write(
-                    "one:\t\t" + one + "\nTwo:\t\t" + two + "\nRecall:\t\t" + str(
-                        recall) + "\nSmatch:\t\t" + str(
-                        smatch_pred) + "\nSource:\n" + graph_source + "\nSummary:\n" + graph_summary + "\n\n")
-            if precision is not None:
                 txt.write(
-                    "one:\t\t" + one + "\ntwo:\t\t" + two + "\nPrecision:\t" + str(
-                        precision) + "\nSmatch:\t\t" + str(
+                    "one:\t\t" + one + "\nTwo:\t\t" + two + "\nResult:\t\t" + str(
+                        result) + "\nSmatch:\t\t" + str(
                         smatch_pred) + "\nSource:\n" + graph_source + "\nSummary:\n" + graph_summary + "\n\n")