Refactor evaluation a bit

evaluation.py

@@ -15,7 +15,11 @@ def evaluate(timeline, gold_timelines):
     timeline = tilse_format(timeline)
     ground_truth = timelines.GroundTruth([tilse_format(gold_timeline) for gold_timeline in gold_timelines])
 
-    return evaluator.evaluate_align_date_content_costs_many_to_one(timeline, ground_truth)
+    eval_result = evaluator.evaluate_align_date_content_costs_many_to_one(timeline, ground_truth)
+    ar1_f = eval_result['rouge_1']['f_score']
+    ar2_f = eval_result['rouge_2']['f_score']
+
+    return ar1_f, ar2_f
 
 
 def date_f1(timeline, gold_timeline):

run.py

@@ -17,9 +17,7 @@ def main():
             print(f'Topic {topic}, gold timeline {gold_timeline_name}')
             timeline = timeline_generation.make_timeline(articles, gold_timeline, keywords)
             timeline_generation.print_timeline(timeline)
-            eval_result = evaluation.evaluate(timeline, [gold_timeline])
-            ar1_f = eval_result['rouge_1']['f_score']
-            ar2_f = eval_result['rouge_2']['f_score']
+            ar1_f, ar2_f = evaluation.evaluate(timeline, [gold_timeline])
             date_f1 = evaluation.date_f1(timeline, gold_timeline)
             print(f'{ar1_f},{ar2_f},{date_f1}')
             break