removed huggingface req

scoring-scripts/compute_MCC.py

@@ -1,19 +1,30 @@
 from sklearn.metrics import matthews_corrcoef
 import numpy as np
-def compute_MCC(references_dataset, predictions_dataset, ref_col='ner_tags', pred_col='pred_ner_tags'):
-    # computes the Matthews correlation coeff between two datasets
-    
-    # sort by id
-    references_dataset = references_dataset.sort('unique_id')
-    predictions_dataset = predictions_dataset.sort('unique_id')
-    
+def compute_MCC_jsonl(references_jsonl, predictions_jsonl, ref_col='ner_tags', pred_col='pred_ner_tags'):
+    '''
+    Computes the Matthews correlation coeff between two datasets in jsonl format (list of dicts each with same keys).
+    Sorts the datasets by 'unique_id' and verifies that the tokens match.
+    '''
+    # reverse the dict
+    ref_dict = {k:[e[k] for e in references_jsonl] for k in references_jsonl[0].keys()}
+    pred_dict = {k:[e[k] for e in predictions_jsonl] for k in predictions_jsonl[0].keys()}
+
+    # sort by unique_id
+    ref_idx = np.argsort(ref_dict['unique_id'])
+    pred_idx = np.argsort(pred_dict['unique_id'])
+    ref_ner_tags = np.array(ref_dict[ref_col], dtype=object)[ref_idx]
+    pred_ner_tags = np.array(pred_dict[pred_col], dtype=object)[pred_idx]
+    ref_tokens = np.array(ref_dict['tokens'], dtype=object)[ref_idx]
+    pred_tokens = np.array(pred_dict['tokens'], dtype=object)[pred_idx]
+
     # check that tokens match
-    assert(references_dataset['tokens']==predictions_dataset['tokens'])
-    
+    for t1,t2 in zip(ref_tokens, pred_tokens):
+        assert(t1==t2)
+
     # the lists have to be flattened
-    flat_ref_tags = np.concatenate(references_dataset[ref_col])
-    flat_pred_tags = np.concatenate(predictions_dataset[pred_col])
-    
+    flat_ref_tags = np.concatenate(ref_ner_tags)
+    flat_pred_tags = np.concatenate(pred_ner_tags)
+
     mcc_score = matthews_corrcoef(y_true=flat_ref_tags,
                                   y_pred=flat_pred_tags)