Update entity marking

collect_entity.py

@@ -3,16 +3,10 @@ import re
 import os
 
 
-def extract_movie_ids(text):
-    res = []
-    for m in re.finditer(r'@(\d+)', text):
-        res.append(m.group(1))
-    return res
-
 def collect_entity_mappings():
     files = [
-        "./test_data_dbpedia.jsonl",
-        "./train_data_dbpedia.jsonl",
+        "test_data_dbpedia.jsonl",
+        "train_data_dbpedia.jsonl",
         "valid_data_dbpedia.jsonl"
         ]
     entityName2entity = {}
@@ -20,30 +14,21 @@ def collect_entity_mappings():
     with open('entity2id.json', 'r', encoding='utf-8') as f:
         entity2id = json.load(f)
 
-    movieId2entity = {}
     for path in files:
         with open(path, 'r') as f:
             for line in f:
                 dialog = json.loads(line)
-
                 for message in dialog["messages"]:
-                    movieIds = extract_movie_ids(message["text"])
                     for name, entity in zip(message['entity_name'], message['entity']):
                         entityName2entity[name] = entity
                     for movie_name, entity in zip(message['movie_name'], message['movie']):
                         entityName2entity[movie_name] = entity
-                    for movieId, entity in zip(movieIds, message['movie']):
-                        movieId2entity[movieId] = entity
+
 
     entityName2id = {k: entity2id[entityName2entity[k]] for k in entityName2entity}
     with open('entityName2id.json', 'w') as f:
         json.dump(entityName2id, f)
 
-    movieId2id = {k: entity2id[movieId2entity[k]] for k in movieId2entity}
-    with open('movieId2id.json', 'w') as f:
-        json.dump(movieId2id, f)
-
-
 
 if __name__ == "__main__":
     collect_entity_mappings()

redial_unicrs.py

@@ -4,6 +4,22 @@ from typing import List
 import html
 import datasets
 
+ENTITY = 'entity'
+ENTITY_PATTERN = r'<entity> {} </entity>'
+
+
+def markup_entity(utt: str, entities: List[str]):
+    # If entities like "action movie" and "action" appear at the same time, we only mark the longer one
+    entities = sorted(list(set(entities)), key=lambda x: len(x), reverse=True)
+    for i, entity in enumerate(entities):
+        valid = entity not in ENTITY
+        for prev in entities[:i]:
+            if entity in prev:
+                valid = False
+        if valid:
+            utt = re.sub(entity, ENTITY_PATTERN.format(entity), utt)
+    return utt
+
 logger = datasets.logging.get_logger(__name__)
 
 
@@ -13,7 +29,6 @@ class RedialConfig(datasets.BuilderConfig):
     def __init__(self, features,
                  initiator_prefix='User: ',
                  respondent_prefix='System: ',
-                 separator='<sep>',
                  **kwargs):
         """BuilderConfig for ReDIAL (used in UniCRS).
 
@@ -26,7 +41,6 @@ class RedialConfig(datasets.BuilderConfig):
         self.features = features
         self.initiator_prefix = initiator_prefix
         self.respondent_prefix = respondent_prefix
-        self.separator = separator
 
 _URL = "./"
 _URLS = {
@@ -97,7 +111,7 @@ class ReDIAL(datasets.GeneratorBasedBuilder):
                     return '<movie>'
                 movie_name = movieid2name[movieid]
                 movie_name = ' '.join(movie_name.split())
-                return self.entity_pattern.format(movie_name)
+                return movie_name
             else:
                 return match.group(0)
 
@@ -108,24 +122,8 @@ class ReDIAL(datasets.GeneratorBasedBuilder):
 
         return utt
 
-    def _mark_entity(self, utt: str, entities: List[str]):
-        # If entities like "action movie" and "action" appear at the same time, we only mark the longer one
-        entities = sorted(list(set(entities)), key=lambda x: len(x), reverse=True)
-        for i, entity in enumerate(entities):
-            valid = True
-            for prev in entities[:i]:
-                if entity in prev:
-                    valid = False
-            if valid:
-                utt = re.sub(entity, self.entity_pattern.format(entity), utt)
-        return utt
-
     movie_pattern = re.compile(r'@(\d+)')
     default_movie_entity = '<movie>'
-    entity_pattern = "<entity>{}</entity>"
-    def map_movie_pattern_to_entity(self, id2entity, id: str):
-        entity = id2entity[id] if id in id2entity else self.default_movie_entity
-        return self.entity_pattern.format_map(entity)
 
     def _generate_examples(self, filepath, entity2id):
         """This function returns the examples in the raw (text) form."""
@@ -154,7 +152,7 @@ class ReDIAL(datasets.GeneratorBasedBuilder):
                         turn_j = turn_i
                         while turn_j < len(messages) and messages[turn_j]['senderWorkerId'] == worker_id:
                             utt = self._process_utt(messages[turn_j]['text'], movieid2name, replace_movieId=True)
-                            utt = self._mark_entity(utt, messages[turn_j]['entity_name'])
+                            utt = markup_entity(utt, messages[turn_j]['entity_name']+messages[turn_j]['movie_name'])
                             utt_turn.append(utt)
                             movie_ids = [entity2id[movie] for movie in messages[turn_j]['movie'] if
                                          movie in entity2id]