Merge remote-tracking branch 'origin/master'

NaturalLanguageAnalyzer.py

@@ -1,35 +1,28 @@
-import jsgf
+from convlab.base_models.t5.nlu import T5NLU
+import requests
+
+
+def translate_text(text, target_language='en'):
+    url = 'https://translate.googleapis.com/translate_a/single?client=gtx&sl=auto&tl={}&dt=t&q={}'.format(
+        target_language, text)
+    response = requests.get(url)
+    if response.status_code == 200:
+        translated_text = response.json()[0][0][0]
+        return translated_text
+    else:
+        return None
 
 
 class NaturalLanguageAnalyzer:
-    # def process(self, text):
-    #     user_act = None
-    #     if ("imie" in text or "imię" in text) and "?" in text:
-    #         user_act = "request(firstname)"
-    #     return user_act
-
     def process(self, text):
-        with open('grammar_1.jsgf', 'r', encoding='utf-8') as f:
-            content = f.read()
-        book_grammar = jsgf.parse_grammar_string(content)
-        matched = book_grammar.find_matching_rules(text)
-        if matched:
-            return self.get_dialog_act(matched[0])
-        else:
-            return {'act': 'null', 'slots': []}
+        # Inicjalizacja modelu NLU
+        model_name = "ConvLab/t5-small-nlu-multiwoz21"
+        nlu_model = T5NLU(speaker='user', context_window_size=0, model_name_or_path=model_name)
 
-    def get_slots(self, expansion, slots):
-        if expansion.tag != '':
-            slots.append((expansion.tag, expansion.current_match))
-            return
+        # Automatyczne tłumaczenie na język angielski
+        translated_input = translate_text(text)
 
-        for child in expansion.children:
-            self.get_slots(child, slots)
+        # Wygenerowanie odpowiedzi z modelu NLU
+        nlu_output = nlu_model.predict(translated_input)
 
-        if not expansion.children and isinstance(expansion, jsgf.NamedRuleRef):
-            self.get_slots(expansion.referenced_rule.expansion, slots)
-
-    def get_dialog_act(self, rule):
-        slots = []
-        self.get_slots(rule.expansion, slots)
-        return {'act': rule.grammar.name, 'slots': slots}
+        return nlu_output

archives/iobes_slot.py

@@ -0,0 +1,29 @@
+import pandas as pd
+
+# Wczytanie danych z pliku TSV
+data = pd.read_csv("combined_df.tsv", sep="\t")
+
+# Inicjalizacja pustej listy do przechowywania tagów IOBES
+tags = []
+
+# Iteracja po każdym wierszu danych
+for index, row in data.iterrows():
+    # Podział akcji na pojedyncze słowa
+    words = row['act'].split()
+    # Początkowy tag IOBES to 'O' dla każdego słowa
+    current_tags = ['O'] * len(words)
+    # Ustawienie tagu 'B' dla pierwszego słowa
+    current_tags[0] = 'B'
+    # Ustawienie tagu 'E' dla ostatniego słowa
+    current_tags[-1] = 'E'
+    # Ustawienie tagu 'I' dla pozostałych słów, jeśli są
+    if len(words) > 2:
+        current_tags[1:-1] = ['I'] * (len(words) - 2)
+    # Dodanie tagów do listy
+    tags.extend(current_tags)
+
+# Dodanie kolumny z tagami do danych
+data['tags'] = tags
+
+# Zapisanie danych z tagami do nowego pliku TSV
+data.to_csv("nazwa_pliku_z_tagami.tsv", sep="\t", index=False)

evaluate.py

@@ -1,5 +1,7 @@
 import os
 import pandas as pd
+import re
+
 from NaturalLanguageAnalyzer import NaturalLanguageAnalyzer
 
 data_directory = 'data'
@@ -14,9 +16,38 @@ for file_name in file_list:
 
 combined_df = pd.concat(dfs, ignore_index=True)
 
-for text, act in zip(combined_df["value"].values, combined_df["act"].values):
+change_act_format = {
+    "thankyou": "thank",
+    "bye": "thank",
+    "hello": "greet",
+    "inform": "inform",
+    "request": "request",
+    "reqmore": "request"
+}
+
+correct = 0
+incorrect = 0
+for text, ground_act in zip(combined_df["value"].values, combined_df["act"].values):
     nla = NaturalLanguageAnalyzer()
-    user_act = nla.process(text)
-    print(user_act)
-    print(act)
+    nla_output = nla.process(text)
+    predicted_act = set([i[0] for i in nla_output])
+
+    pattern = re.compile(r'([^(&]+)(?=\()')
+    matches = re.findall(pattern, ground_act)
+    ground_act_processed = set()
+    for match in matches:
+        if match in change_act_format:
+            ground_act_processed.add(change_act_format[match])
+
+    for i in ground_act_processed:
+        if i in predicted_act:
+            correct += 1
+        else:
+            incorrect += 1
+
+    print("Predicted:", predicted_act)
+    print("Ground truth:", ground_act_processed)
     print()
+
+accuracy = correct/(correct+incorrect)
+print("Accuracy: ", accuracy)