265 lines
8.4 KiB
Python
265 lines
8.4 KiB
Python
import jsgf
|
|
import codecs
|
|
from conllu import parse_incr
|
|
from tabulate import tabulate
|
|
import os.path
|
|
|
|
|
|
from flair.data import Corpus, Sentence, Token
|
|
from flair.datasets import SentenceDataset
|
|
from flair.embeddings import StackedEmbeddings
|
|
from flair.embeddings import WordEmbeddings
|
|
from flair.embeddings import CharacterEmbeddings
|
|
from flair.embeddings import FlairEmbeddings
|
|
from flair.models import SequenceTagger
|
|
from flair.trainers import ModelTrainer
|
|
|
|
import random
|
|
import torch
|
|
random.seed(42)
|
|
torch.manual_seed(42)
|
|
|
|
if torch.cuda.is_available():
|
|
torch.cuda.manual_seed(0)
|
|
torch.cuda.manual_seed_all(0)
|
|
torch.backends.cudnn.enabled = False
|
|
torch.backends.cudnn.benchmark = False
|
|
torch.backends.cudnn.deterministic = True
|
|
|
|
class ML_NLU:
|
|
def __init__(self, acts, arguments):
|
|
self.acts = acts
|
|
self.arguments = arguments
|
|
|
|
def nolabel2o(self, line, i):
|
|
return 'O' if line[i] == 'NoLabel' else line[i]
|
|
|
|
def conllu2flair(self, sentences, label=None):
|
|
fsentences = []
|
|
|
|
for sentence in sentences:
|
|
fsentence = Sentence()
|
|
|
|
for token in sentence:
|
|
ftoken = Token(token['form'])
|
|
|
|
if label:
|
|
ftoken.add_tag(label, token[label])
|
|
|
|
fsentence.add_token(ftoken)
|
|
|
|
fsentences.append(fsentence)
|
|
|
|
return SentenceDataset(fsentences)
|
|
|
|
|
|
def predict(self, model, sentence):
|
|
csentence = [{'form': word} for word in sentence]
|
|
fsentence = self.conllu2flair([csentence])[0]
|
|
model.predict(fsentence)
|
|
return [(token, ftoken.get_tag('slot').value) for token, ftoken in zip(sentence, fsentence)]
|
|
|
|
def setup(self):
|
|
|
|
if os.path.isfile('slot-model/final-model.pt'):
|
|
model = SequenceTagger.load('slot-model/final-model.pt')
|
|
else:
|
|
fields = ['id', 'form', 'frame', 'slot']
|
|
|
|
with open('Janet_test.conllu', encoding='utf-8') as trainfile:
|
|
trainset = list(parse_incr(trainfile, fields=fields, field_parsers={'slot': self.nolabel2o}))
|
|
with open('Janet_test.conllu', encoding='utf-8') as testfile:
|
|
testset = list(parse_incr(testfile, fields=fields, field_parsers={'slot': self.nolabel2o}))
|
|
|
|
tabulate(trainset[0], tablefmt='html')
|
|
|
|
corpus = Corpus(train=self.conllu2flair(trainset, 'slot'), test=self.conllu2flair(testset, 'slot'))
|
|
tag_dictionary = corpus.make_tag_dictionary(tag_type='slot')
|
|
|
|
embedding_types = [
|
|
WordEmbeddings('pl'),
|
|
FlairEmbeddings('pl-forward'),
|
|
FlairEmbeddings('pl-backward'),
|
|
CharacterEmbeddings(),
|
|
]
|
|
|
|
embeddings = StackedEmbeddings(embeddings=embedding_types)
|
|
tagger = SequenceTagger(hidden_size=256, embeddings=embeddings,
|
|
tag_dictionary=tag_dictionary,
|
|
tag_type='slot', use_crf=True)
|
|
|
|
trainer = ModelTrainer(tagger, corpus)
|
|
trainer.train('slot-model',
|
|
learning_rate=0.1,
|
|
mini_batch_size=32,
|
|
max_epochs=10,
|
|
train_with_dev=False)
|
|
|
|
model = SequenceTagger.load('slot-model/final-model.pt')
|
|
|
|
|
|
return model
|
|
|
|
def test_nlu(self, utterance):
|
|
|
|
model = self.setup()
|
|
|
|
if utterance:
|
|
return tabulate(self.predict(model, 'doktor lekarza rodzinnego najlepiej dzisiaj w godzinach popołudniowych dziś '.split()), tablefmt='html')
|
|
|
|
else:
|
|
return 'Critical Error'
|
|
|
|
class Book_NLU: #Natural Language Understanding
|
|
"""
|
|
Moduł odpowiedzialny za analizę tekstu. W wyniku jego działania tekstowa reprezentacja wypowiedzi użytkownika zostaje zamieniona na jej reprezentację semantyczną, najczęściej w postaci ramy.
|
|
|
|
Wejście: Tekst
|
|
|
|
Wyjście: Akt użytkownika (rama)
|
|
"""
|
|
def __init__(self, acts, arguments, book_grammar):
|
|
self.acts = acts
|
|
self.arguments = arguments
|
|
self.book_grammar = book_grammar
|
|
|
|
def get_dialog_act(self, rule):
|
|
slots = []
|
|
self.get_slots(rule.expansion, slots)
|
|
return {'act': rule.grammar.name, 'slots': slots}
|
|
|
|
def get_slots(self, expansion, slots):
|
|
if expansion.tag != '':
|
|
slots.append((expansion.tag, expansion.current_match))
|
|
return
|
|
|
|
for child in expansion.children:
|
|
self.get_slots(child, slots)
|
|
|
|
if not expansion.children and isinstance(expansion, jsgf.NamedRuleRef):
|
|
self.get_slots(expansion.referenced_rule.expansion, slots)
|
|
|
|
def analyze(self, text):
|
|
"""
|
|
Analiza Tekstu wprowadzonego przez użytkownika i zamiana na akt (rama)
|
|
"""
|
|
print("Analiza Tekstu: " + text)
|
|
act = "(greetings()&request(name))"
|
|
print("Akt to: " + act)
|
|
#przerobienie na wektor
|
|
act_vector = [[0],[1,0]] #1 wektor to greetings, a 2 wektor to request z argumentem "name"
|
|
print("Zamiana na: ")
|
|
print(act_vector)
|
|
return act_vector
|
|
|
|
def test_nlu(self, utterance):
|
|
matched = self.book_grammar.find_matching_rules(utterance)
|
|
print(matched)
|
|
|
|
if matched:
|
|
return self.get_dialog_act(matched[0])
|
|
|
|
else:
|
|
return {'act': 'null', 'slots': []}
|
|
|
|
class DST: #Dialogue State Tracker
|
|
"""
|
|
Moduł odpowiedzialny za śledzenie stanu dialogu. Przechowuje informacje o tym jakie dane zostały uzyskane od użytkownika w toku prowadzonej konwersacji.
|
|
|
|
Wejście: Akt użytkownika (rama)
|
|
|
|
Wyjście: Reprezentacja stanu dialogu (rama)
|
|
"""
|
|
def __init__(self, acts, arguments):
|
|
self.acts = acts
|
|
self.arguments = arguments
|
|
self.frame_list= []
|
|
|
|
|
|
def store(self, rama):
|
|
"""
|
|
Dodanie nowego aktu do listy
|
|
"""
|
|
print("\nDodanie do listy nowej ramy: ")
|
|
print(rama)
|
|
self.frame_list.append(rama)
|
|
|
|
|
|
def transfer(self):
|
|
print("Przekazanie dalej listy ram: ")
|
|
print(self.frame_list)
|
|
return self.frame_list
|
|
|
|
class DP:
|
|
"""
|
|
Moduł decydujący o wyborze kolejnego aktu, który ma podjąć system prowadząc rozmowę.
|
|
|
|
Wejście: Reprezentacja stanu dialogu (rama)
|
|
|
|
Wyjście: Akt systemu (rama)
|
|
"""
|
|
def __init__(self, acts, arguments):
|
|
self.acts = acts
|
|
self.arguments = arguments
|
|
|
|
|
|
def choose_tactic(self, frame_list):
|
|
"""
|
|
Obieranie taktyki na podstawie aktów usera. Bardzo ważna jest kolejność dodawanych do frame_list wartości.
|
|
"""
|
|
act_vector = [0, 0]
|
|
return act_vector
|
|
|
|
class NLG:
|
|
"""
|
|
Moduł, który tworzy reprezentację tekstową aktu systemowego wybranego przez taktykę dialogu.
|
|
|
|
Wejście: Akt systemu (rama)
|
|
|
|
Wyjście: Tekst
|
|
"""
|
|
def __init__(self, acts, arguments):
|
|
self.acts = acts
|
|
self.arguments = arguments
|
|
|
|
|
|
def change_to_text(self, act_vector):
|
|
"""
|
|
Funkcja zamieniająca akt systemu na tekst rozumiany przez użytkownika.
|
|
"""
|
|
if(act_vector == [0, 0]):
|
|
return "Cześć, mam na imię Janet"
|
|
return "Nie rozumiem"
|
|
|
|
|
|
class Janet:
|
|
def __init__(self):
|
|
self.acts={
|
|
0: "greetings",
|
|
1: "request",
|
|
}
|
|
self.arguments={
|
|
0: "name"
|
|
}
|
|
self.nlg = NLG(self.acts, self.arguments)
|
|
self.dp = DP(self.acts, self.arguments)
|
|
self.dst = DST(self.acts, self.arguments)
|
|
self.nlu = Book_NLU(self.acts, self.arguments, jsgf.parse_grammar_file('book.jsgf'))
|
|
self.nlu_v2 = ML_NLU(self.acts, self.arguments)
|
|
|
|
def test(self, command):
|
|
out = self.nlu_v2.test_nlu(command)
|
|
return out
|
|
|
|
def process(self, command):
|
|
act = self.nlu.analyze(command)
|
|
self.dst.store(act)
|
|
dest_act = self.dp.choose_tactic(self.dst.transfer())
|
|
return self.nlg.change_to_text(dest_act)
|
|
|
|
janet = Janet()
|
|
print(janet.test('chciałbym się umówić na wizytę do Piotra Pająka na jutro')) #Testowy print na start
|
|
while(1):
|
|
print('\n')
|
|
text = input("Wpisz tekst: ")
|
|
print(janet.test(text)) |