model.py

'''
Creating a TfidfVectorizer and SVM model for the chatbot
'''

#importing libraries

import json
import warnings
warnings.filterwarnings('ignore')

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.preprocessing import LabelEncoder
from sklearn.svm import SVC
from sklearn.pipeline import Pipeline
import pickle

with open(r"data\profile_intents.json") as file:
    intents = json.load(file)

# print(intents)

#initializing the arrays
patterns = []
responses = []
labels = []

#getting the data from json to arrays
for intent in intents["intents"]:
    for pattern in intent["patterns"]:
        patterns.append(pattern)
        labels.append(intent["tag"])
    responses.append({intent['tag']:intent['responses']})

#print(patterns,'\n\n\n',responses,'\n\n\n',lables)

# encode the Lables
label_encoder = LabelEncoder()
encoded_lables = label_encoder.fit_transform(labels)

#print(encoded_lables)

#Creating a pipeline
pipeline = Pipeline([
    ('tfidf',TfidfVectorizer(stop_words = 'english')),
    ('svm',SVC(kernel='rbf',probability = True))
])

#Train the pipeline
pipeline.fit(patterns, encoded_lables)

#Saving the pipeline
with open("pickles/chatbot_pipeline.pkl","wb") as file:
    pickle.dump(pipeline,file)
with open("pickles/label_encoder.pkl","wb") as file:
    pickle.dump(label_encoder,file)
with open("pickles/responses.pkl","wb") as file:
    pickle.dump(responses,file)

print("Model trained and Saved")