Update app.py

app.py

@@ -1,56 +1,54 @@
+import json
+import pickle
 from sklearn.feature_extraction.text import CountVectorizer
-from sklearn.naive_bayes import MultinomialNB
 import gradio as gr
 
-# Example data
-train_queries = [
-    "How do I activate my card?",
-    "What is the age limit for opening an account?",
-    "Do you support Apple Pay or Google Pay?",
-]
-train_labels = [0, 1, 2]
+# Load the Naive Bayes model and vectorizer
+with open("naive_bayes_model.pkl", "rb") as model_file:
+    clf = pickle.load(model_file)
 
+with open("vectorizer.pkl", "rb") as vectorizer_file:
+    vectorizer = pickle.load(vectorizer_file)
+
+# Define the responses dictionary
 responses = {
-    0: "To activate your card, please go to the app's settings.",
-    1: "The age limit for opening an account is 18 years.",
-    2: "Yes, we support Apple Pay and Google Pay.",
+    "activate_my_card": "To activate your card, log in to the app and navigate to the 'Activate Card' section.",
+    "age_limit": "The minimum age to use this service is 18 years.",
+    "apple_pay_or_google_pay": "Yes, you can use both Apple Pay and Google Pay with your card.",
+    "atm_support": "Your card is supported by ATMs that display the Visa or Mastercard logo.",
+    "automatic_top_up": "You can enable automatic top-up in the app under the 'Top-Up Settings' section.",
+    "balance_not_updated_after_bank_transfer": "If your balance hasn’t updated after a bank transfer, please wait for 24 hours. If the issue persists, contact support.",
+    "beneficiary_not_allowed": "Ensure that the beneficiary details are correct. Some accounts may have restrictions; check with customer support for clarification.",
+    "cancel_transfer": "To cancel a transfer, go to the 'Transaction History' section in the app and select the transfer you wish to cancel.",
+    "card_about_to_expire": "If your card is about to expire, a replacement will be sent automatically. Contact support if you haven’t received it.",
+    "card_arrival": "New cards usually arrive within 7-10 business days after being issued.",
+    "card_not_working": "If your card isn't working, ensure it is activated and has sufficient balance. Contact support if the issue persists.",
+    "change_pin": "You can change your PIN using the app or at any ATM with your card.",
+    "contactless_not_working": "Ensure your card supports contactless payments and check if the terminal accepts it. If the issue persists, contact support.",
+    "country_support": "Your card is supported in all countries where Visa/Mastercard is accepted. Check the app for restrictions.",
+    "declined_card_payment": "Card payments can be declined due to insufficient balance, incorrect PIN, or restrictions on the merchant. Check your app for details.",
+    "lost_or_stolen_card": "If your card is lost or stolen, block it immediately in the app under the 'Card Management' section and request a replacement.",
+    "pending_card_payment": "Pending payments are usually resolved within 2-3 business days. Contact support if the status doesn't update.",
+    "refund_not_showing_up": "Refunds can take up to 7 business days to appear. Check with the merchant or contact support if it takes longer.",
+    "top_up_failed": "Top-up failures may occur due to incorrect details or insufficient balance in the source account. Check your app for details.",
+    "transfer_not_received_by_recipient": "If the recipient hasn't received the transfer, ensure the details are correct. Contact support for further assistance.",
 }
 
-label_to_intent = {0: "activate_my_card", 1: "age_limit", 2: "apple_pay_or_google_pay"}
-
-# Prepare the Naive Bayes model
-vectorizer = CountVectorizer()
-X_train = vectorizer.fit_transform(train_queries)
-
-clf = MultinomialNB()
-clf.fit(X_train, train_labels)
-
-# Define the chatbot response function
-def naive_bayes_response(user_input):
+# Define chatbot response logic
+def chatbot_response(user_input):
     vectorized_input = vectorizer.transform([user_input])
     predicted_label = clf.predict(vectorized_input)[0]
-    return responses.get(predicted_label, "Sorry, I couldn't understand your query.")
-
-# Define Gradio interface
-def chatbot_interface(user_input):
-    return naive_bayes_response(user_input)
+    return responses.get(predicted_label, "Sorry, I couldn't understand your question.")
 
-# UI design with Gradio
-with gr.Blocks() as demo:
-    gr.Markdown("# Naive Bayes Chatbot")
-    gr.Markdown("This is a chatbot powered by Naive Bayes that handles basic queries.")
-    with gr.Row():
-        with gr.Column():
-            user_input = gr.Textbox(
-                label="Your Query",
-                placeholder="Type your question here...",
-                lines=1,
-            )
-            submit_btn = gr.Button("Submit")
-        with gr.Column():
-            response = gr.Textbox(label="Chatbot Response", interactive=False)
-    submit_btn.click(chatbot_interface, inputs=user_input, outputs=response)
+# Define the Gradio interface
+interface = gr.Interface(
+    fn=chatbot_response,
+    inputs=gr.Textbox(lines=2, placeholder="Ask your fintech question here..."),
+    outputs="text",
+    title="FinTech Chatbot",
+    description="A chatbot to handle fintech-related queries using a Naive Bayes model."
+)
 
-# Run the app
 if __name__ == "__main__":
-    demo.launch()
+    # Launch the Gradio app
+    interface.launch()