File size: 2,763 Bytes
6be3cee e2d1132 beb5a28 6be3cee e2d1132 39f9620 5cfa812 d458480 e2d1132 5cfa812 4855174 d458480 5cfa812 d458480 5cfa812 e2d1132 6be3cee 39f9620 e2d1132 600b0cd e2d1132 600b0cd e2d1132 e216d81 5cfa812 e2d1132 5cfa812 e216d81 e2d1132 39f9620 e2d1132 39f9620 600b0cd e2d1132 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 |
import gradio as gr
import pandas as pd
from sentence_transformers import SentenceTransformer, util
# Load FAQ
try:
faq_df = pd.read_csv("lic_faq.csv", encoding="utf-8")
except UnicodeDecodeError:
faq_df = pd.read_csv("lic_faq.csv", encoding="ISO-8859-1")
model = SentenceTransformer('all-MiniLM-L6-v2')
faq_embeddings = model.encode(faq_df['question'].tolist(), convert_to_tensor=True)
# Policy recommendations
policy_suggestions = {
"term": "💡 You might consider LIC Tech Term Plan for pure protection at low cost.",
"money back": "💡 LIC Money Back Policy is great for periodic returns along with insurance.",
"endowment": "💡 LIC New Endowment Plan offers savings and insurance benefits together.",
"ulip": "💡 LIC SIIP and Nivesh Plus are good ULIP options with market-linked returns.",
"pension": "💡 LIC Jeevan Akshay and PM Vaya Vandana Yojana are best for pension seekers."
}
def chatbot(history, query):
query_lower = query.lower().strip()
# Handle greetings
if query_lower in ["hi", "hello", "hey", "good morning", "good evening"]:
response = "👋 Hello! I’m your LIC Assistant. Ask me anything about policies, claims, onboarding, or commission."
else:
query_embedding = model.encode(query, convert_to_tensor=True)
scores = util.pytorch_cos_sim(query_embedding, faq_embeddings)[0]
best_score = float(scores.max())
best_idx = int(scores.argmax())
if best_score < 0.6:
response = "🤖 I'm not confident I have the right answer for that. Please ask about LIC policies, claims, commissions, onboarding, or KYC."
else:
response = faq_df.iloc[best_idx]['answer']
for keyword, suggestion in policy_suggestions.items():
if keyword in query_lower:
response += f"\n\n{suggestion}"
break
history.append((query, response))
return history, history
with gr.Blocks(title="LIC Agent Chatbot") as demo:
gr.Markdown(
"<h1 style='text-align:center;color:#0D47A1;'>🧑💼 LIC Agent Assistant</h1>"
"<p style='text-align:center;'>Ask me anything about policies, claims, commissions, onboarding, and KYC.</p>"
)
chatbot_ui = gr.Chatbot(label="LIC Assistant", height=450, bubble_full_width=False, avatar_images=("🧑", "🤖"))
with gr.Row():
msg = gr.Textbox(placeholder="Ask your question here...", show_label=False, scale=8)
send = gr.Button("Send", variant="primary", scale=2)
clear = gr.Button("Clear Chat")
state = gr.State([])
send.click(fn=chatbot, inputs=[state, msg], outputs=[chatbot_ui, state])
clear.click(lambda: ([], []), None, [chatbot_ui, state], queue=False)
demo.launch()
|