import gradio as gr
import pandas as pd
import numpy as np
from sklearn.metrics.pairwise import cosine_similarity
from sentence_transformers import SentenceTransformer
# Load your model
model = SentenceTransformer('sentence-transformers/distiluse-base-multilingual-cased-v2')
# Load embeddings and DataFrames (replace these with your paths)
embeddings_hotels = np.load("embeddings_hotels.npy")
embeddings_ar = np.load("embeddings_ar.npy")
df_hotels = pd.read_csv("hotels.csv")
df_ar = pd.read_csv("arabic_data.csv")
def search_in_combined(query_text, model, k=5):
query_embedding = model.encode(query_text, convert_to_tensor=True).cpu().numpy().reshape(1, -1)
similarities_hotels = cosine_similarity(query_embedding, embeddings_hotels).flatten()
similarities_ar = cosine_similarity(query_embedding, embeddings_ar).flatten()
top_indices_hotels = np.argsort(similarities_hotels)[::-1][:k]
top_indices_ar = np.argsort(similarities_ar)[::-1][:k]
top_hotels = df_hotels.iloc[top_indices_hotels].copy()
top_ar = df_ar.iloc[top_indices_ar].copy()
top_hotels['similarity'] = similarities_hotels[top_indices_hotels]
top_ar['similarity'] = similarities_ar[top_indices_ar]
combined_top_results = pd.concat([top_hotels, top_ar], ignore_index=True)
combined_top_results = combined_top_results.sort_values(by='similarity', ascending=False)
combined_top_results['google_maps_url'] = combined_top_results.apply(
lambda row: f"https://www.google.com/maps/search/?api=1&query={row['hotel_name'].replace(' ', '+')}"
if 'hotel_name' in row and not pd.isna(row['hotel_name']) else '',
axis=1
)
return combined_top_results.head(k)
def format_results(results):
formatted_results = []
for _, row in results.iterrows():
if not pd.isna(row.get('hotel_name', '')):
google_maps_url = f"https://www.google.com/maps/search/?api=1&query={row.get('hotel_name', 'N/A').replace(' ', '+')}"
result = (
f"Hotel Name: {row.get('hotel_name', 'N/A')}
"
+ (f"Description: {row.get('hotel_description', 'N/A')}
" if not pd.isna(row.get('hotel_description', '')) else "")
+ f"Review Title: {row.get('review_title', 'N/A')}
"
f"Review Text: {row.get('review_text', 'N/A')}
"
f"Rating: {row.get('rate', 'N/A')}
"
f"Trip Date: {row.get('tripdate', 'N/A')}
"
f"Price Range: {row.get('price_range', 'N/A')}
"
f"Location: {row.get('locality', 'N/A')} , {row.get('country', 'N/A')}
"
f"Hotel Website URL: Link
"
f"Google Maps: View on Maps
"
f"Image: 
"
)
else:
google_maps_url = f"https://www.google.com/maps/search/?api=1&query={row.get('name', 'N/A').replace(' ', '+')}"
result = (
f"Name: {row.get('name', 'N/A')}
"
f"Location: {row.get('location', 'N/A')}
"
f"Price: {row.get('price', 'N/A')}
"
f"Price For: {row.get('price_for', 'N/A')}
"
f"Room Type: {row.get('room_type', 'N/A')}
"
f"Beds: {row.get('beds', 'N/A')}
"
f"Rating: {row.get('rating', 'N/A')}
"
f"Rating Title: {row.get('rating_title', 'N/A')}
"
f"Google Maps: View on Maps
"
f"Number of Ratings: {row.get('number_of_ratings', 'N/A')}
"
f"Hotel Website URL: Link
"
f"Additional Info: {row.get('cm', 'N/A')}
"
)
formatted_results.append(result)
return "
".join(formatted_results)
def search_interface(query_text):
results = search_in_combined(query_text, model, k=5)
return format_results(results)
iface = gr.Interface(
fn=search_interface,
inputs=gr.Textbox(label="Enter your search query"),
outputs=gr.HTML(label="Search Results"),
title="Hotel Search Beta(v0.1)",
description="Enter a query to search for your appropriate hotel. The results will show the top matches based on similarity and provide a Google Maps URL for hotel locations ,and anouther info about the hotel.",
examples=["Riyadh", "Deluxe Room"]
)
iface.launch()