Thanks for your reply.
Actually, I used the postgresql, and I got the data with this function :
def text_from_row(row):
field_labels = {
“vid_vin”: “Vehicle Identification Number”,
“date_min”: “Started date”,
“date_max”: “End date”,
“days_seen”: “So, days seen”,
“listing_stock”: “Stock”,
“listing_price”: “Price”,
“listing_type”: “Used/New”,
“listing_mileage”: “Mileage”,
“vehicle_year”: “Year”,
“vehicle_make”: “Made Company”,
“vehicle_model”: “Model”,
“vehicle_trim”: “Trim”,
“vehicle_style”: “Style”,
“vehicle_color_exterior”: “Color Exterior”,
“vehicle_color_interior”: “Color Interior”,
“va_seller_id”: “Seller id”,
“va_seller_name”: “Seller name”,
“va_seller_address”: “Seller address”,
“va_seller_city”: “Seller City”,
“va_seller_state”: “Seller State”,
“va_seller_zip”: “Seller Zip”,
“va_seller_county”: “Seller County”,
“va_seller_country”: “Seller Country”,
“va_seller_websites”: “Seller Websites”,
“va_seller_phones”: “Seller Phones”,
“va_seller_type”: “Seller Type”,
“va_seller_makes”: “Seller makes”,
“va_seller_inventory_count_total”: “Seller inventory count total”,
“va_seller_inventory_count_new”: “Seller inventory count new”,
“va_seller_inventory_count_used”: “Seller inventory count used”,
“vdp_url”: “Vehicle Detail Page”,
“ymmt_id”: “Year, Make, Model, and Trim (YMMT) identifier”,
“listing_description”: “Vehicle Description”,
“listing_features”: “Vehicle Features”,
“vehicle_title”: “Vehicle Title”,
“vehicle_subtitle”: “Vehicle Subtitle”,
“vehicle_type”: “Vehicle Type”,
“vehicle_truck_cab_style”: “Vehicle Truck Cab Style”,
“vehicle_truck_bed_style”: “Vehicle Truck Bed Style”,
“vehicle_engine”: “Vehicle Engine”,
“vehicle_engine_size”: “Engine size”,
“vehicle_engine_cylinders”: “Cylinzers”,
“vehicle_transmission”: “Vehicle Transmission”,
“vehicle_transmission_type”: “Transmission Type”,
“vehicle_transmission_speed”: “Transmission Speed”,
“vehicle_drivetrain”: “Vehicle Drivetrain”,
“vehicle_doors”: “Doors”,
“vehicle_fuel_type”: “Fuel Type”,
“vehicle_fuel_efficiency”: “Fuel Efficiency”,
“vehicle_fuel_efficiency_highway”: “Fuel Efficiency highway”,
“vehicle_fuel_efficiency_city”: “Fuel Efficiency city”,
“vehicle_history_description”: “History Description”,
“vehicle_history_critical_count”: “History critical count”,
“vehicle_history_accident_count”: “History accident count”,
“vehicle_history_theft_count”: “History theft count”,
“vehicle_history_salvage_count”: “History salvage count”,
“vehicle_history_service_count”: “History Service count”,
“vehicle_history_owner_count”: “History owner count”,
“portal_urls”: “Portal Urls”,
“portal_deal_ratings”: “Portal Deal Ratings”,
“portal_days_online”: “Portal Days Online”,
“portal_prices”: “Portal Prices”,
“portal_titles”: “Portal Titles”,
“vdp_url_last_crawled”: “Vehicle Detail Page Url Last Crawled”,
“vehicle_history_report_urls”: “Vehicle history report urls”,
“va_seller_portal_websites”: “Seller portal websites”,
“va_seller_portal_urls”: “Seller portal Urls”,
}
formatted_data = “”
for field_name, label in field_labels.items():
if field_name in row:
formatted_data += f"{label}: {row[field_name]}\n"
return formatted_data
def get_embedding(text):
try:
response = openai.Embedding.create(
model=“text-embedding-ada-002”, input=text.replace(“\n”, " “)
)
embedding = response[“data”][0][“embedding”]
embedding_array = np.array(embedding)
return embedding_array
except Exception as e:
print(f"An error occurred: {str(e)}”)
return None
And this is the function that gets the embedding from the formatted_data.
I used pgvector.
That’s all.
What I want is if the customer ask “Recommend the most expensive white car.”
In that case, I want the chatbot to reply the most expensive white car.
