Skip to content

ChatGPT Plugin to Semantically Search Google Maps

Notifications You must be signed in to change notification settings

wonhyeongseo/chatmaps

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

43 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

chatmaps

chatmaps is a chatgpt plugin that allows you to semantically search google maps. currently is geo-locked to San Francisco, California.

Screenshot 2023-04-28 at 12 07 28 PM

Usage

Interact with the ChatMaps by installing this URL into the ChatGPT plugin interface: chatmaps.sdan.io

  • "Kinda like Blue Bottle"
  • "Sweetgreen but not sweetgreen"
  • "Find a romantic Italian restaurant in North Beach"
  • "Sushi places with outdoor seating in the Mission District"
  • "Cozy coffee shop with Wi-Fi in the Castro"
  • "Vegan-friendly lunch spot in the Financial District"
  • "Family-friendly pizza place in Noe Valley"
  • "Late-night dessert options in the Marina"
  • "Dog-friendly brewery in Dogpatch"
  • "Trendy cocktail bar with a view in Russian Hill"
  • "Farm-to-table brunch spot in Pacific Heights" "Casual Mexican taqueria in the Haight-Ashbury"

Overview

this chatgpt plugin currently provides personalized restaurant recommendations based on your search queries. It fetches embeddings from the embeddings database to find relevant information such as names, addresses, ratings, opening hours, reviews, if they do takeout/delivery/etc, and the types of cuisine served at each restaurant.

Features

  • Gets better over time -- more embeddings that are slowly added will increasingly improve search results
  • Fetch and display personalized restaurant recommendations
  • Show detailed information about each recommended restaurant
  • No installation required. Just add chatmaps.sdan.io as an unverified plugin on ChatGPT's UI

How it works

  • Instead of making costly Google Maps API calls at runtime, I instead embed all locations into a hosted ChromaDB instance running on Railway.
  • Some of the metadata that is tagged along with each location includes:
    • Name
    • Address
    • Editorial Summary
    • Rating
    • Type of establishment
    • Total number of ratings
    • Price level
    • Opening hours
    • Reviews(usually top 3 relevant)
    • Dine in
    • Delivery
    • Takeout
  • User queries are embedded then matched with relevant restaurant locations, which are then sent back to ChatGPT
  • At a high level I use a technique similar toRAG(Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks)

Enjoy exploring new dining experiences with the ChatMaps Restaurant Recommendations Plugin!

About

ChatGPT Plugin to Semantically Search Google Maps

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 81.4%
  • HTML 16.2%
  • Dockerfile 2.4%