This chatbot is designed to assist people with choosing a recipe to cook. It is connected to a database of over 200,000 recipes from Food.com. You can ask the chatbot for suggestions of recipes based on criteria like ingredients, meal, cooking time, dietary restrictions, food ethnicity, etc.
Kaggle Dataset: Recipes from Food.com
We began by loading the data into a Pandas dataframe. The features we decided to use are
- recipe name
- average rating (out of 5)
- cook time (in minutes)
- tags (such as "vegetarian" or "main course")
- description of the dish
- number of ingredients
- list of ingredients
- cooking steps
- nutrition information
We combined these features into a single text string for each recipe, then used a FAISS document store to create vector embeddings for our recipes.
The app is deployed using Chainlit, Docker and Ploomber Cloud.
Ben Marsh
Eva Draganova
Miao Zheng