Recipe Embedding Recommendations for Faster Online Grocery Browsing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current online concierge systems face challenges in efficiently selecting recipes due to increased time required for users to browse through numerous recipes, which discourages further interaction.
Innovation Solution
An online concierge system utilizes a recipe graph to maintain connections between recipes based on attributes and user interactions, generating recipe embeddings to personalize recipe recommendations using a neural network model, and leveraging user embeddings to suggest recipes based on similarity and interaction history.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the online concierge system organizes recipes into different sections to simplify user browsing, then the ease of operation is improved, but the time required for users to select a recipe increases
Solution Approach 1:
The system pre-generates recipe embeddings and user embeddings before the user actually needs to select a recipe. These embeddings capture the essential features and preferences in advance, so when the user visits, the matching process is already prepared and can quickly retrieve relevant recipes without requiring the user to spend time browsing through organized sections
Solution Approach 2:
The patent replaces the mechanical browsing system (where users manually navigate through organized recipe sections) with an automated information retrieval system using embeddings. Instead of users physically moving through categorized lists, the system uses vector space matching to automatically present relevant recipes based on user preferences and recipe characteristics
2Adaptability or versatility
If the online concierge system provides more recipes in different sections, then the adaptability is improved, but the time required for users to identify a recipe of interest increases
Solution Approach 1:
The system transforms recipes and user preferences into a different parameter space (embedding vectors) that captures essential characteristics in a condensed form. By operating in this transformed vector space rather than the original recipe description space, the system can efficiently compare and match even when the number of recipes is large, maintaining both variety and speed
Data Source
AI summary
An online concierge shopping system identifies recipes to users to encourage them to include items from the recipes in orders. The online concierge system maintains user embeddings for users and recipe embeddings for recipes. For users who have not placed orders, recipes are recommended based on global user interactions with recipes. Users who have previously ordered items from recipes are suggested recipes selected based on a similarity of their user embedding to recipe embeddings. Users who have purchased items but not from recipes are compared to a set of similar users based on the user embeddings, and recipes with which users of the set of similar users interacted are used for identifying recipes to the users. A recipe graph may be maintained by the online concierge system to identify similarities between recipes for expanding candidate recipes to suggest to users.