Image-Based Intent Recommendations for Complementary Product Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional search engines provide broad and unhelpful search results, requiring users to search for multiple items related to a specific job or goal individually, leading to frustration and missed sales opportunities for retailers.
Innovation Solution
An intent-based shopping guidance system using augmented reality and artificial intelligence to analyze user data and provide filter suggestions, augmented search results, and AR images of recommended items in a user-selected setting, enhancing the shopping experience by predicting and recommending complementary items.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional search engines present a wide variety of product entries in response to queries, then the search results cover more product options, but the results become very broad and unhelpful to the user
Solution Approach 1:
The patent segments search results into multiple hierarchical levels: primary items directly matching the query, complementary items needed to complete a job, and substitute items that can replace primary items. This segmentation organizes the broad product variety into structured, easily navigable categories that help users find what they need without overwhelming them
Solution Approach 2:
The system performs preliminary analysis of user queries to identify the intended job or goal before presenting results. By pre-processing queries to understand context and requirements, the system can proactively organize and present relevant items across multiple levels, reducing the need for users to perform multiple separate searches
2Measurement precision
If users search for each related item one at a time, then the user can find specific items, but the process becomes time-consuming and tedious
Solution Approach 1:
The patent merges multiple search results into a single comprehensive presentation by combining primary items with their associated complementary and substitute items. This allows users to discover multiple related items in one search operation rather than conducting separate searches for each item needed to complete a job
Solution Approach 2:
The system provides feedback to users by presenting items organized around their intended job or goal. By showing complementary items that complete the user's objective and substitute items that offer alternatives, the system helps users make informed decisions quickly without needing to perform multiple iterative searches
3Adaptability or versatility
If traditional search engines provide broad search results, then more product options are available, but the retailer loses potential sales for related items and services
Solution Approach 1:
The patent applies local quality by providing different types of item recommendations in different sections of the search results. Primary items receive standard presentation, while complementary items are highlighted as needed to complete the job, and substitute items are presented as alternatives. This differentiated presentation ensures that related items receive appropriate visibility without compromising the overall product variety
Solution Approach 2:
The system performs preliminary identification of complementary and substitute items based on the user's query and the primary items found. By pre-analyzing relationships between items and the user's intended goal, the system can proactively present relevant related items that the user may not have known they needed, thereby capturing potential sales opportunities
Data Source
AI summary
Examples include identifying a primary item from an initial query and item attributes associated with the primary item; obtaining an image and identify setting details associated with a setting depicted in the image using a computer vision (CV) model; determining, using a machine learning (ML) model, a micro-level intent and a macro-level intent for the primary item; selecting, using the ML model, a plurality of recommended items for use in conjunction with the primary item, wherein the plurality of recommended items includes at least one micro-item belonging to a same or similar item-type as the primary item selected based on the micro-level intent and at least one macro-item belonging to an item-type outside of the item-type of the primary item selected based on the macro-level intent; and presenting each of the plurality of recommended items as a selectable option for including in a selection experience including the primary item.