Visual Search Personal Assistant with Knowledge Graph
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional search methods are inefficient due to their text-based nature, leading to overwhelming irrelevant results and difficulties in communicating search intent, especially in multi-turn dialogues, as they were not designed to handle the scale of online searches which have ballooned to billions of possible products.
Innovation Solution
An intelligent personal assistant system leveraging scalable AI, a knowledge graph, and machine learning to understand user intents, allowing for intuitive and personalized interactions through various input modalities such as text, voice, and images, and integrating visual search with knowledge graph-based searches to provide accurate and relevant product suggestions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional text-based search methods are used, then search functionality is provided, but search efficiency deteriorates and user time is lost due to overwhelming irrelevant results
Solution Approach 1:
The patent replaces traditional text-based search mechanics with image-based visual search mechanics. Users can upload images to search for similar products, and the system uses image recognition and comparison algorithms to retrieve relevant results, eliminating the need for manual text query formulation and result filtering.
Solution Approach 2:
The patent introduces a visual search intermediary system that acts as a mediator between user intent and product results. The system includes image upload interfaces, visual comparison algorithms, and result ranking mechanisms that filter and prioritize relevant products, reducing the time users spend browsing irrelevant results.
2Ease of operation
If text-based search interfaces are used, then basic search capability is provided, but ease of operation deteriorates due to difficulty in communicating search intent
Solution Approach 1:
The patent replaces text-based input mechanics with image-based input mechanics. Users can directly upload product images to express their search intent, which is more intuitive and accurate than describing products in text. The system processes these images through visual recognition algorithms to understand and fulfill user search intent.
Solution Approach 2:
The patent uses image copying and comparison mechanisms where users can upload images of target products, and the system searches for similar products by comparing visual features. This copying approach preserves the exact visual characteristics users are interested in, ensuring accurate communication of search intent.
3Adaptability or versatility
If conventional search tools are used, then simple search functionality is provided, but adaptability deteriorates when handling billions of online products
Solution Approach 1:
The patent segments the large-scale product catalog search problem into manageable components: image preprocessing, feature extraction, similarity computation, and result ranking. Each component handles a specific aspect of the search process, making the overall system adaptable to billions of products while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The patent transitions from traditional text-based one-dimensional search to multi-dimensional visual search by incorporating image features, color, texture, shape, and other visual attributes. This dimensional expansion enables the system to adapt to large-scale product catalogs by searching across multiple visual dimensions simultaneously.
Data Source
AI summary
Systems, methods, and computer program products for identifying a relevant candidate product in an electronic marketplace. Embodiments perform a visual similarity comparison between candidate product image visual content and input query image visual content, process formal and informal natural language user inputs, and coordinate aggregated past user interactions with the marketplace stored in a knowledge graph. Visually similar items and their corresponding product categories, aspects, and aspect values can determine suggested candidate products without discernible delay during a multi-turn user dialog. The user can then refine the search for the most relevant items available for purchase by providing responses to machine-generated prompts that are based on the initial search results from visual, voice, and/or text inputs. An intelligent online personal assistant can thus guide a user to the most relevant candidate product more efficiently than existing search tools.