Image-Based Article Recommendation System for Absent Commodity Discovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing online shopping systems lack the ability to suggest commodities based on images, specifically recognizing articles absent in an input image for personalized recommendations.
Innovation Solution
A recommendation system that analyzes input images to recognize article sections, specifies categories, extracts identification information for unrecognized article sections, and outputs relevant commodity information, including URLs for search results, allowing users to discover associated commodities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If image analysis is used to recognize articles, then personalized recommendation capability is improved, but system complexity increases
Solution Approach 1:
The system divides the recommendation process into distinct functional modules: image input processing, article recognition, category specification, and commodity information extraction. Each module handles a specific aspect of the recommendation task, making the complex system manageable and maintainable while enabling personalized recommendations based on image analysis
Solution Approach 2:
The patent introduces an intermediary database structure that links images, articles, categories, and commodities. This intermediary layer facilitates the complex relationships between different data types, enabling the system to process image inputs and generate personalized recommendations without directly managing all the complexity in a single monolithic system
2Measurement precision
If article recognition is performed on input images, then commodity search accuracy is improved, but processing time increases
Solution Approach 1:
The system performs preliminary processing by extracting and storing article information from images in advance, organizing it in a database structured by categories. This pre-processing enables faster commodity search operations during actual user queries, as the recognition results are already available and can be directly queried without re-processing the entire image dataset
Solution Approach 2:
The patent applies different processing depths to different parts of the system based on local requirements. Image analysis is performed only when needed for new recommendations, while existing article recognition results are cached and reused for subsequent searches, optimizing the balance between accuracy and processing time for different operational contexts
3Quantity of substance
If multiple article sections are extracted from image, then recommendation comprehensiveness is improved, but information processing complexity increases
Solution Approach 1:
The system segments the extracted article information into distinct categories and stores them in a structured database format. This segmentation allows the system to process multiple articles efficiently by querying specific categories independently, reducing the complexity of handling comprehensive recommendation data while maintaining thoroughness
Solution Approach 2:
The patent transforms the raw extracted article information into standardized database parameters and categories. By converting diverse article data into uniform parameter formats, the system can handle comprehensive recommendation data without increasing processing complexity, as all data is normalized to the same structure for efficient retrieval and processing
Data Source
AI summary
An object is to conduct a search for an article absent in an input image. A recommendation system according to an embodiment of the present invention is provided with an input unit to achieve input of a still image, an article recognizing unit to analyze the input image to recognize an article section according to an article included in the image, a category specifying unit to specify a category according to the input image on the basis of the recognized article section, an article information extracting unit to extract identification information of an article section except for the recognized article section among article sections associated with the specified category, and an output unit to output the extracted identification information of the article section and/or information corresponding thereto. This configuration allows the recommendation system of the embodiment to provide a function (article search function) to achieve input of a still image including one or more articles and output of identification information of an article section according to an article absent in the image and/or other information.