Picture-Based Friend Recommendation Using Attribute Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current friend recommendation methods on social networking sites often result in a large amount of irrelevant friend recommendations, wasting bandwidth and storage space, as they do not accurately reflect users' interests in establishing friendships.
Innovation Solution
A method and apparatus that utilize picture-based friend recommendations, where a server determines users associated with a picture through attribute information, such as face recognition, and sends friend recommendation information to users who are not already friends, increasing the likelihood of relevant connections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional friend recommendation methods (based on friends list or registration information) are used, then the recommendation coverage is broad, but the relevance of recommended friends to users is low
Solution Approach 1:
The patent introduces picture attribute information as an intermediary factor to connect users. By analyzing pictures that users have uploaded or shared, the system identifies common visual elements (landmarks, activities, objects) that serve as mediators to recommend friends with similar picture attributes, thereby improving relevance while maintaining recommendation coverage
Solution Approach 2:
The patent changes the recommendation parameters from traditional attributes (friends list, registration info) to picture-based attributes (visual content, scene characteristics). This parameter transformation enables the system to capture user interests and social context more accurately, resolving the contradiction between quantity and relevance
2Quantity of substance
If a large number of friend recommendations are sent to users, then the recommendation coverage increases, but the bandwidth and storage space consumed increase
Solution Approach 1:
The patent extracts and processes picture attribute information from the vast amount of user-generated content. By extracting key visual features and using them as filtering criteria, the system can reduce the recommendation pool to high-quality candidates, thereby reducing the volume of recommendation data that needs to be transmitted and stored
3Ease of manufacture
If friend recommendations are based on common registration information (place of origin, school), then the recommendation process is simple, but the accuracy of reflecting user friendship needs is low
Solution Approach 1:
The patent introduces picture analysis as an intermediary that bridges the gap between simple registration data and complex user friendship needs. Pictures serve as a richer intermediary that captures user interests, activities, and social context, enabling more accurate friendship recommendations while maintaining process simplicity through automated image processing
Data Source
AI summary
A method is for friend recommendation which includes obtaining a first picture sent by a user, and determining one or more users associated with the first picture based on attribute information of the first picture. The method also includes, when it is determined that a total number of the users associated with the first picture is two or more, detecting whether a first user and a second user from the one or more users associated with the first picture are friends. Further, the method includes, when it is detected that the first user and the second user are not friends, sending friend recommendation information to one of the first user and the second user, wherein the friend recommendation information contains information of the other of the first user and the second. It is more likely that the friend recommendation information is of real interest of the users receiving the information.