Dynamic Vector Recommendation System for Fashion Trends
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional information providing methods using curated databases can only satisfy 60% to 70% of users, failing to meet the needs of superfans and advanced users who seek new experiences, as they are based on static data and general scales, unable to adapt to changing fashion trends and individual preferences.
Innovation Solution
An information providing system that generates item lists based on user and item profiles, using session information to calculate a target social position, updating user and item vectors dynamically, and proposing items in descending order of satisfaction scores, to cater to a broader range of user groups, including superfans and advanced users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a curated database is used for information provision, then information can be provided to follower group users, but the system cannot satisfy superfans and advanced users who seek new experiences
Solution Approach 1:
The patent segments users into different groups (follower group, superfans, advanced users) and applies different recommendation strategies for each segment. The system divides the recommendation task into multiple processing stages including vector generation, social position calculation, and dynamic adjustment, allowing tailored approaches for different user types without requiring a completely different system for each group.
Solution Approach 2:
The patent implements dynamic user vectors that are continuously updated based on user interactions, session information, and social position changes. This dynamic approach allows the system to adapt to evolving user preferences and trends in real-time, enabling it to satisfy both traditional follower group users and advanced users seeking new experiences by reflecting current fashion trends and individual preferences.
2Adaptability or versatility
If static database data is used, then the system is simple to maintain, but it cannot adapt to changing fashion trends and user preferences
Solution Approach 1:
The patent implements feedback mechanisms where user interactions, selections, and session information are continuously fed back into the system to update user vectors and item vectors. This feedback loop enables the system to automatically adapt to changing fashion trends and user preferences without manual intervention, maintaining high adaptability while managing data processing through efficient vector update algorithms.
Solution Approach 2:
The patent changes the state of data from static to dynamic by continuously updating user vectors and item vectors based on new information. The system adjusts vector parameters reflecting user preferences, social positions, and item attributes in real-time, enabling adaptation to trending topics and changing user tastes while maintaining computational efficiency through targeted updates rather than complete reprocessing.
3Measurement precision
If general scale proposal is used, then follower group users are satisfied, but superfans with strong intentionality cannot be satisfied
Solution Approach 1:
The patent applies local quality by calculating specific social positions for users within the fashion industry context and generating tailored recommendation lists based on individual user characteristics. Instead of uniform general-scale proposals, the system computes personalized user vectors, determines social positions relative to fashion trends, and generates customized item lists that precisely match each user's specific preferences and intensity levels.
Data Source
AI summary
An information providing system 1 includes: an item list generator 55 that generates a match-pair list including items associated with a subject user based on information registered in a user profile database 51 and information registered in an item profile database 52; a session information processing system 6 that acquires session information in a session with the subject user via a user interface 2 and calculates a target social position of the subject user in a social space formed by reflection of sense of values of a plurality of registered users registered in the user profile database 51, based on the session information; and an item proposer 7 that proposes items to the subject user based on the match-pair list and the target social position.