User Experience-Based Content Generation with Multi-Model Feedback Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI-based content generation methods struggle to produce creative content that accurately reflects a user's personal experiences and thoughts due to unrefined language inputs and mismatched user needs.
Innovation Solution
A content generation platform server utilizing multiple learning models to process user inputs, generate reconstructed content, and incorporate user feedback to refine and align generated content with personal experiences and thoughts, including image captioning and morphological analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If AI-based content generation methods are used, then content can be generated automatically, but the generated content does not accurately reflect user experiences and thoughts due to unrefined language inputs
Solution Approach 1:
The patent implements a feedback mechanism where user preferences and selections from generated content are fed back into the learning model to continuously refine and improve future content generation, ensuring better alignment with user experiences and thoughts over time
Solution Approach 2:
The system performs preliminary processing of user inputs including language refinement and structuring before content generation, and provides preliminary content options for user selection before final output, improving both efficiency and accuracy
2Measurement precision
If multiple learning models and processing steps are implemented to improve content quality, then content alignment with user intentions improves, but system complexity increases
Solution Approach 1:
The patent combines multiple learning models (language model, recommendation model, generation model) into an integrated content generation system that operates as a unified platform, managing complexity through systematic integration rather than separate independent systems
3Measurement precision
If user feedback is continuously incorporated to refine content, then content creativity and user concept formation are maximized, but processing time and computational resources increase
Solution Approach 1:
The system performs preliminary content generation and structuring before user feedback is received, preparing multiple candidate options in advance that can be quickly refined based on user input, reducing overall processing time
Solution Approach 2:
The learning model operates continuously in the background, learning from user feedback and preferences without requiring complete system stops or reinitialization, maintaining continuous improvement while minimizing disruption to content generation workflow
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure aims to provide a process in which images having maximum creativity are generated through a learning model generating the images, based on a text, in a direction desired by a user by repeating a process for forming an archive reflecting a user's own concept, based on the user's experiences and thoughts.