Stepwise Sentence Purification Using AI and User Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for purifying malicious online content are inadequate in addressing psychological and emotional damage, failing to create a safe and positive online environment for free expression.
Innovation Solution
A method and device using generative AI for stepwise purification of sentences, employing a deep learning-based interactive model with reinforcement learning from human feedback (RLHF) to generate multiple purified sentences, allowing users to select the final purified sentence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If simple detection or single-step purification methods are used, then the system complexity is low, but the psychological and emotional damage caused by malicious content is not effectively addressed
Solution Approach 1:
The purification process is divided into multiple stages: detection of malicious content, generation of multiple purification candidates, user selection, and final purification. This segmented approach allows comprehensive addressing of psychological harm while maintaining manageable system complexity through modular design.
Solution Approach 2:
The system dynamically adapts the purification process based on user feedback and selections. Multiple purification candidates are generated dynamically, and the system adjusts to user preferences, enabling effective harm reduction while keeping the interface simple and user-controlled.
2Reliability
If multiple types of purification are applied stepwise, then the purification effectiveness is improved, but the processing time and complexity increase
Solution Approach 1:
Multiple purification candidates are generated in advance before user selection. This preliminary action allows comprehensive purification to be prepared upfront, reducing the need for iterative processing and minimizing actual processing time experienced by the user.
Solution Approach 2:
The system incorporates user feedback through selection of preferred purification candidates. This feedback mechanism allows the system to learn and adapt, improving purification effectiveness over time while optimizing processing efficiency based on actual user needs rather than applying fixed multi-step processing to all cases.
3Ease of operation
If user selection of purified sentences is implemented, then the user satisfaction and control are improved, but the interaction complexity increases
Solution Approach 1:
Multiple purification candidates are generated as copies of the original malicious content with different purification approaches. Users can review and select from these copies without the system needing to implement complex interaction logic, as the candidates are pre-prepared variations that can be directly compared and selected.
4Reliability
If generative AI models with RLHF are used, then the purification quality is improved, but the computational resources and energy consumption increase
Solution Approach 1:
The system uses generative AI models with RLHF to generate multiple purification candidates, which may be more processing than any single user needs. This excessive action ensures high purification quality and user choice, while the actual computational burden is distributed and optimized through efficient model deployment and candidate selection mechanisms.
Data Source
AI summary
Provided is a method of generating stepwise purified sentences using a deep learning-based interactive model. The method includes receiving input sentences by a user's terminal, detecting malicious words in the input sentences, and generating a plurality of stagewise purified sentences using a plurality of types of purification including purification of the malicious words according to stepwise sentence purification, by a language purification model part of a stepwise purified sentence generation system, and presenting, by the terminal, the plurality of stepwise purified sentences to a user to allow the user to select one of the plurality of stepwise purified sentences, and transmitting, by the terminal, the selected stepwise purified sentence to the stepwise purified sentence generation system to determine a final purified sentence.