An explainable scoring and directional re-generation method and system based on implication reasoning chain for education evaluation scenarios
By constructing a formalized implied reasoning chain and a cross-attention mechanism, the problems of uninterpretability and resource waste in existing automatic scoring systems are solved, enabling interpretable scoring and targeted regeneration, and improving the transparency of scoring and the accuracy of learning feedback.
CN122152519APending Publication Date: 2026-06-05BEIJING PROSHINE TECH CO LTD
View PDF 0 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING PROSHINE TECH CO LTD
- Filing Date
- 2026-03-05
- Publication Date
- 2026-06-05
Smart Images
Figure FT_1 
Figure FT_2
Abstract
The application discloses an education evaluation scene-oriented explainable scoring and directional regeneration method and system based on implication reasoning chain. The method constructs an implication reasoning chain from student answers to scoring standards for each scoring point, and labels the implication relationship type and confidence at each step. The score is calculated through an implication chain strength function, which naturally supports partial scoring. The implication chain breaking point and contradiction point are coded as a structured defect descriptor, which drives the content generation module to perform directional regeneration on the defect interval and provides accurate learning feedback for students. The application is applied to education scenes such as test paper marking and composition scoring. Experimental verification shows that the QWK reaches 0.86, and the directional regeneration saves 78% of the computing resources.
Need to check novelty before this filing date? Find Prior Art