Automated Diagram Grading via Vectorized Entity Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual grading of diagrams in web-based learning and instructional settings is inefficient and lacks consistency, particularly for complex diagrams with multiple elements, making it difficult to provide real-time feedback and grading.
Innovation Solution
A software system that converts student submissions into vectorized entity representations, uses a pre-authored grading rubric to compute grades, and employs algorithms like Optimal Weighted Bipartite Matching and Munkres Assignment to automatically assess diagrams by matching elements and providing feedback, ensuring scalability for large and complex diagrams.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual grading is used for diagrams, then grading accuracy and feedback quality are maintained, but grading efficiency and scalability deteriorate
Solution Approach 1:
The system enables automatic self-grading of diagrams by converting student submissions into structured entity representations and using algorithms to autonomously compare these representations against solution criteria, eliminating the need for manual instructor grading while maintaining scalability
Solution Approach 2:
The patent replaces the mechanical process of manual human grading with an automated computational system that uses vectorized entity representation and matching algorithms to perform grading functions, significantly improving efficiency and reducing time loss
2Reliability
If brute force matching is used to interpret student submissions, then all possible interpretations are considered, but computational complexity becomes unscalable for diagrams with 10 or more elements
Solution Approach 1:
The patent segments the complex matching problem into manageable components by representing diagrams as structured entity lists with defined properties and relationships, allowing the system to process and compare diagram elements systematically without requiring exhaustive brute force evaluation of all permutations
Solution Approach 2:
The system changes the parameters of the matching algorithm by using optimized computational approaches that reduce time complexity from factorial (n!) to polynomial (n^4) scale, enabling reliable interpretation of diagrams with many elements while maintaining accuracy through structured entity comparison
3Loss of time
If real-time grading is implemented for complex diagrams, then feedback timeliness is improved, but system computational load increases
Solution Approach 1:
The system performs preliminary actions by pre-processing student submissions into optimized entity representations before comparison, and by pre-establishing solution criteria and matching rules, which reduces the computational energy required during the actual grading process and enables real-time feedback delivery
Data Source
AI summary
A software-based method and system enables the creation and automatic assessment of an image such as a free-body diagram by adding pre-made or user-created components to create the image. The input to the software may be a vectorized entity representation of constituent elements of a user's submitted image, including element properties. A pre-authored correct solution and a grading rubric may be used to compute a grade to be returned to the user as output. The assigned grade can be proportional to the severity of any errors in accordance with the grading rubric. The method and system may also generate feedback to the user indicating where and why grade points were lost.