FIELD NOTES TOPIC
automated grading
2 notes, newest first.
Automated grading of Linux/bash examinations using large language models: a four-level cognitive taxonomy approach
The paper evaluates if four large language models can accurately grade Linux/bash command responses in a way that approximates expert judgment. The authors use a four-level cognitive taxonomy to test the models on real…
Grading the Grader: Lessons from Evaluating an Agentic Data Analysis System
The research evaluates automated graders for agentic data analysis systems using LAMBDA, a multi-agent system, on 153 numerical QRData tasks. The study introduces a three-layer human-AI grading cascade and an iterative…
