AI Paper Atlas

Research map / Large language models / Agents, code and math

Challenges in the Automatic Analysis of Students’ Diagnostic Reasoning

Claudia Schulz, Christian M. Meyer, Iryna Gurevych

AAAI 2019 · Large language models › Agents, code and math · approach: Inductive and abstract reasoning evaluation

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use Concat rather than the other neural architecturesEpistemic activity identification · MeDM_A: 23.01 vs. 22.21 and 21.25Improved
Use Concat rather than the other neural architecturesEpistemic activity identification · TeDM_A: 21.68 vs. 19.96 and 18.34Improved
Use Concat rather than Multi-OutputEpistemic activity identification · MeDhamming loss: 0.07 -> 0.06Improved
Use Concat rather than Multi-OutputEpistemic activity identification · TeDhamming loss: 0.07 -> 0.06Improved
Use Concat rather than SeparateEpistemic activity identification · MeDM_O EE: 66.43 vs. 66.39No clear effect
Use a single-label preference baseline that ignores overlapsBiLSTM-CRF epistemic activity identification · MeDM_S EE: 75.46 vs. 79.96Hurt
Use a single-label preference baseline that ignores overlapsBiLSTM-CRF epistemic activity identification · TeDM_S EE: 77.21 vs. 78.87Hurt
Predict overlapping activities jointly rather than only one activity per tokenEpistemic activity identification · MeD and TeDM_OMixed
Replace neural architectures with the majority baselineEpistemic activity identification · MeD and TeDhamming loss: 0.07/0.06 -> 0.11Hurt
Evaluate with challenge-specific metrics in addition to overall hamming lossEpistemic activity identification evaluation · MeD and TeDchallenge diagnosisImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy