Reports & Results

LLM Experiment Reports

• LLM Experiments: Extraction & Review

  Detailed report on using LLM agents for workflow extraction from research papers and systematic review of the resulting ATWL representations. Describes the multi-agent collaborative process, error patterns observed, and validation with fresh LLM instances.

  LLM_experiments_extract_review.pdf

• Systematic Review of 16 Workflows

  Full audit report: for each workflow, the LLM reviewer identified issues and proposed corrections; the expert evaluated each proposal. Documents the collaborative refinement process that shaped both the library and the language definition.

  review_16_workflows.pdf

• LLM Recommendations Report

  Report on the 2×2 recommendation experiment — comparative analysis of Prose vs. Formal library conditions across two problems. Corresponds to Appendix E of the paper.

  LLM_recommendations_report.pdf

Workflow Recommendations

• Recommendation: Bike-Sharing Analysis

  Complete recommended workflow for analysing spatio-temporal patterns of public bike usage and developing a relocation strategy. Natural-language description with library references per phase. Corresponds to Appendix F of the paper.

  recommendation_bikes.pdf

• Recommendation: Diploma Project (VIS Trends)

  Recommended workflow for a student's diploma project on research-topic evolution in IEEE VIS publications. Corresponds to Appendix G of the paper.

  recommendation_diploma_project.pdf