SOLID Principles Assessment in UML Diagrams via LLMs

Oskar Picus's replication package provides data and scripts for evaluating SOLID principles in UML class diagrams using Large Language Models. The 69.1 MB package includes Excel files, Python scripts, and documentation, last updated in April 2026. It supports the replication of research on automated code quality assessment.

Use Cases

• Analyze the relationship between UML diagram features and SOLID principle compliance scores.
• Benchmark LLM performance on classifying design patterns or violations from textual or structured data in the Excel files.
• Train or validate models to predict code quality metrics from diagram attributes.
• Replicate experimental results on LLM-based assessment of object-oriented design principles.

Strengths

• Includes replication scripts (Python) and documented analysis procedures.
• Dataset is shared under a permissive CC-BY-4.0 license for open reuse.
• Packaged at 69.1 MB with multiple file formats (XLSX, PY, TXT) for different analysis stages.

Limitations

• Small scale at 69.1 MB, limiting the volume of training or evaluation data.
• Specific row counts, column names, and dataset structure are not detailed in the provided metadata.
• Temporal and geographic context for the source UML diagrams is unspecified.

Provenance

Source

Oskar Picus, published on figshare.

Collection Method

Created as a replication package for academic research on LLM assessment of SOLID principles.

Freshness

Last updated on 2026-04-08.