From Semantic Layers to AI-Native Data Interfaces

Manuscript source, figures, and supplementary data artifacts for:

V. M. Sangaraju, "From Semantic Layers to AI-Native Data Interfaces: Evolution, Challenges, and Governance," submitted to PeerJ Computer Science.

Repository structure

.
├── manuscript/
│   ├── SemanticLayerstoAI-NativeDataInterfaces.tex   # LaTeX source
│   ├── references.bib                                 # Bibliography
│   └── Fig1.jpg ... Fig9.jpg, Fig4.png                 # Figures (kept alongside the .tex so the
│                                                        # existing \includegraphics{FigN.jpg} paths
│                                                        # resolve without editing the source)
├── supplementary_data/
│   ├── table1_systems_taxonomy.csv / .md               # 151-system data-interface taxonomy, with per-row sources
│   ├── table2_benchmark_synthesis.csv / .md             # 41-study text-to-SQL/agentic benchmark synthesis, with per-row sources
│   └── Supplementary_Data_Methodology_Note.md            # Provenance & data-integrity notes (read first)
├── scripts/
│   ├── build_table1.py                                  # Regenerates Table 1 from source data
│   └── build_table2.py                                  # Regenerates Table 2 from source data
├── LICENSE
└── README.md

Supplementary data — read before citing

Both tables carry two extra columns beyond the original spec: Primary Source (a real citation — arXiv ID, DOI, ACL Anthology link, or official documentation URL — for that row) and Verified This Pass (whether that citation was freshly confirmed via web search during compilation, or recalled from general knowledge and flagged for a manual spot-check). Row counts are not round numbers by design — entries that couldn't be independently sourced were cut rather than kept.

supplementary_data/Supplementary_Data_Methodology_Note.md explains this in full, including exactly what was removed from an earlier draft and why. Read it, and treat any row marked Verified: No as "confirm before formally citing," before relying on this data independent of this repository.

Reproducing the supplementary tables

cd scripts
python3 build_table1.py   # writes table1_systems_taxonomy.{csv,md}
python3 build_table2.py   # writes table2_benchmark_synthesis.{csv,md}

Building the manuscript

The manuscript uses the IEEE Open Journal (ieeeojies) LaTeX class. Compile with:

cd manuscript
pdflatex SemanticLayerstoAI-NativeDataInterfaces.tex
bibtex SemanticLayerstoAI-NativeDataInterfaces
pdflatex SemanticLayerstoAI-NativeDataInterfaces.tex
pdflatex SemanticLayerstoAI-NativeDataInterfaces.tex

License

This repository is released under the Creative Commons Attribution 4.0 International License (CC BY 4.0) — see LICENSE. This matches PeerJ's default open-access license, so no relicensing should be needed on acceptance. Change it before pushing if you'd prefer different terms.

Contact

Venkata M. Sangaraju — sangaraju1988@gmail.com