Metadata-Version: 2.1
Name: nesydep
Version: 1.0.0
Summary: NeSyDep: fast neural-symbolic discovery of data dependencies (FDs, CFDs, and more)
Keywords: functional dependencies,CFD,data profiling,data quality,data dependencies
Author: NeSyDep Authors
License: MIT License
         
         Copyright (c) 2026 NeSyDep Authors
         
         Permission is hereby granted, free of charge, to any person obtaining a copy
         of this software and associated documentation files (the "Software"), to deal
         in the Software without restriction, including without limitation the rights
         to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
         copies of the Software, and to permit persons to whom the Software is
         furnished to do so, subject to the following conditions:
         
         The above copyright notice and this permission notice shall be included in all
         copies or substantial portions of the Software.
         
         THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
         IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
         FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
         AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
         LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
         OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
         SOFTWARE.
         
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Science/Research
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: C++
Classifier: Topic :: Scientific/Engineering
Classifier: Topic :: Database
Project-URL: Homepage, https://github.com/Y-Staring/NeSyDep
Project-URL: Documentation, https://y-staring.github.io/NeSyDep/
Project-URL: Repository, https://github.com/Y-Staring/NeSyDep
Requires-Python: >=3.10
Requires-Dist: pandas>=1.5
Requires-Dist: numpy>=1.24
Requires-Dist: pydantic>=2.5
Requires-Dist: typer>=0.9
Requires-Dist: rich>=13
Provides-Extra: bn
Requires-Dist: pgmpy>=1.0.04; extra == "bn"
Requires-Dist: networkx>=3.0; extra == "bn"
Requires-Dist: joblib>=1.3; extra == "bn"
Provides-Extra: transformer
Requires-Dist: torch>=2.0; extra == "transformer"
Requires-Dist: sentence-transformers>=2.2; extra == "transformer"
Provides-Extra: lsh
Requires-Dist: datasketch>=1.6; extra == "lsh"
Provides-Extra: all
Requires-Dist: pgmpy>=1.0.04; extra == "all"
Requires-Dist: networkx>=3.0; extra == "all"
Requires-Dist: joblib>=1.3; extra == "all"
Requires-Dist: torch>=2.0; extra == "all"
Requires-Dist: sentence-transformers>=2.2; extra == "all"
Requires-Dist: datasketch>=1.6; extra == "all"
Provides-Extra: dev
Requires-Dist: pytest>=7.4; extra == "dev"
Requires-Dist: pytest-cov; extra == "dev"
Requires-Dist: ruff>=0.4; extra == "dev"
Requires-Dist: mypy>=1.8; extra == "dev"
Requires-Dist: pre-commit; extra == "dev"
Requires-Dist: mkdocs-material; extra == "dev"
Requires-Dist: mkdocstrings[python]; extra == "dev"
Description-Content-Type: text/markdown

# NeSyDep

**Ne**ural-**Sy**mbolic **Dep**endency discovery: fast mining of data dependencies
(functional dependencies, conditional functional dependencies, and more to come).

NeSyDep turns the research prototypes behind our papers into a usable, extensible
open-source tool for data scientists and data engineers:

- **BSFD** — fast FD discovery guided by Bayesian-network structure learning
  ([paper](https://github.com/Y-Staring/BSFD), ICDE 2026)
- **SCFDM** — fast CFD discovery via Transformer-guided relation partitioning
  ([paper](https://github.com/chen760316/CFDs-2026))

Mining kernels are C++ (exposed through pybind11); orchestration, sampling,
correlation extraction and evaluation are pure Python.

> Status: early development. See the roadmap below.

## Install

```bash
pip install nesydep            # core (FD/CFD mining + lightweight correlation)
pip install "nesydep[bn]"      # + Bayesian-network correlation (BSFD default)
pip install "nesydep[transformer]"  # + Transformer AttrFinder (large, pulls torch)
pip install "nesydep[all]"     # everything
```

## Quick start

```python
import pandas as pd
import nesydep as nd

df = pd.read_csv("hospital.csv")

# One-liner
fds = nd.discover(df, algo="bsfd", support=100, confidence=0.95)

# Full control
miner = nd.SCFDM(correlation="pearson", strategy="FD-First-DFS-dfs",
                 support=0.001, confidence=0.9)
result = miner.discover(df)
result.to_csv("cfds.csv")
```

CLI:

```bash
nesydep discover data.csv --algo bsfd -o fds.csv
nesydep algorithms            # list registered algorithms and parameters
```

## Roadmap

- **v0.1 (MVP)**: BSFD, SCFDM (lightweight correlation), TANE/DFD/CTane baselines,
  evaluation, CLI.
- **v0.2**: Transformer AttrFinder, all 10 SCFDM strategies, prebuilt wheels.
- **v1.0**: PyPI release, full docs, plugin tutorial.
- Later: GCFD and other dependency types (OD, DC, UC) as plugins.

Out of scope for v1: direct database connections, distributed (MPI) wheels.

## Citing

If you use NeSyDep in academic work, please cite the papers listed in
[CITATION.cff](CITATION.cff).

## License

MIT — see [LICENSE](LICENSE).
