v0.1.1

🐛 GraphFlex – Bugfix Update

This minor update focuses on improving robustness of the BFS feature extraction logic when working with graphs that use dictionary-based neighbourhood representations.

🔧 What’s Fixed

Resolved an issue in extractor.py where graphs containing a dictionary neighbourhood were not handled correctly.
The BFS extractor now preserves a list of values per relationship key, enabling proper downstream processing and feature aggregation.

✨ Impact

Correct BFS traversal for heterogeneous or multi-relational graphs
Improved compatibility with graph backends that expose neighbourhoods as dictionaries
No breaking changes to the public API

📦 Installation / Update

pip install --upgrade graphflex

v0.1.0

📦 GraphFlex – Initial Release

We’re excited to introduce the first official release of GraphFlex – a Flexible Framework for Graph Feature Engineering in Python!

This initial version lays the foundation for seamless graph-based feature engineering, fully compatible with scikit-learn pipelines and modern graph backends such as DGL, Neo4j, and RDFLib-HDT.

---

✨ Highlights

• Modular GraphFlex class with plug-and-play architecture
• Built-in feature functions and postprocessing filters
• Scikit-learn compatibility: Pipeline, GridSearchCV, etc.
• Support for multiple graph backends via connector modules:
  • ✅ DGL
  • ✅ Neo4j (optional)
  • ✅ RDFLib-HDT (optional)
• Clean and extensible API for research and production use
• Optional dependency groups for flexible installation

---

📦 Installation

pip install graphflex