Downloads
Source Code
Anubis Twitter v2.5 prototype: 47 Python files, ~2,800 LOC. Stylometric modality for Arabic Twitter data.
GitHub RepositoryResearch Paper
AnubisX v3.0.0 Scientific Re-Architecture (1032 pages including appendices).
Download MarkdownDataset
31 Egyptian Twitter account JSON files included in the repository. Publicly available tweet data only. Used for prototype validation experiments EXP-001 through EXP-015.
Dependencies
- Python 3.10+
- FAISS (Facebook AI Similarity Search)
- sentence-transformers
- NumPy
- scikit-learn
- pandas
Known Issues
Hash non-determinism: Python hash() non-determinism across interpreter sessions affects character n-gram features (100 of 372 dimensions). A SHA-256-based replacement has been identified as a required fix and is planned for the next release. This issue affects reproducibility of feature vectors and should be considered when interpreting current experimental results.