Introducing Potnia: A Python language library for the conversion of ancient texts to Unicode
summary
Potnia is an Apache‑2.0 Python library on PyPI that converts Romanized transliterations to Unicode using per‑script YAML mappings, tokenisation, regex rules and an API/CLI/Gradio GUI. It currently handles Linear B (corpus ~6,500 inscriptions) and Linear A (~1,200) and ships with ~350 pytest fixtures achieving 100% coverage for covered scripts. Users can choose optional regularisation; the tool successfully converted the entire Linear B corpus for embedding analysis and PCA visualization.
AI-written summary · about
chapters (21)
- 0:00Introducing Potnia
- 0:48Our research focus
- 1:51Why Potnia was created
- 2:35Name and symbolism
- 3:18Talk outline
- 4:40Challenges with transliterated texts
- 6:47Unicode for machine learning
- 7:50Problems with transliteration
- 9:14Existing tools limitations
- 10:18How Potnia bridges gaps
- 11:01Implementation details
- 12:04Tokenization process
- 13:27Regularization and annotations
- 14:51Flexibility for use cases
- 15:35Testing and validation
- 15:55User interfaces
- 16:37Supported scripts and future plans
- 16:58Research applications
- 17:39Case study: Linear B and PCA
- 19:02Access and contribute
- 19:23Q&A
related talks
29mHow to learn Japanese with Python
36m12 years of Pylint (or How I learned to stop worrying about bugs)
01h 39mNatural Language Processing (NLP) in Python - From Zero to Hero
35mPython, Locales and Writing Systems
29mOptimal Transport in Python: A Practical Introduction with POT
33mPython's Transition to Multilingual Programming
45mPython at Massive Scale
43mPL/Python -- Python inside the PostgreSQL RDBMS