README

This repository was created for a Clio ML learning session on August 7, 2024.

It contains example code for different approaches to the problem of text alignment. Given different versions of the same text, how does one algorithmically match a transformed chunk of the text across the different versions?

These approaches range from the ultra-naive (approach_00.py) to the totally overcooked (approach_04.py). None of them are production-ready. Some may contain bugs, none are optimized for performance on large datasets, and all can be improved. However, they illustrate core concepts and how the solution complexity can grow.

Running the code

Use poetry with python 3.11.9.

Within the poetry shell:

• Run the tests: pytest ./alignment/tests/
• Run a specific approach: python ./alignment/approach_XX.py
  • The script will use the approach to align the sequences provided in example.py
  • It will print out results showing where the approach succeeded and where it failed

Extending the code

You can extend this project by increasing the complexity of the example (./alignments/example.py) or by adding a new approach (./alignments/approach_XX.py).

I recommend test-driven development for any new approach.