Anki_Tagger

This is a project I started as a fourth year medical student. Anki was a huge part of my medical education, and this felt like a productive way to give back for everything others have done.

Big picture, this set of scripts will parse a lecture guide and identify the most relevant anki cards within a premade Anki deck.

While there are some fantastic deck’s out there to support medical education, aligning the content of these decks with preclinical curriculum is a persistent challenge and source of anxiety for students. This project aims to alleviate that stress by selecting the best cards for each lecture to study alongside their preclinical curriculum. Using the OpenAI API, I was able to quickly tag over 200 lectures to cover the entirety of M1 and M2 with minimal human intervention. While it is not the perfect solution, hopefully others can use and improve upon the code and make their own class tags!

Workflow

1. Create a personal OpenAI account, and obtain an API key.
2. In anki, export the deck you wish to tag as an anki_deck.apkg.
3. In anki, export the deck using the Notes as plain text funcion, and select to include a unique identifier: anki.txt
4. python embed_anki_deck.py Returns: anki_embeddings.csv This will create the embeddings of your deck: These are required for a first pass crude search of your deck to minimize API costs.
5. python make_learning_objectives.py or Returns: anki_learning_objectives.csv Create a list of summary learning objectives, the filename of the pdf will be the tag for the learning objective. Generally one lecture guide results in 10-30 questions.
6. python select_cards.py Returns: anki_cards.csv This will create a list of cards from your deck scoring them on their relevance to each learning objective.
7. python tag_deck.py Will tag the deck, and return the original deck apkg file.
8. Import into Anki and enjoy!

zachalmers