Spaced repetition tool

This holds a commandline tool to help memorize 'things' via spaced repetition. Currently only text-based representations of information are supported. It can handle standard 'flashcard' style facts, but it is generalized to a card with any number of sides. Main use case for this is situations where multiple representations of the same information are to be learned, for example a number in binary, hexadecimal as well as decimal. Each unique representation gets assigned a unique identifier and transformation get one as well, this will allow for creation of questions that are not explicitly defined, but instead derived from following matching representation and transformation pairs.

Note from self (trying to use this over 2.5 years after initially writing it): Definitely a bit clunky and clearly one of my earlier Rust projects when I thought everything needed a trait, even though that abstraction is not used anywhere in the codebase. Use at your own risk =)

Quick Start

The example_files folder contains an English to French conjucation file. Generate the learnables by running

cargo run --example generate_from_text -- example_files/eng_fr_verbs.txt --output example_files/eng_fr_learnables.yaml

Run the tui example

cargo run --example tui -- /tmp/log.yaml example_files/eng_fr_learnables.yaml

Design

The traits.rs file describes the main concepts;

• Representation this represents a concept/fact to learn in a particular representation, so 0x13, 19 or 10011 would be different representations of the same Learnable concept.
• Transform A particular transform that can map one representation to another one, so hexadecimal to decimal, or decimal to binary.
• Selector is what algorithms should implement.
• Recorder can record (and load) past performance on questions.

Algorithms

Started with implementing the algorithm described in Enhancing human learning via spaced repetition optimization, but only the first half of the paper is implemented at the moment, it uses the forgetting curve that gets adjusted to weight random selection from the learnables. Currently, this Selector never declares the session complete, but still shows the hardest questions the most often.

The second algorithm (currently default in the cli example) is the SuperMemo2 algorithm. Which is well known and also implemented by other spaced repetition software. This Selector does terminate the session if there's no questions to be asked, this is the default.

Examples

The hosted example hosts a webserver with a minimalistic webinterface that one can use to practice. It can act as a iOS 'webapp' when added to the homescreen. It supports multiple users and each user can have multiple decks available to them. There is no listing of the available users, but this url would be for the user Ivor (case sensitive). If no user is provided default is used as a username.

The generate_deck example can be used to create the yaml files consumed by the hosted example or text ui example. Two types of files are accepted, a .txt file where each line is one learnable, with the answer delimited by a | character. Or a more elaborate yaml file. Examples for both are found in ./example_files/.

The decks used by the hosted example need to be created by running the following to convert the simple txt format:

cargo run --example generate_deck -- ./example_files/hex_dec_conversions.txt --output /tmp/hex.yaml

and converting the yaml file to the deck with:

cargo run --example generate_deck -- ./example_files/learnables_elaborate.yaml  --output /tmp/elaborate.yaml

The hosted example can then be ran with:

cargo r  --example hosted -- ./examples/hosted/example_config.yaml

After which going to http://localhost:8080/ should show these two decks for the default user.