A collection of algorithms to model memory and retention of facts.

Install the nuget package in your project with dotnet add package obliviate.

Ebisu provides a simple model that must be attached with each fact the user is trying to memorise. See the notes on EbisuModel on choosing the parameters.

A learning/quizzing app will need to store the model, schedule reviews and keep it fresh with observations from each review session. Ebisu provides two primary APIs for these tasks. First, PredictRecall attempts to find recall probability of the existing model at a given time. E.g. will I remember this fact after X time units from the last review? Second, assume we reviewed the fact n times with k successful reviews after t time units from last review. UpdateRecall updates the previous model with these additional observations.

Ebisu provides fantastic documentation here. We highly recommend a read if you're planning to use the algorithm.

• Ebisu: https://fasiha.github.io/ebisu/ v2.0.0 (Public domain)
  • Ebisu v2.1.0 support with soft binary quizzes and half life rescale
• Memorize: https://github.com/Networks-Learning/memorize (MIT)
• Duolingo Halflife: https://github.com/duolingo/halflife-regression (MIT)
• SM-2 and related family of algorithms

We plan to support these algorithms along with benchmarks in future. Contributions and suggestions are most welcome!

MIT