alosi's Introduction

alosi

About

alosi is a Python package providing utilities for building and interacting with, components of the ALOSI adaptive learning architecture. These include:

Recommendation engine core utilities
Engine API
Bridge API

Setup

Install using pip:

pip install alosi

Or to install as an editable project:

git clone https://github.com/harvard-vpal/alosi
pip install -e ./alosi

Getting started

Recommendation Engine module

Implementing a recommendation engine

You'll need to subclass BaseAdaptiveEngine and implement all the empty methods. For example:

import numpy as np
from alosi.engine import BaseAdaptiveEngine

class LocalAdaptiveEngine(BaseAdaptiveEngine):

    def __init__(self):
        self.Scores = np.array([
            [0, 0, 1.0],
            [0, 1, 0.7],
        ])
        self.Mastery = np.array([
            [0.1, 0.2],
            [0.3, 0.5],
        ])
        self.MasteryPrior = np.array([0.1, 0.1])

    def get_guess(self, activity_id=None):
        GUESS = np.array([
            [0.1, 0.2],
            [0.3, 0.4],
            [0.5, 0.6]
        ])
        if activity_id is not None:
            return GUESS[activity_id]
        else:
            return GUESS
    
    # implement more methods here ...
...

# Usage

# instantiate a new engine subclass instance
engine = LocalAdaptiveEngine()

# Recommend an activity
engine.recommend(learner_id=1)

# Perform learner mastery Bayesian update for a new score
engine.update_from_score(learner_id=0, activity_id=0, score=0.5)

# Perform estimation and update guess/slip/transit matrices and prior mastery:
engine.train()

See the full example in examples/example_engine.py