Mab2Rec: Multi-Armed Bandits Recommender

Mab2Rec (AAAI'24) is a Python library for building bandit-based recommendation algorithms. It supports context-free, parametric and non-parametric contextual bandit models powered by MABWiser and fairness and recommenders evaluations powered by Jurity. The library is designed with rapid experimentation in mind, follows the PEP-8 standards, and is tested heavily.

Mab2Rec is built on top of several other open-source software developed at the AI Center at Fidelity:

• MABWiser to create multi-armed bandit recommendation algorithms (Bridge@AAAI'24, TMLR'22, IJAIT'21, ICTAI'19).
• TextWiser to create item representations via text featurization (AAAI'21).
• Selective to create user representations via feature selection (CPAIOR'21, DSO@IJCAI'21).
• Seq2Pat to create user representations via sequential pattern mining (AI Magazine'23, AAAI'22, Bridge@AAAI'23, Frontiers'22, KDF@AAAI'22, CMU Blog Post)
• Jurity to evaluate recommendations including fairness metrics (LION'23, CIKM'22, ICMLA'21).

An introduction to content- and context-aware recommender systems and an overview of the building blocks of the library is presented at AAAI 2024 and All Things Open 2021. There is a corresponding blogpost to serve as a starting point for practioners to build and deploy bandit-based recommenders using Mab2Rec.

Documentation is available at fidelity.github.io/mab2rec.

Usage Patterns

Mab2Rec supports prototyping with a single bandit algorithm or benchmarking with multiple bandit algorithms. If you are new user, the best place to start is to experiment with multiple bandits using the tutorial notebooks.

Quick Start

Single Recommender

# Example of how to train an singler recommender to generate top-4 recommendations

# Import 
from mab2rec import BanditRecommender, LearningPolicy
from mab2rec.pipeline import train, score

# LinGreedy recommender to select top-4 items with 10% random exploration 
rec = BanditRecommender(LearningPolicy.LinGreedy(epsilon=0.1), top_k=4)

# Train on (user, item, response) interactions in train data using user features 
train(rec, data='data/data_train.csv', 
      user_features='data/features_user.csv')

# Score recommendations for users in test data. The output df holds 
# user_id, item_id, score columns for every test user for top-k items 
df = score(rec, data='data/data_test.csv', 
           user_features='data/features_user.csv')

Multiple Recommenders

# Example of how to benchmark multiple recommenders to generate top-4 recommendations

from mab2rec import BanditRecommender, LearningPolicy
from mab2rec.pipeline import benchmark
from jurity.recommenders import BinaryRecoMetrics, RankingRecoMetrics

# Recommenders (many more available)
recommenders = {"Random": BanditRecommender(LearningPolicy.Random()),
                "Popularity": BanditRecommender(LearningPolicy.Popularity()),
                "LinGreedy": BanditRecommender(LearningPolicy.LinGreedy(epsilon=0.1))}

# Column names for the response, user, and item id columns
metric_params = {'click_column': 'score', 'user_id_column': 'user_id', 'item_id_column':'item_id'}

# Performance metrics for benchmarking (many more available)
metrics = []
for top_k in [3, 5, 10]:
    metrics.append(BinaryRecoMetrics.CTR(**metric_params, k=top_k))
    metrics.append(RankingRecoMetrics.NDCG(**metric_params, k=top_k))

# Benchmarking with a collection of recommenders and metrics 
# This returns two dictionaries; 
# reco_to_results: recommendations for each algorithm on cross-validation data
# reco_to_metrics: evaluation metrics for each algorithm
reco_to_results, reco_to_metrics = benchmark(recommenders,
                                             metrics=metrics,
                                             train_data="data/data_train.csv",
                                             cv=5,
                                             user_features="data/features_user.csv")

Usage Examples

We provide extensive tutorials in the notebooks folder with guidelines on building recommenders, performing model selection, and evaluating performance.

1. Data Overview: Overview of data required to train recommender.
2. Feature Engineering: Creating user and item features from structured, unstructured, and sequential data.
3. Model Selection: Model selection by benchmarking recommenders using cross-validation.
4. Evaluation: Benchmarking of selected recommenders and baselines on test data with detailed evaluation.
5. Advanced: Demonstration of advanced functionality such as persistency, eligibility, item availability, and memory efficiency.

Installation

Mab2Rec requires Python 3.8+ and can be installed from PyPI using pip install mab2rec or by building from source as shown in installation instructions.

Support

Please submit bug reports and feature requests as Issues.

License

Mab2Rec is licensed under the Apache License 2.0.