‘Microsoft Recommenders’ + ‘Linux Foundation of AI and Data’ = ‘recommenders-team’

Microsoft Recommenders was introduced a few years ago, and it provided guidelines and a collection of software libraries relating to recommender systems. Today I found the announcement that Microsoft Recommenders is joining forces with Linux Foundation of AI and Data (LF AI & Data). On GitHub, they created a new organization, “Recommenders Team“. Currently, this organization has two repositories, the “recommenders” repository, and the “artwork” repository.

Recommenders aims to aid researchers, developers, and enthusiasts in prototyping, experimenting, and transitioning both classic and advanced recommendation systems to production. It’s a project nestled under the Linux Foundation of AI and Data, housing a repository filled with exemplar practices and recommendations for constructing these systems, illustrated through Jupyter notebooks. The shared insights span five crucial tasks: data preparation for each algorithm, model building employing classical to deep learning algorithms like Alternating Least Squares (ALS) or eXtreme Deep Factorization Machines (xDeepFM), offline metric evaluation, hyperparameter tuning for model optimization, and transitioning models to a production sphere on Azure. Within Recommenders, utilities for common tasks like dataset loading as per algorithmic requirements, model output evaluation, and training/test data segmentation are provided. It encompasses implementations of several leading-edge algorithms for self-study and tailoring in individual applications, as detailed in the Recommenders documentation and further elucidated in the repository’s wiki page documents.

For more details visit https://github.com/recommenders-team/recommenders.