gorse-io/gorse

AI powered open source recommender system engine supports classical/LLM rankers and multimodal content via embedding

Gorse – Open‑source Recommender System Engine

What it is – Gorse is a Go‑written, AI‑powered recommender‑system server. It ingests users, items and interaction data, trains recommendation models automatically, and serves personalized suggestions via a REST API or a web dashboard.

Key capabilities

  • Multiple recommendation strategies – latest‑item, user‑to‑user, item‑to‑item, classic collaborative filtering, etc.
  • Multimodal support – can handle text, images, video embeddings as item features.
  • AI‑powered pipelines – includes classical recommenders, optional reranker APIs and decision‑making APIs.
  • GUI dashboard – visual pipeline editor, system monitoring, data import/export.
  • RESTful API – CRUD for users/items/feedback and endpoints for real‑time recommendations.
  • Scalable architecture – single‑node or distributed deployment with a master node (training & config), server nodes (online inference) and worker nodes (offline batch recommendations). Data can be stored in MySQL/MariaDB, PostgreSQL or MongoDB; caches use Redis or the same databases.

Getting started

# Run a ready‑made demo (GitHub‑repo recommendations) in a container
docker run -p 8088:8088 zhenghaoz/gorse-in-one --playground
  • The container downloads a sample dataset, starts the dashboard at http://localhost:8088, and trains an item‑to‑item model.
  • You can add feedback (e.g., a user “bob” starring LLM‑related repos) with a simple curl POST to /api/feedback and then request ten recommendations for that user via /api/recommend/bob?n=10.

Typical use‑cases

  • Product or content recommendation for e‑commerce, media, or code‑hosting platforms.
  • Personalised feed generation where items have rich multimodal metadata.
  • Rapid prototyping of recommendation pipelines without building ML infrastructure from scratch.

Documentation & community

Project health

  • CI badge shows automated tests; code coverage badge indicates test coverage.
  • Actively maintained contributors (see GitHub contributors graph).
  • Open‑source license (not specified in the excerpt but typical for the project).

Bottom line – Gorse provides a ready‑to‑run, Go‑native recommendation engine that combines classic collaborative‑filtering methods with modern AI‑style embeddings, exposing everything through a clean API and a visual dashboard, making it a practical building block for any service that needs personalised suggestions.

Related

  • Project
    NVIDIA/recsys-examples

    A collection of optimized recommender models and GPU libraries for large-scale ranking, retrieval, and generative recommendation, featuring agent-driven PyTorch optimization.

  • Project
    gorse-io/gitrec

    GitRec is an open‑source recommendation system for GitHub repositories built on the Gorse engine. It provides a web UI, a Gorse admin dashboard, and Chrome/Edge/Firefox extensions that suggest repos based on a user’s starred projects or show related repos for popular projects. The service can be self‑hosted via Docker‑Compose, uses a MySQL data dump for initial repo data, and requires GitHub OAuth credentials and optional OpenAI settings.

  • Project
    go-kratos/blades

    A multimodal AI agent framework for the Go language that enables developers to build complex AI workflows with pluggable models, tools, and memory.

  • Project
    NVIDIA-Merlin/HugeCTR

    A GPU-accelerated recommender framework designed for the efficient training and inference of large deep learning models with massive embedding tables.

  • Project
    recommenders-team/recommenders

    Recommenders is a toolkit of classical and deep learning recommendation algorithms, along with utilities for data preparation, evaluation, hyperparameter tuning, and deployment. It helps researchers and developers prototype and bring recommendation systems to production.