gorse

Why we included this project

Gorse is a self-hosted recommendation engine you can wire into your own product rather than routing user behavior data through a third-party service. You push items, users, and interaction events through its REST API, and it trains models on its own to produce personalized output such as related items or 'recommended for you' feeds. Under the hood it blends fresh-item feeds, collaborative filtering, and item-to-item similarity, and can fold text, image, or video embeddings into one multimodal pipeline, with classical and LLM-backed rankers available depending on your setup. A GUI dashboard lets you edit the pipeline, inspect tasks, and monitor online evaluation results from recent feedback, so teams see what the model is doing instead of trusting a black box. If you want self-hosted recommendations you can inspect, and your stack can call HTTP APIs (Go included), this saves you from building a recommender from scratch.