Welcome to Frappe’s documentation!¶
Frappe is a Django application that provides item recommendations for users. The simplest form of recommendations now are based on collaborative filtering tensor factorisation (TF). We use test.fm framework for computing the models. Frappe then implements a logic to use the models in order to serve recommendations. The framework ships with a set of filters and re-rankers to implement business rules such as diversity. Filters are modules that implement a peace of logic such as “filter out items that users has already installed”, whereas, rerankers implement more complex logic such as enforcing diversity or discounting values for some items.
The framework was originally designed for Firefox OS marketplace but is flexible enough to adapt to the most of the application domains. Within the framework it is possible to develop and deploy “recommendation” plugins that improve the recommendations, make additional business rules, etc.
Project’s home¶
Check for the latest release of this project on Github.
Please report bugs or ask questions using the Issue Tracker.