Index

TRecSys is a tabular similarity search server. Install it into your project to have your very own recommendation engine.

You can:

  • customize the filters, encoders, and user encoders for the similarity search server.

  • index a list of items based on feature values

  • search by item to get the other items most similar to it

  • save your custom similarity search models to disk

  • load your custom similarity search models from disk

Note

TRecSys is created by argmaxml. We are focused on creating software the enables you to integrate recommendation engines into your product to increase customer engagement.

Check out the welcome for further information.

Welcome

TRecSys Documentation

TRecSys is a tabular similarity search server. Install it into your project to have your very own recommendation engine.

You can:

  • customize the filters, encoders, and user encoders for the similarity search server.

  • index a list of items based on feature values

  • search by item to get the other items most similar to it

  • save your custom similarity search models to disk

  • load your custom similarity search models from disk

Filters

HARD FILTERS. USUALLY BY GEO OR LANGUAGE

What they are Why use them How to use them Examples

Encoders

What they are Why use them How to use them Examples

User Encoders

What they are Why use them How to use them Examples

Index Values

each item should be a dict mapping an item feature to its value.

What does it mean to index values (what are the values) Why do it How to do it Examples

Query

Gets a single item and returns its k nearest neighbors. What does it mean to save to disk When do it Why do it How to do it

Save Model

Save your custom similarity search models to disk. What does it mean to save to disk When do it Why do it How to do it

Load Model

Load your custom similarity search models from disk.

What does it mean to load from disk (where does it load to) When do it Why do it How to do it

Get Started

Installation

Using pip

You can install TRecSys in several ways using pip. Choose one of the following.

Install from PyPl.

Install from GitHub.

Run Server

After installing, run a server. .. code:

python -m TRecSys

Browse to http://127.0.0.1:5000/docs.

You should see a swagger interface for the REST API.

The package is accessible via rest or python bindings.

FAQ

FAQ

Contribution Guidelines

Contribution Guidelines

Using TRecSys

Server

Server