The algorithm used by this engine combines collaborative filtering with content-based techniques. The relative importance between the two methods is determined automatically, depending on the available data, so as to optimize performance. The collaborative filtering side of the algorithm is based on matrix factorization and uses the users' past interactions with the items. On the other hand, the content-based side allows numerical and categorical data to be included for both the users and the items, as well as tags.
This engine is particularly effective when a new user or item is included in a so-called cold-start scenario. A prototypical example of how this engine can be used is for movie recommendations: in addition to the user-item interactions (i.e., ratings), there are also descriptive fields about the users and the movies.
Accepted interaction types:
Tags extraction and training
The tags extraction is done by training the model with the Content Based engine. It generates the tags used for the rest of engines.
In order to make recommendations, the engine needs to be trained. Then the model is ready to make recommendations and will be able to calculate the recommendations for new users/items are imported and have not been trained yet.
You can find more details about the training in Training.
After the model has been trained, recommendations can be made. This engine can make recommendations for the following user-item combinations:
Model tags are required to apply diversification.
Every entity included in the list of recommendations comes with the expected interaction value for continuous and categorical interactions and a percent match score between 0 and 100 for the rest of interaction types (unary, binary, categorical) that indicates the degree of affinity between the recommended entity and the recipient.