This page describes serving configs. A serving config is a serving entity that you associate with a model. It is used when generating your recommendation results.
Relationship with models
When you create a serving config, you select a model to attach. Serving configs are invoked by your site when surfacing recommendation results. Discovery for Media references the serving config's associated model at serving time to determine the recommendations to generate.
Models
A Discovery for Media Recommendations serving config can have a single model associated with it. However, any model can be associated with multiple serving configs, enabling you to deploy the same model on different pages via different serving configs.
Media Recommendations options for serving configs
The following options allow you to change the behavior of a Media Recommendations serving config.
Demotion
You can demote recommendations based on usage data or content age. Recommendations that meet the demotion criteria that you specify are demoted to the bottom of the results list.
Demotion based on usage data
You can choose one of a number of options for demoting recommendations based on usage data. If your content is a detail page, you can choose to have it demoted if the user has viewed it. If your content is playable media, you can choose to have it demoted depending on the amount of time the user played it or the percentage of the content that the user played.
Demotion based on content age
You can demote playable content based on its age. You set the age threshold in days, and any content that reaches that age is demoted to the bottom of the results list.
Diversification
If you want to ensure that results returned from a single prediction request are more diversified rather than looking similar, you can enable diversification. Generally, diversification reduces the likelihood that similar products are shown in a recommendation panel, at the risk of removing some good recommendations. Diversification settings can be edited after creating a serving config. It is disabled by default.
Two types of diversification are available: rule-based diversity and data-driven diversity.
Rule-based diversity
Rule-based diversity relies on categories of your datastore. Use rule-based diversity to recommend products from a variety of categories. Diversification is configured by level, with higher levels of diversification causing fewer items to be displayed per category. This diversification type works best if your datastore provides high-quality product categories.
| Diversification level | Maximum items per category |
|---|---|
| None | Unlimited |
| Low | 3 |
| Medium | 2 |
| High | 1 |
| Auto | Depends on datastore |
Data-driven diversity
Use data-driven diversity to produce recommendation results that balance relevance and diversity. Data-driven diversity learns from datastore metadata, such as titles or categories. Instead of relying on the title or category words, data-driven diversity captures semantic similarity to produce better-performing diversification.
| Diversification level | Maximum similar items |
|---|---|
| None | Unlimited |
| Low | 3 |
| Medium | 2 |
| High | 1 |
| Auto | Depends on datastore |