Types of data Amazon Personalize can use - Amazon Personalize
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Types of data Amazon Personalize can use

The following topics introduce the different types of data that you can import into Amazon Personalize.

Interactions data

An interaction is an event that you record and then import as training data. Amazon Personalize generates recommendations primarily based on the interactions data. Interactions data can include the following:

  • Event type and event value data

  • Contextual metadata

  • Impressions data

You import interactions data into an Item interactions dataset. For more details about interactions datasets, see Item interactions dataset.

Item data

The item metadata that Amazon Personalize can use includes the following:

  • Numerical data about each item, such its price.

  • Categorical metadata about each item, such as the item's genre or color.

  • Creation timestamp data for each item.

  • Unstructured text metadata, such as product descriptions or movie synopses.

You import metadata about your items into an Items dataset. For more information about Items datasets, see Items dataset.

User data

The user metadata Amazon Personalize can use includes the following:

  • Numerical data about each user, such as their age.

  • Categorical metadata about each user, such as their gender or loyalty membership status.

You import metadata about your users into a Users dataset. For more information about Users datasets, see Users dataset.

Actions data

The action data Amazon Personalize can use includes the following:

  • The business value or importance of each action.

  • Categorical metadata for each action, such as seasonality or action exclusivity.

  • Action expiration timestamp data that specifies when Amazon Personalize should stop recommending each action.

  • Repeat frequency data that specifies long Amazon Personalize should wait before recommending each action after a user interacts with it.

You import data about your actions into a Actions dataset. For more information about Actions datasets, see Actions dataset.

Actions interactions data

The data Amazon Personalize can use from user interactions with actions includes the following:

  • Event type data

  • Categorical metadata

You import interactions data into an Action interactions dataset. For more details about Action interactions datasets, see Action interactions dataset.