Filter expressions - Amazon Personalize
Creating filter expressionsFilter expression examples

Filter expressions

To configure filters, you must use a properly formatted filter expression. Filter expressions are composed of dataset and property identifiers in dataset.property format, along with logical operators, keywords, and values. For values, you can specify fixed values or add placeholder parameters set the filter criteria when you get recommendations.

For a complete list of filter expression elements, see Filter expression elements. For examples of filter expressions, see Filter expression examples.

Note

Amazon Personalize ignores case only when matching event types.

Creating filter expressions

The general structure of a filter expression is as follows: 


EXCLUDE/INCLUDE ItemID/UserID WHERE dataset type.property IN/NOT IN (value/parameter)

You can either manually create filter expressions or get help with expression syntax and structure by using the Expression builder in the console. You can use filter expressions to filter items or users based on data from the following datasets:

  • Interactions: Use filter expressions to include or exclude items that a user has interacted with from recommendations (for example, user events such as click or stream). Or include or exclude users that have taken certain actions from user segments (interactions with certain event types). Interactions filters can't be used with Item-Attribute-Affinity recipe.

    Amazon Personalize considers up to 200 historical interactions for a user, and up to 100 streamed interactions you record for the user with the PutEvents operation. Additionally, the number of historical interactions Amazon Personalize considers for a user depends on the max_user_history_length_percentile and min_user_history_length_percentile hyperparameters you defined before training.

    For example, if you used .99 for the max_user_history_length_percentile, and 99% of your users have at most 4 interactions, Amazon Personalize will only filter based on the user's most recent 4 historical interactions. If a user has less than the number historical interactions at the min_user_history_length_percentile, Amazon Personalize doesn't consider the user's interactions when filtering.

    To filter based on up to 200 historical interactions for a user, set the max_user_history_length_percentile to 1.0 and retrain the model. For more information on hyperparameters, see Step 1: Choosing a recipe and navigate to the Properties and Hyperparameters section for your recipe.

  • Items: Use filter expressions to include or exclude items based on specific item conditions. You can't use filters to include or exclude items based on unstructured textual item metadata such as product descriptions.

  • Users:

    • For item recommendations for users, use filter expressions to include or exclude items from recommendations based on properties of the user you are getting recommendations for (the CurrentUser).

      If you have a Users dataset, you can add an IF condition to your expression to check conditions for the CurrentUser regardless of the dataset that is being used in the expression.

    • For user segments, use filter expressions to include or exclude users from user segments based on properties of users.

    For filtering recommendations,

When creating a filter expression, note the following limitations:

  • You can't use filters to include or exclude items based on unstructured textual item metadata such as product descriptions.

  • You can't chain Interaction and Item datasets into one expression. To create a filter that filters by Interaction and then Item datasets (or the opposite), you must chain two or more expressions together. For more information, see Combining multiple expressions.

  • You can't create filter expressions that filter using values with a boolean type in your schema. To filter based on boolean values, use a schema with a field of type String and use the values "True" and "False" in your data. Or you can use type int or long and values 0 and 1.

Filter expression elements

Use the following elements to create filter expressions:

INCLUDE or EXCLUDE

Use INCLUDE to limit recommendations to only items that meet the filter criteria OR use EXCLUDE to remove all items that meet the filter criteria.

ItemID/UserId

Use ItemID after the INCLUDE or EXCLUDE element for filtering item recommendations. Use UserId for filtering users from user segments.

WHERE

Use WHERE to check conditions for items or users. You must use the WHERE element after the ItemID or UserID.

AND/OR

To chain multiple conditions together within the same filter expression, use AND or OR. Conditions chained together using AND or OR can only affect properties of the dataset used in the first condition.

Dataset.property

Provide the dataset and the metadata property that you want to filter recommendations by in dataset.property format. For example, to filter based on the genres property in your Items dataset, you would use Items.genres in your filter expression. You can't use filters to include or exclude items based on textual item metadata.

IF condition

Use an IF condition only to check conditions for the CurrentUser and only once at the end of an expression. However, you can extend an IF condition using AND.

CurrentUser.property

To filter item recommendations based on the user you are getting recommendations for, in only an IF condition, use CurrentUser and provide the user property. For example, CurrentUser.AGE.

IN/NOT IN

Use IN or NOT IN as comparison operators to filter based on matching (or not matching) one or more string values. Amazon Personalize filters only on exact strings.

Comparison operators

Use =, <, <=, >, >= operators to test numerical data for equality.

Asterisk (*) character

Use * to include or exclude interactions of all types. Use * only for filter expressions that use the EVENT_TYPE property of an Interactions dataset.

Pipe separator

Use the pipe separator (|) to chain multiple expressions together. For more information, see Combining multiple expressions.

Parameters

For expressions that use = and IN operators, use the dollar sign ($) and a parameter name to add a placeholder parameter as a value. For example, $GENRES. For this example, when you get recommendations, you supply the genre or genres to filter by. For information on the number of parameters you can use, see Service quotas.

Note

You define a parameter name when you add it to an expression. The parameter name does not have to match the property name. We recommend that you use a parameter name that is similar to the property name and easy to remember. You use the parameter name (case sensitive) when you apply the filter to recommendations requests.

Filter expression examples

Use the filter expressions in the following sections to learn how to build your own filter expressions.

Item recommendation filter expressions

The following filter expressions show how to filter item recommendations based on interactions, item metadata, and user metadata. They are organized by data type.

Interaction data

The following expression excludes items based on an event type (such as click) or event types that you specify when you get recommendations using the $EVENT_TYPE parameter. 

EXCLUDE ItemID WHERE Interactions.EVENT_TYPE IN ($EVENT_TYPE)

The following expression excludes items that a user clicked or streamed.

EXCLUDE ItemID WHERE Interactions.EVENT_TYPE IN ("click", "stream")

The following expression includes only items that the user has clicked.

INCLUDE ItemID WHERE Interactions.EVENT_TYPE IN (""click")

Item data

The following expression excludes items based on a category or categories that you specify when you get recommendations using the $CATEGORY parameter. 

EXCLUDE ItemID WHERE Items.CATEGORY IN ($CATEGORY)

The following expression excludes items in the shoe category that do not have a description of boot. 

EXCLUDE ItemID WHERE Items.CATEGORY IN ("shoe") AND Items.DESCRIPTION NOT IN ("boot")

The following expression includes only items with a genre or genres that you specify when you get recommendations using the $GENRE parameter. 

INCLUDE ItemID WHERE Items.GENRE IN ($GENRE)

User data

The following expression excludes items with a genre or genres that you specify when you get recommendations using the $GENRE parameter, but only if the current user's age is equal to the value that you specify when you get recommendations using the $AGE parameter. 

EXCLUDE ItemID WHERE Items.GENRE IN ($GENRE) IF CurrentUser.AGE = $AGE

The following expression includes only items in the watch category, with a description of luxury, if the current user's age is over 18. 

INCLUDE ItemID WHERE Items.CATEGORY IN ("watch") AND Items.DESCRIPTION IN ("luxury") IF CurrentUser.AGE > 18

User segment filter expressions

The following filter expressions show how to filter user segments based on interactions data and user metadata. They are organized by data type.

User data

The following filter expression includes only users with a membership status equal to the value that you specify when you get user segments. 

INCLUDE UserID WHERE Users.MEMBERSHIP_STATUS IN ($MEMBERSHIP)

The following filter expression excludes users with an AGE less than 18. 

EXCLUDE UserID WHERE Users.AGE < 18

Interaction data

The following filter expression includes only users who have clicked or rated items.

INCLUDE UserID WHERE Interactions.EVENT_TYPE IN ("click, rating")

The following filter expression excludes users from user segments who have interactions with an event type you specify when you get user segments. 

EXCLUDE UserID WHERE Interactions.EVENT_TYPE IN ($EVENT_TYPE)

Combining multiple expressions

To filter by Items and Interactions datasets with one filter, combine multiple expressions together using a pipe separator (|). Each expression is first evaluated independently and the result is either the union or the intersection of the two results.

Matching expressions example

If both expressions use EXCLUDE or both expressions use INCLUDE, the result is the union of the two results as follows (A and B are different expressions):

  • Exclude A | Exclude B is equal to Exclude result from A or result from B

  • Include A | Include B is equal to Include result from A or result from B

The following example shows how to combine two expressions that use INCLUDE. The first expression includes only items with a category or categories that you specify when you get recommendations using the $CATEGORY parameter. The second expression includes items the user has marked as a favorite. Recommendations will include only items with the category you specify along with items that the user has marked as a favorite. 

INCLUDE ItemID WHERE Items.CATEGORY IN ($CATEGORY) | INCLUDE ItemID WHERE Interactions.EVENT_TYPE IN ("favorite")

INCLUDE and EXCLUDE example

If one or more expression uses INCLUDE and one more expression uses EXCLUDE, the result is the subtraction of the EXCLUDE expression result from the INCLUDE expression result as follows (A, B, C, and D are different expressions).

  • Include A | Exclude B is equal to Include result from A - result from B

  • Include A | Include B | Exclude C | Exclude D is equal to Include (A or B) - (C or D)

The following example shows how to combine an INCLUDE expression and a EXCLUDE expression. The first expression includes only items with a genre or genres that you specify when you get recommendations using the $GENRE parameter. The second expression excludes items that the user has clicked or streamed. Recommendations will include only items with a genre that you specify that have not have been clicked or streamed. 

INCLUDE ItemID WHERE Items.GENRE IN ($GENRE) | EXCLUDE ItemID WHERE Interactions.EVENT_TYPE IN ("click", "stream")