hey! We use cookies only to ensure our site works.

Media Personalization

Used in recommendations personalization
for media e-commerce digital

The Media Personalization schema is used to capture very granular view and usage behavior. The data is used to train machine learning models for personalization and recommendation purposes: a personalized list of recommended articles and videos to read, based on a user's behavior. Media poses specific challenges, as something like a list of articles might not be sensitive in itself, but could fingerprint users based on specific patterns in their behavior.

Like every strm data schema it has two main components, event contract and the serialization schema.

Event Contract reference


Schema reference


Event contract overview

Key fieldsession_idThe session ID is used as key to atomically encrypt events
PII fieldcustomer_id1customer_id needs consent 1
PII fielddevice_id1device_id needs consent 1
PII fieldsession_id1session_id needs consent 1
ValidationsnoneNo validations applied

Serialization schema fields

event_typeIndicates an interaction with a recommended item
brand_sourceThe brand or media source, like the brand of a newspaper or website
platformThe platform of the user, e.g. web or app
osThe Operating System of the user, e.g. a OS X or iOS
versionThe OS version number, e.g. iOS 13.1.1
device_idA device ID like a UUID.
customer_idThe customer ID
session_idSession ID as set by producer, e.g. the browser cookie, or app open ID
context_idA unique identifier for the context (UUID)
swimlane_idA unique identifier for the lane
swimlane_rankDisplayed rank of the swimlane. TBD start at 1
article_rankTBD start at 1
article_in_viewContains time > x% of item was in viewport. TODO tbd.
article_actionclick, ...
followable_idA server-set item id for that specific item
followable_rankDisplayed rank of the item. TBD start at 1
recommendation_idAn unique identifier for a recommendation (UUID)