Inference & Profiling · Company practice
Inferences generated via ML and LLMs for personalization and advertiser audiences
The Times generates inferences about user preferences (topics, writers, teams) using machine learning and third-party large language models, using them for personalization, service improvement, Times marketing, and creating third-party advertiser audiences.
Supporting quotations
We infer new information from other data we collect, including using automated means such as machine learning, which may include third party machine learning services or large language models. These inferences include information about your likely preferences or other characteristics (e.g., article topics, writers, teams or markets of interest to you). As described below, we use these inferences for purposes of personalization, to improve our services, to market Times Services to you and to create audiences for third party advertisers and for marketing on other sites and services.
Qualifications & exceptions
- Employs automated means including machine learning and third-party large language models.
Inference & Profiling · Company practice
Audience models group users by attributes and pass random IDs to ad servers
Survey/registration data is combined with browsing tracking data to build data models measuring demographic and interest attributes, assigning users to groups identified by random IDs passed to ad servers.
Supporting quotations
We also identify groups of users to whom to serve personalized ads on behalf of our advertisers. To do this, we combine information we collect through surveys or registration with information we collect automatically using tracking technologies while you browse our sites and apps. This combined information is used to build models. These data models are then used to measure users’ attributes, like their demographic information or their interests. Working with service providers, we use these measurements to group users by common attributes. Each group is associated with a random ID which is then passed to our ad server for use in targeting ad campaigns on our sites and apps.
Qualifications & exceptions
- Groups associated with random IDs passed to ad servers.