Facebook, Inc.

Meta Platforms, Inc.

Junbiao Tang

Avi Samuel Gavlovski

Zhurun Zhang

Ewa Dominowska

Anand Sumatilal Bhalgat

Hao Zhang

Hao Song

Yevgeniya Solyanik

Anwar Saipulla

James Theodore Kleban

David Benjamin Lue

Hua Chen

Jennifer Anne Abrahamson

Zhonghua Qu

Abhishek Agarwal

An online system receives multiple candidate components for including in content items to be presented to online system users. Upon identifying an opportunity to present content to a subject user of the online system, the online system dynamically generates an optimal content item for presentation to the subject user that includes one or more candidate components. Candidate components included in the optimal content item are associated with a predicted marginal effect on a performance metric associated the optimal content item. This marginal effect may be predicted using a machine-learned model that is trained using historical performance information about content items that were presented to viewing users of the online system having at least a threshold measure of similarity to the subject user and one or more features associated with candidate components included in these content items and in the optimal content item.

Selecting one or more components to be included in a content item optimized for an online system user

An online system receives multiple candidate components for including in content items to be presented to online system users. Upon identifying an opportunity to present content to a subject user of the online system, the online system dynamically generates an optimal content item for presentation to the subject user that includes one or more candidate components. Candidate components included in the optimal content item are selected by predicting an affinity of the subject user for each candidate component. The affinity of the subject user for a candidate component may be predicted using a machine-learned model that is trained using historical performance information about content items including the candidate component that were presented to viewing users of the online system having at least a threshold measure of similarity to the subject user. Components of content items used to train the model may be selected using a heuristic (e.g., Thompson sampling).

Generating data to predict user affinity for a component to be included in a content item optimized for an online system user

An online system maintains information identify a context in which sponsored content items were presented to users. A context in which a sponsored content item was presented to a user identifies additional content presented to the user prior to the sponsored content item, and may identify additional content presented in conjunction with the sponsored content item. The online system identifies users to whom at least one sponsored content item was presented in a context and generates characteristics for the context based on characteristics of users who were presented with at least one sponsored content item in the context. When the online system receives a request to present sponsored content items in the context that does not identify an online system user, the online system selects sponsored content items for the request based on the generated characteristics for the context.

Generating characteristics of users of an online system presented with content in a context relative to other content

Growing community of inventors

Mercer Island, WA, United States of America

Junbiao Tang