Multi-surface applications are applications where multiple surfaces share common infrastructure and user management, each surface focusing on particular product goals. Casual users are a segment of users who have limited activities on a given surface. Data for casual users is usually sparse and noisy, making it hard to capture the casual user’s true interest and to recommend items of interest to them. This disclosure describes a graph-based multi-surface recommender for casual user understanding and transfer, e.g., techniques of user understanding that transfer user interests, both latent and as disclosed by the users, across multiple surfaces. Leveraging centrally managed user accounts, a graph-theoretic understanding of the user is developed based on the diverse activities the user may have on different surfaces, each reflecting the user’s interests or preferences. The user’s interests are captured and modeled for personalization across surfaces.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.