Netflix Inc.

Justin Derrick Basilico

Prasanna Padmanabhan

Faisal Zakaria Siddiqi

Dong-Bang Tsai

Mohammad Hossein Taghavi

Gaurav Agrawal

Mehmet Yilmaz

Vivek Kaushal

David Gevorkyan

Richard Wellington

Ajinkya More

Christopher Alvino

Chao-Yuan Wu

Alexander J Smola

Gaurav Agrawa

In various embodiments, an optimization engine regenerates items included in an interactive page while the user is interacting with the interactive page. In operation, an optimization engine displays a portion of the interactive page during a viewing session. Subsequently, the optimization engine computes a probability distribution for the viewing session over a set of interests based on model parameters and operations performed by the user during the viewing session. The optimization engine then regenerates items that are included in a second portion of the interactive page based on the probability distribution for the viewing session. The optimization engine displays a least a part of the resulting regenerated interactive page. Advantageously, by regenerating items included in the interactive page based on operations performed by the user during the viewing session, the optimization engine reduces the time required for the user to view an item that piques an interest.

Regenerating an interactive page based on current user interaction

The disclosed computer-implemented method may include accessing updated data structures that are to be included in a user interface functionality test, where the updated data structures contribute to a user interface. The method may also include accessing live or snapshotted data captured from services running in a production environment, initiating generation of a first user interface instance using the updated data structures and using the accessed live or snapshotted data, and initiating generation of a second user interface instance using a different version of the data structures and using the same accessed live or snapshotted data. The method further includes comparing the first user interface instance to the second user interface instance to identify differences and then determine which outcome-defining effects the updated data structures had on the user interface based on the identified differences between the user interfaces. Various other methods, systems, and computer-readable media are also disclosed.

Page simulation system

A system for utilizing models derived from offline historical data in online applications is provided. The system includes a processor and a memory storing machine-readable instructions for determining a set of contexts of the usage data, and for each of the contexts within the set of contexts, collecting service data from services supporting the media service and storing that service data in a database. The system performing an offline testing process by fetching service data for a defined context from the database, generating a first set of feature vectors based on the fetched service data, and providing the first set to a machine-learning module. The system performs an online testing process by fetching active service data from the services supporting the media streaming service, generating a second set of feature vectors based on the fetched active service data, and providing the second set to the machine-learning module.

Feature generation for online/offline machine learning

A content provider system ranks media content items with respect to a particular user based on selection scores determined for each of the media content items. The selection scores may be determined using a particular model that calculates a predicted selection score based on feature values associated with the content item with respect to the particular user. The feature values may indicate properties of the media content item, the particular user, or the particular user's relationship with the content item, including information about the novelty of the media content item with respect to the user. The particular model may be trained with sample user consumption data points that represent various combinations of media content items and users. The data point information evaluated during the training of the particular model may cause the model to assign higher selection scores to content items that are novel in particular ways.

Media content rankings for discovery of novel content

Growing community of inventors

Saratoga, CA, United States of America

Justin Derrick Basilico