Netflix Inc.

Microsoft Technology Licensing, LLC

Vijay Bharadwaj

Nelly L Porter

Himanshu Soni

Gopinathan Kannan

Carlos A Gomez-Uribe

Carlos Alberto Gomez Uribe

Mohammad Sabah

Siddharth Angrish

Ryan Segeberg

Antonio Molins Jiménez

Sasi Parthasarathy

Arthur H Baker

One embodiment sets forth technique for computing a similarity score between two digital items is computed based on interaction histories associated with global users and interaction histories associated with local users. Global counts indicating the number of interactions associated with each unique pair of digital items are weighted based on a mixing rate. The weighted global counts are then combined with local counts to compute total counts. An effective interaction probability indicating the likelihood of a user interacting with one digital item in the pair of digital items after interacting with the other digital item in the pair is computed based on the total counts. The effective interaction probability is then corrected for noise, resulting in a similarity score indicating the similarity between the pair of digital items.

Identifying similar items based on global interaction history

One embodiment of the present invention sets forth techniques for generating recommendation sets for a first client device. A recommendation system receives, from the first client device, a first selection of a first recommended item included in a plurality of recommended items. The recommendation system identifies a second recommended item included in the plurality of recommended items that has not been selected. The recommendation system retrieves an authenticated negative item from a plurality of authenticated negative items. The recommendation system stores one or more entries in a log file comprising a plurality of entries, based on at least one of the first recommended item, the second recommended item, and the authenticated negative item. One advantage of the disclosed techniques is that the use of authenticated negative examples, also referred to herein as authenticated negative items, provides a more relevant set of recommendations for the user.

Relationship-based search and recommendations via authenticated negatives

A data processing method comprises receiving title interaction data, wherein the title interaction data specifies, an order in which users interacted with a plurality of titles; generating a plurality of statistical models, each statistical model of the plurality of statistical models specifying a plurality of probabilities, wherein the plurality of probabilities represent, for each first title of the plurality of titles and each second title of the plurality of titles, a likelihood that a user will interact with the first title then next interact with the second title; refining the plurality of statistical models based on the title interaction data; determining a plurality of weight values corresponding to the plurality of statistical models for a particular user; identifying, for the particular user, one or more recommended titles of the plurality of titles based on the plurality of weight values and the plurality of statistical models.

Personalized Markov chains

Techniques are described for determining relationships between user activities and determining search results and content recommendations based on the relationships. A plays-related-to-searches application may determine a relationship score between plays of a media title and searches of a query by determining a distance between a projection of the search onto the space of the users and a projection of plays of the media title onto the space of the users. A plays-after-searches application may determine a score for plays of the streaming media title given the search by multiplying a number of times plays of the media title occur after the query is entered by the number of times any play occurs, and dividing by a product of the number of times plays of the media title occur after any query is entered and the number of times plays of any media title occur after the query is entered.

Relationship-based search and recommendations

In an approach, a method comprises receiving title interaction data, wherein the title interaction data specifies, an order in which users interacted with a plurality of titles; generating a plurality of statistical models, each statistical model of the plurality of statistical models specifying a plurality of probabilities, wherein the plurality of probabilities represent, for each first title of the plurality of titles and each second title of the plurality of titles, a likelihood that a user will interact with the first title then next interact with the second title; refining the plurality of statistical models based on the title interaction data; determining a plurality of weight values corresponding to the plurality of statistical models for a particular user; identifying, for the particular user, one or more recommended titles of the plurality of titles based on the plurality of weight values and the plurality of statistical models.

Growing community of inventors

Belmont, CA, United States of America

Vijay Bharadwaj