Facebook, Inc.

Meta Platforms, Inc.

Shengbo Guo

James Li

Ying Zhang

Yixian Zhu

David Vickrey

Sara Lee Su

Jiun-Ren Lin

Khalid Bakry El-Arini

Lu Wang

Christopher John Leggetter

Mark Warren McDuff

Annie Hsin-Wen Liu

Michael Charles Bailey

Jiayi Wen

Seyoung Park

Yue Zhuo

Grace Louise Jackson

Kristin S Hendrix

Cheng Ju

Sijian Tang

Bram Wasti

Chenzhang He

Exemplary embodiments relate to the automatic generation of captions for visual media, including photos, photo albums, non-live video, and live video. The visual media may be analyzed to determine contextual information (such as location information, people and objects in the video, time, etc.). A system may integrate this information with information from the user's social network and a personalized language model built using public-facing language from the user. The personalized language model captures the user's way of speaking to make the generated captions more detailed and personalized. The language model may account for the context in which the video was generated. The captions maybe used to simplify and encourage content generation, and may also be used to index visual media, rank the media, and recommend the media to users likely to engage with the media.

Automatic personalized story generation for visual media

Systems, methods, and non-transitory computer readable media are configured to receive a uniform resource locator. A time and one or more features associated with the uniform resource locator can be provided to a first machine learning model. A prediction relating to a quantity of views the uniform resource locator achieves by the time can be received from the first machine learning model.

Systems and methods for scraping URLs based on viewport views

The disclosed computer-implemented method may include (1) maintaining a heterogeneous graph that represents (a) objects of a first type, (b) objects of a second type, and (c) relationships between the objects of the first type and objects of the second type, (2) using features of the objects of the first type to train a first embedding model to generate embeddings of the first type that (a) predict a label of objects of the first type and (b) predict, when combined with embeddings of the second type, graphical relationships in the graph, and (3) using features of each of the objects of the second type to train a second embedding model to generate the embeddings of the second type that predict, when combined with the embeddings of the first type, the graphical relationships in the graph. Various other methods, systems, and computer-readable media are also disclosed.

Systems and methods for using heterogeneous graphs to derive object embeddings

The disclosed computer-implemented method may include (1) sampling links from an online system, (2) receiving, from a human labeler for each of the links, a label indicating whether the human labeler considers a landing page of the link to be a low-quality webpage, (3) generating a link graph from a crawl of the links, (4) using the link graph to derive a graph-based feature for each of the links, (5) using the label and the graph-based feature of each of the links to train a model configured to predict a likelihood that a link is to a low-quality webpage, (6) identifying content items that are candidates for a content feed of a user, (7) applying the model to the content items to determine a ranking, and (8) displaying the content items in the content feed based on the ranking. Various other methods, systems, and computer-readable media are also disclosed.

Systems and methods for using link graphs to demote links to low-quality webpages

An online system ranks topic-groups for users and presents content items in topic-based feeds. A topic group corresponds to one or more topic(s) and can be used to generate a feed for presenting the content items related to the topic(s). For a particular user, the topic groups are ranked according to the likelihood of the user interacting with content items included in the topic groups. The topic groups are ranked using information of the users and/or users' historical interaction data such as click-based interaction data, post-based interaction data, or engagement-based interaction data. The online system generates and provides a user interface for presenting the topic groups to the client device. Content items that are related to the topic(s) corresponding to the topic group are presented in each topic-based feed such that the user can switch between different topic-based feeds.

Topic ranking of content items for topic-based content feeds

An online system receives posts that include links to various external pages and presents those posts to users of the online system. When the online system determines an opportunity to present a post to a particular viewing user of the online system, the online system determines a quality metric and an associated value score for the post. The quality metric is determined as a likelihood that the viewing user will view the external page for less than a threshold time period, and is used to adjust the associated value score. The online system compares the value score of the post to the value scores of other posts and selects one or more of the compared posts for presentation to the viewing user of the online system.

Selecting content with an external link for presentation based on user interaction with external content

Growing community of inventors

San Jose, CA, United States of America

Shengbo Guo