Abstract: Systems and methods provide an experimentation system, or testing engine, for a real-time search infrastructure. An example method includes generating a snapshot of a production search environment and performing testing of a signal or index change in the snapshot. The change can be specified as parameters passed to the system. The method may include estimating an impact of the change based on the testing and determining, based on the estimate, whether the impact is positive. Responsive to determining the impact is positive the method may include generating a holdback environment of the production search environment, the holdback environment being a portion of the production search environment selected not to receive the change, pushing the change to the production search environment, and monitoring the change by running partial production traffic through the holdback environment.

Abstract: Systems and methods select and provide video snippets in a matrix interface. Example methods include obtaining a portion of a live video stream, assigning the portion to a class, determining that the portion is preview-eligible, and generating a snippet of the live video stream using the portion. Other example methods include determining, using a trained video classifier, a set of live broadcasts that have at least one portion that is classified as preview-eligible, calculating, for each of the live broadcasts in the set, an aggregate score for the live broadcast, selecting live broadcasts with highest aggregate scores, generating a snippet for each of the selected live broadcasts, and providing the snippets in a user interface. Methods may also include selecting portions of live video streams that are responsive to a user-provided parameter and generating a snippet for each live video stream that has a preview-eligible portion.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for limiting load on host servers that implement a social messaging platform. An example user device sends, to a platform comprising a plurality of host servers, a first request. The request is directed to a first endpoint. The user device receives, in response to the first request, a first error that indicates that the first request was not processed. The user device determines a back off time and places subsequent requests to the platform that are initiated before the back off time elapses and that are directed to the first endpoint in a back off queue in an order in which the subsequent requests are initiated. The user device sends, to the platform, the requests in the back off queue after the back off time has elapsed, until the back off queue is empty.

Abstract: A system, method, and computer-readable medium for identifying music-related data based on message data from a messaging platform. The method includes: obtaining a request from a client; identifying context data associated with the request; identifying message data from the messaging platform based on the context data; selecting, based on the context data and the message data from the messaging platform, music-related data for inclusion in a result set; and providing the result set to the client in response to the request.

Abstract: An example system includes a processor and a memory. The system performs sub-pixel convolution that is free of checkerboard artifacts. In one example implementation, the system may execute a method that includes initializing one or more parameters of a sub-kernel of a kernel and copying the one or more parameters of the sub-kernel to other sub-kernels of the kernel. The method may further include performing convolution of an input image with the sub-kernels of the kernel and generating a plurality of first output images. A second output image is then generated based on the plurality of first output images.

Abstract: A method for generating recommendations involves selecting a first platform message, making a first determination that the first platform message is potentially associated with a plurality of topics including a first topic and a second topic, obtaining additional information associated with the first platform message including at least one of information about an account that authored the first platform message and information about third party accounts engaging with the first platform message, making a second determining that the first platform message is associated with the first topic using the plurality of topics and at least a portion of the additional information, wherein the first topic is an initial classification of the first platform message, generating a recommendation for at least one account based on the second determination, and providing the recommendation to at least one account.

Abstract: A method, computer program product, and computer system for receiving, at a computing device, information associated with an entity from one or more social media sites. One or more attributes for the information associated with the entity is identified. A relevance profile associated with the one or more attributes is generated. A plurality of posts from the one or more social media sites is identified, wherein at least a portion of the plurality of posts includes at least a portion of the one or more attributes for the information associated with the entity. At least the portion of the plurality of posts is ordered on a display based upon, at least in part, the relevance profile associated with the one or more attributes.

Abstract: A system and method including: receiving an authorization request originating from an authorization module of an application executing on a client device, where the authorization request includes an identifier identifying the client device; causing transmission, based on the identifier, of a verification message to the client device, where the verification message includes a verification code; receiving a confirmation of the verification code from the authorization module of the application executing on the client device; authenticating the application based on the receiving the confirmation of the verification code; determining that the client device identified by the identifier corresponds to a user account including secure user data associated with a user; and transmitting a unique token verifying that the application is authorized to sign into the user account, where: the unique token uniquely identifies the user account to the application, and the secure user data is not shared with the application.

Abstract: A computing device can receive at least a first post in association with a first account, a second post in association with the first account, the second post being associated with the first post, a third post in association with a second account, the third post being associated with the first post, a fourth post in association with the second account, the fourth post being associated with the first post, and a fifth post in association with a third account, the fifth post being associated with the first post, determine that the first account and the second account are most relevant to a conversation, and based on determining that the first account and the second account are most relevant to the conversation, present the first post, the second post, the third post, and the fourth post without presenting the fifth post.

Abstract: A system and method for message analysis, including: receiving, by a control service, a first modification request to modify a file system of a computing device, wherein the computing device is operating in a read-only state; identifying, by the control service, a request parameter associated with the first modification request; determining, by the control service, that the request parameter satisfies a permission criteria to perform the first modification request; provisioning, by the control service, the computing device to operate in a read/write state in response to determining that the permission criteria has been satisfied, wherein the first modification request is executed to modify the file system while the computing device is operating in the read/write state; and, upon a determination that the first modification request has successfully completed, provisioning, by the control service, the computing device to operate in the read-only state.

Abstract: A real-time messaging platform allows advertiser accounts to pay to insert candidate messages into the message streams requested by account holders. To accommodate multiple advertisers, the messaging platform controls an auction process that determines which candidate messages are selected for inclusion in a requested account holder's message stream. Selection is based on a bid for the candidate message, the message stream that is requested, and a variety of other factors that vary depending upon the implementation. The process for selection of candidate messages generally includes the following steps, though any given step may be omitted or combined into another step in a different implementation: targeting, filtering, prediction, ranking, and selection.

Abstract: In some examples, a method includes receiving, from one or more client devices, a stream of messages composed by one or more users of the one or more client devices, wherein each of the messages includes a particular hashtag, determining, using a set of metrics that are based at least in part on the messages, a trending score that represents a magnitude of a trend for the particular hashtag, in response to determining that the trending score satisfies a threshold, sending, to a content provider system, a set of demographic data that describes one or more of the users who associated with the particular hashtag, and, in response to receiving, from the content provider system, targeted content that is based at least in part on the particular hashtag and the set of demographic data, sending, for display at the one or more of the one or more client devices, the targeted content.

Abstract: A method for determining a circle of trust (CoT) includes receiving a request for the CoT, generating the CoT for the context account by: identifying a primary graph with nodes based on at least one action within a social network for the context account. The method further includes performing random walks through the nodes of the primary graph, each of the random walks including two steps, ranking each of the nodes based on an amount of the random walks that end on each of the nodes, with the CoT including a number of the highest ranking plurality of nodes, filtering content items using the CoT to identify a subset of relevant items, and providing the subset for display on a client device.

Abstract: A system and method including: a request to display a message stream for a context account of a social media platform may be received by a client device; a preview video including a set of video snippets of a source video may be identified by the client device; automatic playback of the preview video in a playback area displayed inline with the message stream may be initiated; a user input indicating a request to play the source video may be received by the client device; and playback of the source video may be initiated in response to the request for the source video.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for delivering event content. One of the methods includes initiating, for one or more client devices, a subscription to a topic; receiving an event relevant to a topic; determining a number of channel servers that correspond to the topic, wherein each channel server is associated with one or more client devices; determining that the number of channel servers that correspond to the topic is greater than a threshold number of channel servers; writing the received event from the event processor to an event fan-out queue, in response to the determination that the number of channel servers that correspond to the topic is greater than the threshold number of channel servers; reading the received event from the event fan-out queue; and sending the received event to the one or more client devices associated with each respective channel server.

Abstract: According to an implementation, a method for digital information retrieval in a social media platform includes transmitting, over a network, information to render a timeline of social content for a user of a client application. The timeline of social content includes messages posted on the messaging platform by user accounts that are connected to a user account of the user in a connection graph. The method includes computing, using a machine-learning algorithm inputted with relevance signals, a relevance level between the user account of the user and a user account not linked to the user account of the user in the connection graph, and transmitting information about a profile of the user to a computing device associated with the user account not linked to the user account of the user in response to the relevance level being greater than a threshold level.

Abstract: Embodiments are provided for enabling a dynamic management of a multi-tenant distributed database. According to certain aspects, a management module supports an interface that enables a customer to configure one or more consistency models for a service to be supported by the distributed database. The management module may determine computing resources within the distributed database that are needed to support the service according to the configured consistency model(s), and may instantiate the computing resources for testing and development of the service by the customer.

Abstract: In some examples, a computing device includes at least one processor and at least one module, operable by the at least one processor to receive, from a client device of a user, a request for one or more advertisements to display at the client device with a set of messages. The set of messages is associated with the user in a social network messaging service. The at least one module may be further operable to determine a probability that the user will select a candidate advertisement using a machine learning model based on point-wise learning and pair-wise learning. The at least one module may be further operable to determine, based on the probability that the user will select the candidate advertisement, a candidate score for the candidate advertisement, determine that the candidate score satisfies a threshold, and send, for display at the client device, the candidate advertisement.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for providing related content on social networking platforms. One of the methods includes displaying multiple messages of a social networking platform, wherein each of the multiple messages of the social networking platform has an associated user-selectable user interface element; receiving, from a user, a first request for content related to a user-selected message, wherein the first request is submitted by selecting the associated user-selectable user interface element; in response to receiving the first request, processing the user's social network profile to obtain multiple user-related signals; processing the user-selected message to obtain multiple message-related signals; and obtaining, based on the multiple user-related and message-related signals, multiple terms and multiple relevant messages; and displaying, in a second client user interface, the multiple terms and the multiple relevant messages.

Abstract: Techniques of the disclosure are directed to a computing device creating and outputting, for display at client devices accessing a social media platform, targeted content. The computing device may receive candidate messages composed by users of a group of client devices, where the candidate messages each include a reference to the requisite product, brand, or market. If a candidate message has a determined interest score that satisfies a threshold interest score, the computing device includes the candidate message into a group of brand messages. The computing device may then send the targeted message to be output for display at another group of client devices, where the targeted message includes both an original portion and a carousel portion. The carousel portion includes a group of transitional windows, where each transitional window includes one of the brand messages from the group of brand messages.

Abstract: A method for distributed processing involves receiving a graph (G) of targets and of influencers, with each influencer related to at least one target, receiving an action graph of actions performed by one or more of the influencers, and key partitioning G across shards. The method further involves transposing the first graph (G) to obtain a first transposed graph (GT), value partitioning GT across the shards, storing the action graph on multiple shards, issuing, to a shard, a request specifying an influencer, to perform an intersection, receiving a response to the request of a set of influencers each of which is related to a target, and determining whether to send a recommendation to the target based on the response.

Abstract: A messaging platform forms an interactive message with one or more components and broadcasts the interactive message to client devices associated with accounts. The interactive message has a plurality of states that update based on event data received from a content source. While the event is in progress, the messaging platform receives event data relating to the event and transitions to an updated state based on the received event data. When the interactive message transitions to the updated state, the messaging platform updates the interactive message broadcast to the client devices. The interactive message may include interactive components that accounts may use to perform various interactions with the event, the content source, or each other.

Abstract: A real time messaging platform identifies an audio snippet of a time-based media (TBM) event. The messaging platform maintains a real time repository of concurrently broadcasting TBM events as well as a historical repository of previously broadcast TBM events. These repositories contain acoustic fingerprints of their respective TBM events. The messaging platform matches an acoustic fingerprint of the audio snippet with one of the stored acoustic fingerprints to identify the TBM event in the recorded snippet. To identify the TBM event, the messaging platform matches multiple overlapping reference audio segments of the reference audio stream with multiple test audio segments of the audio snippet. This allows the platform to account for time delays between the test and reference audio segments that would otherwise hinder the matching process.

Abstract: A method can include presenting a feed, the feed including at least a first post, a second post, and a third post; receiving a selection gesture in an area corresponding to the first post; in response to receiving the selection gesture, expanding a size of the first post from an original size to an expanded size; receiving a scrolling gesture; and in response to receiving the scrolling gesture, moving the first post, the second post, and the third post in a same direction as the scrolling gesture while maintaining the expanded size of the first post.

Abstract: A messaging system includes features of gathering information regarding content accessed across multiple applications and/or devices and making that information available to account holders of the messaging system. The messaging system stores data related to accessed content in a plurality of indices and makes that data available to enable account holders to access previously-accessed content, even if the previous access occurred in a different software application, on a different device, or on a device with a different operating system. Account holders may also be provided with recommendations based on gathered information.

Abstract: A real-time messaging platform allows advertiser accounts to pay to insert candidate messages into the message streams requested by account holders. To accommodate multiple advertisers, the messaging platform controls an auction process that determines which candidate messages are selected for inclusion in a requested account holder's message stream. Selection is based on a bid for the candidate message, the message stream that is requested, and a variety of other factors that vary depending upon the implementation. The process for selection of candidate messages generally includes the following steps, though any given step may be omitted or combined into another step in a different implementation: targeting, filtering, prediction, ranking, and selection.

Abstract: A method for using an in-message application. The method includes: receiving a broadcast message; identifying, in the broadcast message, a reference to an external data provider; obtaining an identifier of the in-message application from the external data provider; using the identifier to identify a set of components of the in-message application, where placement of the set of components is defined by a visual structure of the in-message application, and where each of the set of components is a user interface (UI) element; associating data obtained from the external data provider with a component of the set of components; and serving the broadcast message and the data to a consumer client, where the consumer client renders the in-message application based on the visual structure.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating avatars based on physical sensor measurements. One of the methods includes receiving, by a second user device from a video streaming platform system during a video streaming session between a first user device and the second user device, an updated physical sensor measurement of the first user device. An updated graphical representation of an avatar selected by a first user of the first user device is generated by the second user device based on the updated physical sensor measurement of the first user device. The updated graphical representation of the avatar is presented by the second user device on a display device of the second user device during a video streaming session between the first user device and the second user device.

Abstract: Implementations provide low-latency live-video streams using existing content delivery networks. An example method includes receiving a video broadcast as a series of frames and determining, for each frame, whether the frame is a break frame. Responsive to determining that the frame is a break frame, the method includes removing an in-progress tag from a current segment file in a playlist for the video broadcast. The playlist includes at least a previous segment file, the current segment file, and a next segment file, which also has a respective in-progress tag. The method also includes associating the frame with a next segment file in a playlist and transmitting the playlist to a cache server. Responsive to determining the frame in the series of frames is not a break frame, the method includes associating the frame with the current segment file. The frame is transmitted to the cache server as a chunk.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating and displaying contextual audiovisual content as a reaction video. One of the methods includes receiving a request from a user; and in response: receiving an input from the user identifying an incoming message of an incoming message stream coming into the user device from the platform, receiving an input from the user using the user device, wherein the input selects content from the incoming message as a subject content item; displaying the subject content item on a display of the user device; while displaying the subject content item, recording a user authored video; generating a reaction video, comprising generating a reaction video frame for each user authored video frame, and generating, as the reaction message, a social messaging platform message comprising the reaction video.

Abstract: A method for displaying messages includes receiving, from a social network system, a plurality of messages, and filtering the plurality of messages into at least two filtered sets of messages based on a group where each of the at least two filtered sets of messages include a different subset of the plurality of messages. The method can include assigning each of the at least two filtered sets of messages to a column, triggering concurrent display of the at least two filtered sets of messages in the assigned column, and analyzing data related to an interaction of a user with at least one filtered set of messages from the at least two filtered messages.

Abstract: A real-time messaging platform allows advertiser accounts to pay to insert candidate messages into the message streams requested by account holders. To accommodate multiple advertisers, the messaging platform controls an auction process that determines which candidate messages are selected for inclusion in a requested account holder's message stream. Selection is based on a bid for the candidate message, the message stream that is requested, and a variety of other factors that vary depending upon the implementation. The process for selection of candidate messages generally includes the following steps, though any given step may be omitted or combined into another step in a different implementation: targeting, filtering, prediction, ranking, and selection.

Abstract: A method for displaying messages receiving from a social network system, a plurality of messages, and filtering the plurality of messages into at least two filtered sets of messages based on at least one feature of the plurality of messages where each of the at least two filtered sets of messages includes a different subset of the plurality of messages. The method can include assigning a first of two filtered sets of messages to a first column, analyzing an interaction of the user with the first of the two filtered set of messages, and triggering display of a second of the two filtered sets of messages in a second column in response to the interaction.

Abstract: Various embodiments disclose an in-message application facilitating the use of third-party application programming interfaces (APIs). In particular, these in-message applications (referred to as “cards”) may be designed by a developer to access third-party API functionality. A user may then create a message referencing, e.g., a web address associated with the developer's card. The system may receive the user's message and recognize the reference. The system may then distribute an instantiation of the card to various recipients, the instantiation including elements generated using, or providing the user access to, the third party functionality.

Abstract: A system and method for providing content for broadcast by a messaging platform, including: receiving an identifier selected by a user of a client computing device during composition of a message; identifying a content provider associated with the identifier; providing, by a computer processor, a set of content items of the content provider for presentation by the client computing device, wherein the user of the client computing device selects a content item from the presented set of content items; and broadcasting the message with a reference to the selected content item to a plurality of accounts of the messaging platform.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing dynamic control of social messaging platform client-side and server-side behavior according to current and historical environmental, service, and application conditions.

Abstract: An example system includes prediction workers, training workers, and a parameter server. The prediction workers store a local copy of a machine-learned model and run the mode exclusively in serving mode. The training workers store a local copy of a machine-learned model and a local snapshot and run the local copy exclusively in training mode and compare the local model or state to the snapshot after training to send delta updates to the parameter server after training. The parameter server aggregates received delta updates into a master copy of the model, sends the aggregated updates back to training workers and provides two types of updates; a real-time update based on a comparison of the master model with a local snapshot, and a full update. The real-time update occurs at least an order of magnitude more frequently than the full update and includes a subset of the weights in the model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for detecting sources of computer network failures. One of the methods includes identifying a network flow in a computer network between a source and a destination; performing a first probe to determine whether there is end-to-end connectivity between the source and the destination; in response to determining that there is no end-to-end connectivity between the host and the destination, performing one or more additional probes including a second probe to determine whether each hop in the path of the network flow between the source and the destination is operational including requesting that the source transmit a respective first trace diagnostic packet to each hop in the path of the network flow; and determining whether at least one link of the computer network that is part of the path of the network flow has failed based on the results.

Abstract: A method includes selecting two or more frames from a plurality of frames of a video, downscaling the two or more frames, estimating a flow data based on an optical flow associated with the downscaled two or more frames, upscaling the flow data, generating a refined flow data based on the upscaled flow data and the downscaled two or more frames, upscaling the refined flow data, and synthesizing an image based on the upscaled refined flow data and the two or more frames.