Abstract: A page of content includes instructions that, when executed by a client device presenting the page, obtain a visibility state of the page describing presentation of the page of content to a user and a time when the visibility state was obtained. Execution of the instructions also generates an identifier with which the visibility state and the time are associated. The instructions also obtained updated visibility states, obtains times when the updated visibility states are obtained, and associates the updated visibility states and their corresponding times with the identifier. The client device communicates the obtained visibility state, updated visibility states, and corresponding times to an online system in association with the identifier. Based on the times corresponding to the visibility state and the updated visibility states, the online system determines a duration the page was presented.

Abstract: In an embodiment, a system for accelerated monitoring of optical transceivers includes a monitoring unit included in a port interface module of a network switch. The monitoring unit is configured to receive a monitoring configuration, obtain status information from a plurality of optical transceivers connected to the port interface module at an instance based on the monitoring configuration, and store the obtained status information and at least one associated timestamp in a memory. The memory is included in the port interface module and configured to provide the stored status information to a requestor external to the port interface module.

Abstract: A chatbot can use a knowledge base including question/answer pairs to respond to questions. When a question is asked that does not correspond to a question/answer pair in the knowledge base, the chatbot can send the question to one or more humans to obtain an answer. However, only some people will have the experience, context, knowledge, etc., to answer the question. A model can be trained to select “experts” that are likely to be able to provide a good answer to a question by using both A) a vector comprising characteristics of questions and of the person posing the questions and B) a vector comprising characteristics of a possible expert. The model can trained to produce a value predicting how good an identified expert's answer is likely to be. The model can be trained based on measures of past answers provided for types of questions/questioners.

Abstract: A social networking system selects a set of contacts for presentation to a user of the social networking system. A contact is an additional user or a group of users maintained by the social networking system. To select the set of contacts, the social networking system generates scores associated with additional users of the social networking system and groups of users based on interactions between the user and various additional users and between the user and groups of users. Based on the scores associated with additional users and with groups of users, the social networking system selects a set of contacts from the additional users and the groups of users. Information identifying the set of contacts is communicated from the social networking system to a client device for presentation to the user.

Abstract: An apparatus for minimizing installation footprints of expansion cards may include one or more expansion cards that include a short edge, a long edge that is longer than the short edge and is substantially perpendicular to the short edge, and an edge connector disposed on the short edge. The apparatus may also include an expansion-card frame dimensioned to 1) guide an expansion card toward a printed circuit board of a computing device at a substantially vertical orientation such that the short edge of the expansion card is disposed proximate the printed circuit board of the computing device and the long edge of the expansion card extends away from the printed circuit board and 2) removably couple the edge connector disposed on the short edge of the expansion card to the printed circuit board of the computing device. Various other apparatuses, systems, and methods are also disclosed.

Abstract: The present disclosure is directed toward systems and methods for providing translations of electronic messages via a social networking system. For example, systems and methods described herein involve determining whether to provide an electronic message or a translation of the electronic message to a recipient based on social networking activities of the recipient. Furthermore, systems and methods described herein can provide a translation of an electronic message based on an analysis of social networking activities of one or more recipients of the electronic message.

Abstract: The present disclosure is directed toward systems, methods, and non-transitory computer readable media that dynamically modify content distribution campaigns based on triggering conditions and actions. In particular, systems described herein can provide a user interface for display to a publisher device that includes a plurality of selectable options for setting triggering conditions and/or actions. For example, the disclosed systems can utilize a machine learning model to generate suggested triggering conditions and/or actions for one or more content distribution campaigns of a provider. Moreover, the disclosed systems can generate custom rules based on selected triggering conditions and actions and apply the custom rules during execution of digital content campaigns. For instance, the disclosed systems can monitor performance of content campaigns, detect triggering conditions, and dynamically modify digital content campaigns based on actions corresponding to the triggering conditions.

Abstract: In one embodiment, a method includes extracting a document object model (DOM) for a content page, wherein the DOM comprises a hierarchical tree-based data structure. The method also includes identifying candidate nodes in the DOM based on a context of the nodes, wherein the candidate nodes may correspond to listing items. The method additionally includes for each of the candidate nodes, locating its parent and child nodes by traversing the DOM from the candidate node, extracting information from the candidate node and its parent and child nodes, and assessing whether the candidate node qualifies as a listing item based on whether the extracted information fulfills a required set of characteristics for a listing item.

Abstract: An online system matches a user across multiple online systems based on image data for the user (e.g., profile photo) regardless whether the image data is from the online system, a different but related online system or a third party system. For example, to match the user across a social networking system and INSTAGRAM™ system, the online system compares the similarity between images of the user from both systems in addition to similarity of textual information in the user profiles on both systems. The similarity of image data and the similarity of textual information associated with the user are used by the online system as indicators of matched user accounts belonging to the same user across both systems. The online system applies models trained using deep learning techniques to match a user across multiple online systems based on the image data and textual information associated with the user.

Abstract: Disclosed embodiments include a method of operation of a distributed network system. The method includes nodes of the network system that send messages over a protocol-independent message bus, and other nodes that receive the messages. Content from the received messages can be stored in a database distributed among nodes of the network system. At least some of the content stored in the database is published. The published content can be accessed by one or more applications to perform one or more functions.

Abstract: Systems, methods, and non-transitory computer-readable media can determine that a content item is being presented through a display screen of the computing device. Information describing one or more salient points of interest that appear during presentation of the content item are determined, wherein the salient points of interest are predicted to be of interest to one or more users accessing the content item. The presentation of at least a first salient point of interest is enhanced during presentation of the content item based at least in part on the information.

Abstract: Activity specifications of a plurality of activities to be monitored are received. Each activity specification of the activity specifications identifies properties of a corresponding activity of the activities to be monitored. A fingerprint specification of a computer security risk fingerprint is received. The fingerprint specification identifies a combination of two or more of the activities to be detected. A log of activities to identify occurrences of the activities to be monitored is analyzed. Based on the analysis, the computer security risk fingerprint in the log of activities is detected, including by detecting an occurrence of at least a portion of the combination of the activities identified by the fingerprint specification. A computer security action based on the detection of the computer security risk fingerprint is performed.

Abstract: In one embodiment, a system may access an image and generate a feature map for the image using a neural network. The system may identify regions of interest in the feature map. Regional feature maps may be generated for the regions of interest, respectively. Each of the regional feature maps has a first, a second, and a third dimension. The system may generate a first combined regional feature map by combining the regional feature maps. The combined regional feature map has a first, a second, and a third dimension. The system may generate a second combined regional feature map by processing the first combined regional feature map using one or more convolutional layers. The system may generate, for each of the regions of interest, information associated with an object instance based on a portion of the second combined regional feature map associated with that region of interest.

Abstract: In one embodiment, a method includes receiving a touch input within a particular region of a display area of the computing device. The display area presents a user interface (UI) including a number of views organized in a hierarchy. One or more of the views has a gesture recognizer configured to process a touch input detected within the regions of the display area associated with each of the views. Each of the views corresponds to one or more regions of the display area. One or more of the gesture recognizers has one or more intervention conditions. A particular one of the views is a default view to process the touch input. The method also includes evaluating the intervention conditions to identify a particular one of the views from among the number of views to process the touch input.

Abstract: An online system generates dynamically optimized sponsored content for a target user of the online system. Each sponsored content item comprises optimal creatives selected for a target user from a set of creatives provided by a content provider. Each type of creative (e.g. title, image, video, descriptive text), has a trained creative model to generate a prediction score for a creative of the same type based on the features of the creative and the characteristics of the target user. The prediction score of a creative indicates the likelihood that the target user will interact with a sponsored content item that includes that particular creative. The online system selects a creative for each type and assembles the selected creatives into a sponsored content item for the target user. For a different user, a different sponsored content item having different creatives chosen from the same set of creatives is generated.

Abstract: When a program function is called, if the instructions for that function are not in active memory, a page fault occurs. Resolving a page fault includes a costly process of loading a page of object code instructions, into active memory, including the instructions for the called function. Technology is disclosed to reduce page faults by placing interrelated functions near each other within executable code based on a log of previous function calls. A log of function calls may be from observing the execution of applications over time. Computing devices can compute where to place functions within executable code by: obtaining the function call log; building a call graph based on the function call log; defining multiple node clusters within the call graph; and generating an ordered list of functions by sorting the node clusters. The ordered list of functions can then be provided during linking to determine function placements.

Abstract: A system and method for multi-channel communication with dual in-line memory modules (“DIMMs”) is disclosed. The system retrieves information characterizing a plurality of memory channels, each of each is configurable to facilitate data communication between a DIMM and a memory controller with associated memory channel interfaces. Based on the retrieved information, one of the memory channels is designated as the active memory channel, granting the designated memory channel the ability to issue memory requests or transactions to the DIMM. On a periodic or as-needed basis (e.g., when the active memory channel is stalled or nearly stalled), the system determines whether to designate a different of the memory channels as the active memory channel, thereby enabling the newly-designated active memory channel the ability to issue memory requests or transactions to the DIMM. In some embodiments, only one of the memory channels is active at a time for communication with each DIMM.

Abstract: A method includes detecting a proximity event associated with a first user and a second user, wherein the proximity event includes the second user being in geographic proximity to the first user and calculating an influence score associated with the proximity event, wherein the influence score is based at least in part on a social gravity of the second user and a duration of the proximity event. The method further includes, upon determining that the influence score is greater than a threshold score, identifying, based at least in part on a geographic location of the first user, a content object associated with the second user for provision to the first user and sending the content object to a client system associated with the first user for display.

Abstract: Systems, methods, and non-transitory computer readable media are configured to determine a training set to train a machine learning model. A feature set for the model is determined. The model is trained based on the training set and the feature set to determine a score reflecting a probability that each user in an evaluation set of users is qualified for employment with an organization. A ranking of users in the evaluation set is provided based on the score determined for each user.

Abstract: In one embodiment, a method includes receiving a request to identify public sentiments for one or more entities from a second computing device, where each of the attribute n-grams represents an attribute among a plurality of pre-determined attributes, accessing a table of word vector relationships, looking up entity word vectors corresponding to each of the names for the one or more entities and attribute word vectors corresponding to each of the plurality of attribute n-grams using the table, calculating, for each of the entity word vectors, a similarity metric to each of the attribute word vectors, and sending a response message to the second computing device, where the response message includes calculated similarity metrics corresponding to all the pairs of an entity word vector and an attribute word vector.