Abstract: In one embodiment, a method includes, by a client system, receiving a user input from the first user of an online social network, where the user input is associated with a subject and an intent, determining whether one or more agents associated with the intent are able to resolve the user input, identifying, responsive to determining that the one or more agents are not able to resolve the user input, one or more entities related to the subject, where each identified entity is connected to the first user via the online social network, and presenting, at the client system, one or more recommendations to the first user corresponding to the one or more entities.

Abstract: In one embodiment, a method includes, by a client system, receiving a user input from the first user of an online social network, where the user input is associated with a subject and an intent, determining whether one or more agents associated with the intent are able to resolve the user input, identifying, responsive to determining that the one or more agents are not able to resolve the user input, one or more entities related to the subject, where each identified entity is connected to the first user via the online social network, and presenting, at the client system, one or more recommendations to the first user corresponding to the one or more entities.

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: In one embodiment, a method includes receiving a user input by a user from a client system associated with the user, parsing the user input to identify a subject and intent of the user input, determining whether one or more agents associated with the identified intent are able to resolve the user input, identifying one or more entities related to the identified subject of the user input responsive to determining that the one or more agents are not able to resolve the user input, calculating a subject-association score for each of the one or more entities, ranking the one or more entities based on their respective subject-association scores, and sending, to the client system, instructions for presenting one or more recommendations to the user corresponding to one or more of the entities having a ranking greater than a threshold ranking.

Abstract: In one embodiment, a method includes initiating a communication session with a second client system associated with a second user via a communication network, wherein the communication session is initiated in a first modality, receiving a ping to the first client system from the communication network to evaluate available bandwidth on the communication network, estimating, by the first client system, an amount of bandwidth available on the communication network for use by the first client system, determining, by the first client system, the amount of bandwidth available on the communication network for use by the first client system is insufficient for the first modality, and switching the communication session with the second client system to a second modality by the first client system, wherein the second modality uses less bandwidth than the first modality.

Abstract: An online system receives tracking requests from client devices interacting with a website. The online system analyzes user interactions with websites using the tracking requests. The online system identifies missing parameters in the tracking requests and predicts values of the missing parameters. The online system may also identify parameters that are populated incorrectly and predicts their correct values. The online system uses the predicted parameter values for generating reports describing user interactions with the website. The online system predicts the values of the missing parameters based on metadata extracted from previous tracking requests received from client devices and also via web crawling of websites. The online system generates accurate reports based on the predicted parameters values.

Abstract: In one embodiment, a method includes initiating a communication session with a second client system associated with a second user via a communication network, wherein the communication session is initiated in a first modality, receiving a ping to the first client system from the communication network to evaluate available bandwidth on the communication network, estimating, by the first client system, an amount of bandwidth available on the communication network for use by the first client system, determining, by the first client system, the amount of bandwidth available on the communication network for use by the first client system is insufficient for the first modality, and switching the communication session with the second client system to a second modality by the first client system, wherein the second modality uses less bandwidth than the first modality.

Abstract: In one embodiment, a method includes, by one or more computing systems of a communication network, receiving, from a first client system associated with a first user during a communication session between the first client system and a second client system associated with a second user, a first user communication in a first modality, wherein the first user communication is to be sent via the network to the second client system. The method further includes determining an available bandwidth of the network with respect to the second client system, determining a second modality to send the first user communication to the second client system based on the available bandwidth, generating a second user communication in the second modality by converting the first user communication to the second modality, and sending, to the second client system the second user communication for presentation to the second user.

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: 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 is a method for identifying an action performed by a user in a third party system. Information associated with a form is received by an online system. For instance, hashed values of a plurality of form fields provided by a user and a description of the plurality of form fields are received by an online system. A form is identified based on the received information. Additionally, a determination whether one or more of the received hashed values correspond to stored values by the online system is made. If the received hashed values correspond to stored values in the online system, a user of the online system is identified based on the stored values corresponding to the one or more received hashed values. An identification of an action associated with the identified form and performed by the user in the third party system is stored.

Abstract: An online system receives tracking requests from client devices interacting with a website. The online system analyzes user interactions with websites using the tracking requests. The online system predicts an accurate label for the web page that caused the tracking request to be generated. The online system uses the accurate label for generating reports describing user interactions with the website. The online system predicts the label of a web page received by the client device based on metadata extracted from markup language documents by the client device and provided to the online system via tracking requests. Examples of metadata extracted from markup language documents include labels and description of widgets in the web page that triggered the tracking request from the client device. The online system generates reports describing the quality of the tracking requests.

Abstract: An online system receives tracking requests from client devices interacting with a website to analyze user interactions with the website. The website provides instructions with web pages sent to a client device that cause the client device to send tracking instructions to the online system. The online system sends requests for web pages to the website, receives a plurality of web pages from the website, and determines a count of distinct web pages provided by the website. The online system determines a score for the website indicating a quality of tracking instructions of the website based on various factors, including an aggregate value based on the distinct webpages of the website that include tracking instructions and the count of distinct web pages provided by the website. Based on this score, the online system generates a report describing a quality of the tracking instructions of the website.

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 is a method for identifying an action performed by a user in a third party system. Information associated with a form is received by an online system. For instance, hashed values of a plurality of form fields provided by a user and a description of the plurality of form fields are received by an online system. A form is identified based on the received information. Additionally, a determination whether one or more of the received hashed values correspond to stored values by the online system is made. If the received hashed values correspond to stored values in the online system, a user of the online system is identified based on the stored values corresponding to the one or more received hashed values. An identification of an action associated with the identified form and performed by the user in the third party system is stored.

Abstract: An online system receives tracking requests from client devices interacting with a website. The online system analyzes user interactions with websites using the tracking requests. The online system predicts an accurate label for the web page that caused the tracking request to be generated. The online system uses the accurate label for generating reports describing user interactions with the website. The online system determines a quality of tracking requests generated by the website based on various factors including a number of web pages of the website that generate tracking requests, the type of information provided by the tracking requests, and so on. The online system generates reports describing the quality of the tracking requests. The online system uses a metric indicating the quality of tracking requests of the website to determine whether to use predicted labels instead of labels provided by tracking requests for generating reports of the website.

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: A method is disclosed for attributing conversions among multiple members of a socially connected influence group, such as a household. Data from advertising impressions, including views and clicks, is maintained by an online system. When a conversion is made, the social network of the user creating the conversion event is analyzed. An influence group, defined as a group comprising the users and group of socially connected users whom influence the purchasing decisions of the first user, is created. Conversion data is analyzed for the first user and the other members of the influence group. This data is weighted to determine the propensities of successful conversions among all members of the influence group.

Abstract: An online system determines a metric indicating whether presenting a content item to various users increased a likelihood of other users performing a specific action associated with the content item. To determine the metric, the online system identifies a control set of users who are not presented with the content item and determines measures of affinity for users of the control set with other users of the control set and for users to whom the content item was presented. Based on measures of affinity for users of the control set and for users who were presented with the content item, the online system identifies segments of users of the control set having different measures of affinity for users of the control set and for users presented with the content item. The online system determines the metric bases on occurrences of the specific action by users in different segments.

Abstract: An online system predicts viewability of content items based on user interactions associated with the content items. The online system sends content items for display via client devices. The online system receives a request for a report based on viewability of the content item. The online system receives user interactions with the content item and determines a value of a user interaction metric based on the received user interactions. The online system provides the value of the user interaction metric as input to a correlation model to predict a value of the viewability metric for the content items. The online system may generates report based on the predicted viewability metric value.