Systems and methods relating to a keyboard application on a mobile device are disclosed. The keyboard application generates a search query which it sends to a remote server. Data located based on the search query is received from the remote server. The data is associated with a record comprising a record identifier. The keyboard application associates a GUI element of the keyboard application with the record identifier. An input is received corresponding to selection of the GUI element. A unique tracking link is generated comprising a URI and an identifier associated with the selection of the GUI element. The URI is a deep link into an application that can execute on the mobile device. The unique tracking link is stored in a database in association with the record and user identifier and passed to the application to access content of the application.

Automated processes, computing systems, computing devices and other aspects of a data processing system provide improved reliability in delivering digital media content over the Internet or a similar wide area network without sacrificing data security. Content is initially placed into a secure format (e.g., secure hypertext transport protocol (HTTPS) via transport control protocol (TCP) or the like). Prior to transmission on the network, the secure data packets are encapsulated within connectionless frames, such as user datagram protocol (UDP) frames. The client device that receives the encapsulated packets extracts the underlying secure content from the connectionless frames for further processing. The encapsulation into connectionless data frames permits client and server devices to establish effective streaming sessions while preserving the security of the underlying data.

A client-side system can include a service proxy that can receive a request to perform a computing operation from a web application that is executable in a web browser of the client-side system. The service proxy can determine if the computing operation is executable by a local execution module that is external to the web browser and local to the client-side system. The local execution module may be different from the web application and may be configured to execute one or more computing operations using computing resources local to the client-side system. If the computing operation is executable by a local execution module, the service proxy can transmit a communication to the local execution module for causing the local execution module to execute the computing operation.

Communication between an owner of a lost tracking device and a finder of a lost tracking device is established. A tracking system receives an identifier of a tracking device and a request for a web page from a mobile device that has captured an image of a smart code on the tracking device. The smart code is embedded with a URL to the web page. The system displays the web page on the mobile device and prompts the user to provide a location of the mobile device. The location is associated with the identifier of the tracking device. A notification indicating the location of the tracking device is generated for the owner. The owner is prompted to provide information to establish contact with the finder of the device, and the web page is populated with the owner's provided information for display on the finder's mobile device.

Techniques are described for a contextual natural language understanding (cNLU) framework that is able to incorporate contextual signals of variable history length to perform joint intent classification (IC) and slot labeling (SL) tasks. A user utterance provided by a user within a multi-turn chat dialog between the user and a conversational agent is received. The user utterance and contextual information associated with one or more previous turns of the multi-turn chat dialog is provided to a machine learning (ML) model. An intent classification and one or more slot labels for the user utterance are then obtained from the ML model. The cNLU framework described herein thus uses, in addition to a current utterance itself, various contextual signals as input to a model to generate IC and SL predictions for each utterance of a multi-turn chat dialog.

Techniques are described for a contextual natural language understanding (cNLU) framework that is able to incorporate contextual signals of variable history length to perform joint intent classification (IC) and slot labeling (SL) tasks. A user utterance provided by a user within a multi-turn chat dialog between the user and a conversational agent is received. The user utterance and contextual information associated with one or more previous turns of the multi-turn chat dialog is provided to a machine learning (ML) model. An intent classification and one or more slot labels for the user utterance are then obtained from the ML model. The cNLU framework described herein thus uses, in addition to a current utterance itself, various contextual signals as input to a model to generate IC and SL predictions for each utterance of a multi-turn chat dialog.

An example device for retrieving media data includes a memory configured to store media data; and one or more processors implemented in circuitry and configured to: receive data indicating a packet loss signaling mechanism, the packet loss signaling mechanism comprising at least one of that segments are sent in chunks, that transport object identifiers (TOIs) are sequential, or that a last packet of an object assigned to a TOI has a specific flag set in a header of the last packet, a base URL, a maximum latency, or a synchronization point in a file delivery header; detect loss of a packet using the at least one of the packet loss signaling mechanisms, the lost packet including missing media data; generate a byte range request for the missing media data using information of the file delivery header; and deliver a proper media object to a media application.

Facilitating web page spectroscopy in a communications network is provided herein. A system can comprise a processor and a memory that stores executable instructions that, when executed by the processor, facilitate performance of operations. The operations can comprise receiving first data that describes a first communication packet flow and second data that describes a second communication packet flow. The operations can also comprise training a model based on the first data and the second data, as a result of which the model is trained to detect respective behaviors represented by the first data and the second. Further, the operations can comprise extracting a common parameter from third data that describes a third communication packet flow and fourth data that describes a fourth communication packet flow based on the model.

An illustrative computing system for a weblink content scanning system scans an electronic message for the presence of one or more weblinks. The computing system accesses, in a sandbox computing environment, content linked to the one or more weblinks. The computing system generates a hash of the accessed content and/or content linked to weblinks accessible via the accessed content. The computing system scans the content accessed via the one or more weblinks for a presence of malicious content and categorizes the scanned content accessed via the one or more weblinks (e.g., safe, malicious, and the like), associates the categorization with each corresponding hash, and saves such information to a data store for future analysis. Based on a result of this analysis, the computing system allows delivery of the original electronic message or generates a modified electronic message for delivery to a recipient device.

A communication control method which performs route control in a communication system comprising: a specific network constituting the Internet; a first network configured to accommodate a specific device connected to the specific network; a second network provided between the specific network and the first network; and a processing device configured to perform predetermined processing on the basis of a packet transmitted between the specific network and the first network, the communication control method comprising: causing a path setting device in the communication system to execute a communication route-setting process comprised of, in accordance with first routing information defining a path leading from the first network to the specific network to be branched in the second network, controlling a path so that a first path of the branched path is set as a path via the processing device, and a second path of the branched path is set as a path leading to the specific network.