Abstract: A user interface may be caused to be displayed on a device of a user. The user interface may be configured to allow the user to create an application or web page. Instructions to configure a choice component of the application or web page may be received. The choice component may have a data source associated with one or more database objects of the database system and the choice component may have a display type defining how the choice component is displayable to users of the computing platform. The data source and the display type may be selectable by the user via the user interface. The choice component may be created or modified based on the instructions. The choice component may be published to a storage medium for transmission upon request to client devices accessing the application or web page.

Abstract: A plurality of database access requests are received at a database system within a designated database access scope. The plurality of database access requests may include a first request to read from a first database table and a second request to write to a second database table. The second request may occur during or after the receipt of the first request. A database access dependency rule linking the first database table and the second database table may be created. The database access dependency rule may impose a protected database access framework when executing a subsequent write request to the second database table when the first database table is updated. The protected database access framework may store one or more values associated with the subsequent write request in a staging table that are capable of being used to undo the subsequent write request.

Abstract: Systems and methods are provided for training object detectors of a neural network model with a mixture of label noise and bounding box noise. According to some embodiments, a learning framework is provided which jointly optimizes object labels, bounding box coordinates, and model parameters by performing alternating noise correction and model training. In some embodiments, to disentangle label noise and bounding box noise, a two-step noise correction method is employed. In some examples, the first step performs class-agnostic bounding box correction by minimizing classifier discrepancy and maximizing region objectness. In some examples, the second step uses dual detection heads for label correction and class-specific bounding box refinement.

Abstract: Messages are transmitted to first and second server groups in an on-demand computing services environment. The first message instructs the first server group to reduce a first resource allocation level associated with a network-accessible computing resource, while the second message instructs the second server group to increase a second resource allocation level associated with the resource. The resource allocation levels designate respective proportions of the network-accessible computing resource shared among the server groups. Request traffic associated with providing services via the on-demand computing services environment is transferred from the first server group to the second server group after decreasing the first resource allocation level and increasing the second resource allocation level.

Abstract: This disclosure relates to tools to facilitate the configuration of interactive agents, sometimes referred to as bots, chatbots, virtual robots, or talkbots. Specifically, the disclosure relates to the provision of a map view visualization of an interactive agent. The map view can display a complexity indicator or usage percentage for each dialog and provide an easy mechanism for creation of new dialogs, actions, parameters, rules, and logic.

Abstract: The present disclosure is directed to dynamically routing an event to a component of a hybrid application. For example, a method may include: detecting an event from a first component of a first component type of a hybrid application; transmitting a request to execute a function associated with the event to a plurality of components, the plurality of components being a combination of components of the first component type and components of a second component type different from the first component type, the first component being different from the plurality of components; dynamically determining which component of the plurality of components to assign to execute the function, the dynamically determining being based on which components of the plurality of components are available to execute the function and one or more rules; assigning the function to the determined component; and receiving a result of the function from the determined component.

Abstract: A method for managing features for a search system using declarative metadata. The method includes receiving search metadata including declarative statements identifying at least one search feature to be enabled across a plurality of components of the search system, performing functional verification of the at least one search feature, testing the at least one search feature, and enabling the at least one search feature in at least one of the plurality of components of the search system in response to positive functional verification and positive testing.

Abstract: The present disclosure is directed to integrating mobile application plug-ins. For example, a method may include: at run-time of a host application, initializing, by the host application, a plurality of plug-ins; in response to initializing the plurality of plug-ins, establishing a set of supported services provided between a plurality of components comprising a native component of the host application and the plurality of plug-ins; detecting a request for a service from a first component of the plurality of components; and in response to determining that the requested service is one of the supported services, routing the requested service to a second component of the plurality of components, the second component being different than the first component.

Abstract: A multi-tenant system stores a hierarchy of machine-learned models, wherein each machine-learned model is configured to receive as input a set of search results and generate as output scores for ranking the set of search results. Each machine-learned model is associated with a set of dimensions. The system evaluates search query performance. Performance below a threshold causes a new model to be generated and added to the hierarchy of models. Upon execution of a new search query associated with the same set of dimensions as the newly created model, the new model is used to rank that search query's search results.

Abstract: For a multi-tenant database accessible by a plurality of separate organizations, a system is provided for capturing organization specificities in a model for the multi-tenant database. The system includes a neural network. The system is configured to: receive an organization encoding for one or more separate organizations making previous search queries into the multi-tenant database; generate a vector matrix from the organization encoding to embed organization specificities for training a model of the neural network; and using the vector matrix, train the model of the neural network for processing a present search query into the multi-tenant database. In some embodiments, the model of the neural network is global across the separate organizations accessing the database.

Abstract: A metric data stream for a plurality of metrics may be retrieved from a database system. Each metric may measure a respective computing characteristic. The metric data stream may include a plurality of values for each of a sequence of time intervals. Each value may correspond with a respective one of the metrics. A plurality of metric correlation matrices may be determined for the metrics, each of which is associated with a respective time period in the metric data stream. A subset of comparison metric correlation matrices may be selected from the plurality of metric correlation metric matrices. A designated anomaly score may be determined for a designated time period by comparing a designated metric correlation matrix associated with the designated time period with the selected subset of comparison metric correlation metric matrices.

Abstract: Containerized workflow engines executing metadata for user-defined applications are described. A system utilizes user selections for configuring a user-defined application to identify application definition metadata. The system stores the application definition metadata for the user-defined application into a persistent storage. A workflow engine executes in a software container in response to receiving user context details and an invocation of the user-defined application. The workflow engine retrieves the application definition metadata from the persistent storage. The workflow engine inputs the user context details and executes the application definition metadata. The workflow engine outputs a result based on executing the application definition metadata.

Abstract: Methods, systems, and devices supporting a data model for emissions analysis are described. Some database systems may store emissions data and support a sustainability application. The sustainability application may display reports that track and analyze data related to carbon emissions. In some cases, underlying data for a report is missing from the database system. The system may support extrapolation techniques to estimate the missing data and aggregate the underlying data—including the extrapolated values—according to a data schema of the database to calculate fields in a report. In some cases, a single data record may be used to generate multiple reports. The system may send one or more results to a user device for display in a user interface (e.g., in one or more dashboards). Additionally or alternatively, the system can display underlying calculations (e.g., report calculations, extrapolation calculations, etc.) in the user interface to support auditing activities.

Abstract: System and methods for identifying a text word from a spoken utterance are provided. An ensemble BPE system that includes a phone BPE system and a character BPE system receives a spoken utterance. Both BPE systems include a multi-level language model (LM) and an acoustic model. The phone BPE system identifies first words from the spoken utterance and determine a first score for each first word. The first words are converted into character sequences. The character BPE model converts the character sequences into second words and determines a second score for each second word. For each word from the first words that matches a word in the second words the first and second scores are combined. The text word is the word with a highest score.

Abstract: Methods and systems for building a demo to interact with a customer including causing to display by a processor, a link for accessing the demo; initiating, by the processor, the demo by actuation of the link wherein the demo includes a set of webpage images, each configured as a screen with animations to mimic an interaction between a user and the customer; and presenting, by the processor, a storyboard of a dialog composed of a set of screens mimicking an ongoing customer dialogue for a service wherein the set of screens includes a first subset of webpage images of a customer view and the second subset of webpage images of a backend analytical view that are configured with animation to build on the ongoing customer dialogue based on mimicked data received from a suite of customer apps to enable the demo to have an appearance of actual customer exchange.

Abstract: A total number of messages in a set of messages that are processed by a computing node in a virtual computer cluster during a time interval is determined. The virtual computer cluster is deployed with a cloud computing service and includes the computing node and other computing nodes at an end time of the time interval. It is determined whether the total number of messages in the set of messages processed by the computing node is no less than a maximum per-interval message number threshold. In response to determining that the total number of messages is no less than a maximum per-interval message number threshold, the cloud computing service is caused to start an additional computing node in the virtual computer cluster after the time interval, wherein the additional computing node is used to process one or more subsequent messages to be processed after the time interval.

Abstract: A system and method for protecting access to metadata based on package membership. The system creates at least one customized metadata type for a first software package and assigns a package protected access mode to control access to the at least one customized metadata type by respective components of one or more software packages. The system denies access to the at least one customized metadata type, according to the package protected access mode, by at least one component of a second software package that shares a namespace with the first software package.

Abstract: Methods, systems, and devices supporting joint code validation across code repositories are described. Some systems may store code across multiple code repositories, where the code repositories support repository-specific pull requests and validations. However, code stored in one repository may depend on code stored in a different repository. To accurately validate changes to such code, a system may identify a validation trigger for a first pull request (e.g., indicating one or more code changes to a first code repository), determine a second code repository or second pull request related to the first pull request, and generate a test build across the code repositories. For example, the test build may include code updates in the first code repository according to the first pull request and code from the second code repository (e.g., updated according to the second pull request). The system may perform joint validation tests on the test build.

Abstract: Database systems and methods are provided for visual design and customization of processes for web applications. A method involves providing a page builder graphical user interface (GUI) display at a client device over a network, obtaining a configured set of configurable web components added to an aggregate component region on the page builder GUI display, generating behavioral code corresponding to the configured set of configurable web components, generating presentation code corresponding to a spatial arrangement of the configured set of the configurable components within the aggregate component region, creating an aggregate component object in a database to maintain an association between the presentation code and the behavioral code for the configured set of the configurable components, and updating code for the web page to include a reference to the aggregate component object at the aggregate component region of the web page.

Abstract: A method for data processing includes identifying, for communication of a content object, a segment of entities including entities of a first entity class of a plurality of entity classes defined by a data model that is configured for a tenant of a multi-tenant system, where the data model defines relationships between entity classes of the plurality of entity classes. The method may further include activating for selection at a user interface at least one second entity class that is related to the first entity class based on the relationships and identifying a set of additional entities of the second entity class from the remaining entities that are related to the segment of entities as defined by the data model. The method may include transmitting to a content communication system an indication of the plurality of entity identifiers corresponding to a modified segment of entities that includes the additional entities.