Abstract: Source code is analyzed to identify components. The components are each assigned a complexity score. Documentation for the source code is identified, related to the components, and given a score based on the quantity of the documentation for the component and the complexity score for the component. To determine semantic meaning of the documentation, vector embeddings for the documentation languages may be generated and aligned. Alignment causes the different machine learning models to generate similar vectors for semantically similar words in the different languages. Since the vectors of the words of the other languages are similar to the vectors of the words in a primary language with similar meanings, the vector representation of the documentation in the other languages will match the vector representation of the source code when the documentation is substantially on the same topic.

Abstract: User-generated messages encapsulating selections from each of a plurality of entities for a selection task having a selection option are received via one or more interfaces presented by a selection processing system. For each selection, a different share of a signing key is obtained that has an identity associated with the selection task. For each selection, a different partial signature is generated by signing the messages using the associated share of the signing key. The partial signatures are combined into a threshold signature if a number of entities for a selection option specified by the task exceeds a pre-defined threshold. Data characterizing the results of the selection task can then be provided.

Abstract: A computer-implemented method includes representing a plurality of database tables as respective vectors in a multi-dimensional vector space, receiving an indication that a first database table represented by a first vector and a second database table represented by a second vector are related to each other, moving the respective vectors representing the plurality of database tables in the multi-dimensional vector space in response to the indication, and grouping the plurality of database tables into one or more table clusters based on positions of the respective vectors representing the plurality of database tables in the multi-dimensional vector space.

Abstract: Gradient Boosting Decision Tree (GBDT) successively stacks many decision trees which at each step try to fix the residual errors from the previous steps. The final score produced by the GBDT is the sum of the individual scores obtained by the decision trees for an input vector. Overfitting in GBDT can be reduced by removing the input values that have the least impact on the output from the training data. One way to determine which input variable has the lowest predictive value is to determine the input variable that is used for the first time in the latest decision tree in the GBDT. This method of identifying the low-predictive features to be removed does not require that earlier trees be regenerated to generate the new GBDT. Since the removed feature was already not used in the earlier trees, those trees already ignore the removed feature.

Abstract: Training data for a speech-to-text service can be generated according to a variety of techniques. For example, synthetic speech audio recordings for training a speech-to-text service can be generated in an automated system via linguistic expression templates that are input to a text-to-speech service. Pre-generation characteristics and post-generation adjustments can be made. The resulting adjusted synthetic speech audio recordings can then be used for training and validation. A large number of recordings can easily be generated for development, leading to a more robust service. Domain-specific vocabulary can be supported, resulting in a trained speech-to-text service that functions well within the targeted domain.

Abstract: Systems, methods, and computer media are described for accelerated fact checking using distributed storage platforms (e.g., a blockchain or other distributed ledger) and trusted software providers. A claim for fact checking is received from a user (e.g., organization) of a software application managed by the trusted software provider. User-specific data associated with both the user and the application that is related to the claim can then be accessed (e.g., retrieved from a data store through queries). It can then be determined whether the claim is supported based on the user's own data. If the claim is supported, verification data for the claim can be generated and stored in a distributed storage platform (e.g., blockchain), where the verification data is available for others to use in fact checking.

Abstract: In an example embodiment, a specialized software object, called a behavior toggle, is utilized for stateful applications in cloud environments. The behavior toggle, once enabled and used by an entity, is not able to be disabled anymore for that entity, other than by explicit migration. Unlike feature toggles, whose values are expressed using “true” or “false” flags or the like, behavior toggles are enabled by specific version. Specifically, the value indicates which version of the behavior is enabled. The behavior toggle software objects may then be referenced by, or included in, other software objects, to utilize the behavior toggle software objects with the other software objects and have the behavior toggle be applied to the other software objects.

Abstract: Embodiments facilitate efficient and flexible handling of metadata for processing in a microservice architecture (MSA) environment. A consumer service and a provider service both utilize a data model having metadata. A central repository stores metadata information that is being called for by a consumer service. A dedicated meta service manages the metadata repository by listening to a message queue to obtain model-relevant metadata information from a provider service. That model-relevant metadata information may be static in nature, or may be dynamic (e.g., varying as between different tenants). The meta service recognizes the static/dynamic nature of the model-relevant metadata information (e.g., based upon content such as a tenant ID or other information) in a message from the provider service, and stores the metadata accordingly. Once stored in the central repository, the model-relevant metadata information may be communicated directly to the consumer service.

Abstract: System and methods for enforcing service level agreements (SLAs) between computing platforms engaged in (e.g., Internet-of-Things) data exchange via peer-to-peer (P2P) connections are described. In accordance with various embodiments, data traffic via the P2P connections is monitored, and if an SLA violation is detected, data streams are rerouted to cure the violation.

Abstract: Pairwise entity matching systems and methods are disclosed herein. A deep learning model may be used to match entities from separate data tables. Entities may be preprocessed to fuse textual and numeric data early in the neural network architecture. Numeric data may be represented as a vector of a geometrically progressing function. By fusing textual and numeric data, including dates, early in the neural network architecture the neural network may better learn the relationships between the numeric and textual data. Once preprocessed, the paired entities may be scored and matched using a neural network.

Abstract: In an example embodiment, a data model infrastructure is implemented as a service rather than as an always-running server. Specifically, one of the technical issues with past implementations is that the data models are deployed onto a server that is intended to be “always running”, even if there are no requests to the server. This utilizes memory and processing power. While it may be useful to have an always running server for commonly used applications, for applications that are infrequently used (e.g., 10 times a month), it can mean that memory and processing power is wasted. Thus, by implementing the data model infrastructure as a service rather than an always-running server, the service can be launched only when actually needed, saving both memory and processing power.

Abstract: Methods, systems, and computer-readable storage media for receiving a capture file, the capture file including data representative of a workload executed in a source database system, and processing the capture file to provide a replay file, the replay file being in a format that is executable by a replayer to replay the workload in a target database system, processing the capture file including: processing the capture file to provide a set of intermediate files, and processing the set of intermediate files using in-memory merge sort operations to generate a set of replay files.

Abstract: Methods and apparatus are disclosed for extracting structured content, as graphs, from text documents. Graph vertices and edges correspond to document tokens and pairwise relationships between tokens. Undirected peer relationships and directed relationships (e.g. key-value or composition) are supported. Vertices can be identified with predefined fields, and thence mapped to database columns for automated storage of document content in a database. A trained neural network classifier determines relationship classifications for all pairwise combinations of input tokens. The relationship classification can differentiate multiple relationship types. A multi-level classifier extracts multi-level graph structure from a document. Disclosed embodiments support arbitrary graph structures with hierarchical and planar relationships. Relationships are not restricted by spatial proximity or document layout. Composite tokens can be identified interspersed with other content.

Abstract: A computer-implemented method can receive a software update for a software application deployable on a target system and perform an automatic performance test of the software application incorporating the software update on the target system. The automatic performance test can be integrated in a continuous integration and continuous delivery (CI/CD) pipeline. The method can collect a plurality of performance metrics measured when performing the automatic performance test and determine a performance score based on the plurality of performance metrics. Responsive to detecting the performance score is below a predefined performance threshold, the method can cause the CI/CD pipeline to reject the software update to the software application.

Abstract: Techniques and solutions are described for storing and retrieving metadata for database artefacts. A client request is received that includes a relevancy identifier. The relevancy identifier is useable to determine a subset of metadata that should be provided in response to the request. Metadata, including in metadata objects or database artefacts, can be reviewed to determine whether it is relevant to the request, such as if the metadata is associated with the relevancy identifier. Different requests from the same client can specify different relevancy identifiers, as, for example, a first set of metadata may be relevant to one client process and another set may be relevant to another client process. While some metadata can be specific to a particular client, at least some of the metadata can be accessed by multiple clients, and at least in some cases different clients can specify the same relevancy identifier.

Abstract: Systems and methods providing drill-down access to timeline events to a user by a user interface are disclosed herein. In some embodiments, a timeline is presented to a user by the user interface. The timeline may comprise a plurality of event cards comprising event data associated with a customer. The user may select an input associated with the event card. Upon selection of the input, a second plurality of event cards indicative of an event category and associate with the customer may be displayed.

Abstract: According to some embodiments, systems and methods are provided, comprising receiving, at a latency module, a request to provision a software system; determining a number of end-user locations; and in a case the determined number of end-user locations is one: determining a region of a data center for the determined end-user location; and deploying the software system to the determined region; and in a case the determined number of end-user locations is more than one: determining one or more non-technical parameters; determining one or more technical parameters; based on the determined one or more non-technical parameters and the determined one or more technical parameters, determining the region for the data center; and deploying the software system to the determined region. Numerous other aspects are provided.

Abstract: Technologies are described for determining between synchronous and asynchronous modes for computing service requests. Computing service requests are received by a computing service from clients. The computing service dynamically determines whether to use synchronous mode or asynchronous mode for processing the computing service requests. The computing service makes the dynamic determination of which mode to use (synchronous or asynchronous) based on various criteria, which can include synchronous/asynchronous mode recommendations generated by machine learning models and/or synchronous/asynchronous mode recommendations generated by static rules.

Abstract: A method and system are provided including at least one tabular data set, wherein each data set includes one or more attribute categories defining a record, each attribute category including one or more data values; an anonymization module; and an anonymization processor in communication with the anonymization module and operative to execute processor-executable process steps to cause the system to: receive a first data set at the anonymization module; determine at least one of the attribute categories in the first data set is a first-class identifier; generate a pseudonymous data set by removing the at least one first-class identifier attribute category from the first data set; determine at least two of the attribute categories in the pseudonymous data set are a second-class identifier; manipulate the at least two second-class attribute categories; and generate an anonymized data set based on the at least two manipulated second-class identifier attribute categories. Numerous other aspects are provided.

Abstract: Provided are systems and methods for implementing a database notification and statistics framework. In one example, the method may include receiving a message published by a database, the message including information about a database event performed by the database, determining a publication type of the message from among a plurality of publication types based on attributes included in the message, identifying one or more database components that have subscribed to the identified publication type of the message, and transferring the information about the database event included in the message to the one or more identified database components that have subscribed to the publication type of the message.