Abstract: A method for on-demand access to table partitions may include receiving, at a data execution engine, a query accessing a database table divided into a first partition and a second partition stored at different nodes of the database. The data execution engine may respond to the query by acquiring a lock on the database table and instantiating data structures to support the execution of the query without loading any of the partitions of the database table into a main memory. Instead, the loading of the individual partitions of the database table may be delegated to the operators tasked with performing the operations included in the execution plan for the query. Each operator may refrain from loading those partitions of the database table that are determined to be unnecessary for performing the individual operations. Related systems and computer program products are also provided.

Abstract: An AI platform mesh includes a machine learning control plane hosted in a container orchestration cluster. The machine learning control plane includes a worker plane manager and a set of machine learning services. One or more remote runtime environments configured to execute machine learning workloads can be registered with the machine learning control plane via the worker plane manager. The remote runtime environments can communicate with the set of machine learning services via the worker plane manager.

Abstract: Techniques and solutions are described for implementing virtual table schemas. A virtual table schema includes a logical pointer that is used to target a table that includes data and is defined according to the virtual table schema. Values assigned to the logical pointer can be used to target tables at different locations, such as in a federated database system or in a cache of a database management system of a central computing system. When data associated with the virtual table is requested, or prior to receiving the request, the data can be stored in a table in the cache. The logical pointer can be updated to reference the cache. If the cache is full, the table can be removed from the cache, and the logical pointer can be updated to reference a table at the federated database system.

Abstract: Systems and processes for generating and managing derivation definitions for calculating sustainability footprints are provided. In a method for generating derivation definitions for calculating footprints, inputs, rulesets, and/or coded processes are determined for the derivation definition. A sequence order for applying the rulesets and/or coded processes is determined for the derivation definition, and the derivation definition is stored in a repository for selection to calculate sustainability footprints.

Abstract: Techniques and solutions are described for automatically updating test code based on changes, including additions or deletions, to production code. Requests to analyze production code and a modification to production code are received. The modification is applied to the production code to provide modified production code. Test code specific for the production code is executed on the modified production code. It is determined that a test specified in the test code was failed by the modified production code. A condition responsible for test failure is identified, where the condition specifies at least one expected value or at least one expected result. A modification to update the test code is determined, where the modification includes an actual value or an actual result obtained during execution of the failed test. The modification to the test code is applied to provide modified test code.

Abstract: A computer implemented method can receive a metadata definition of a restricted measure pertaining to a database including a plurality of database tables. The restricted measure has a label, and the metadata definition includes one or more filter criteria configured to filter values contained in the plurality of database tables. In a report designer user interface for a report, the method can present the label of the restricted measure as an option based on the metadata definition. The method can receive a selection of the label of the restricted measure in the report designer user interface. Responsive to the selection, the method can link the metadata definition of the restricted measure to the report. When generated, the report requests access to the values contained in the plurality of database tables via application of the one or more filter criteria of the metadata definition.

Abstract: A computer-implemented method includes detecting attributes and values in rules contained in a rules set. Definitions of the attributes are determined from a data model associated with the rules set. Multiple different data entries having fields corresponding to the attributes are generated by populating the fields with data according to the values detected in the rules and the definitions of the attributes determined from the data model. A labeled dataset is formed using the data entries and logic contained in the rules. At least a portion of the labeled dataset is used to train a machine learning.

Abstract: A method for workload balancing is implemented in a computer system including a primary database system and a secondary database system, the secondary database system being configured as a backup system for the primary database system. The method includes compiling a query received from a client application. The query indicates one or more source tables have corresponding replicate tables. The compiling comprises replace the one or more source tables with the corresponding replicate tables. The method can generate a query execution plan based on the compiled query, determine a per-table result lag based on minimum commit times of the corresponding replicate tables, and compare the per-table result lag with a predefined threshold. Responsive to a determination that the per-table result lag exceeds the predefined threshold, the method can recompile the query by falling back to the one or more source tables from the corresponding replicate tables.

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 positions of 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: 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: The present disclosure relates to computer-implemented methods, software, and systems for managing asynchronous data replication in a multiple availability zone cloud environment. Metadata for files for asynchronous replication at a second availability zone is stored at an in-memory data grid of a first instance of a storage service at a first availability zone at a multiple availability cloud platform that provides storage services. The in-memory data grid includes a queue data structure of metadata records and a map of metadata records. In response to determining that connection from the first availability zone to the second availability zone is available, asynchronous data replication for files identified at the map is executed. A file for replication is identified at the map and provided for replication at a second file storage at the second availability zone through a replication interface of a second instance of the storage service at the second availability zone.

Abstract: Methods, systems, and computer-readable storage media for determining an initial distribution based on a set of survey boundaries and historical O-data, the historical O-data being generated through execution of a computer-implemented service, the historical distribution including a first set of parameter ranges and a second set of parameter ranges defining a set of range cells, adjusting one or more parameters ranges of at least one of the first set of parameter ranges and the second set of parameter ranges, determining an adjusted distribution defining a range space for a computer-based survey, receiving X-data from a set of users participating in the computer-based survey, the X-data including responses of users in the set of users provided to the computer-based survey, providing a set of X-data to O-data range vector correlations, and executing analytics of the computer-implemented service based on the set of X-data to O-data range vector correlations.

Abstract: The present disclosure involves systems, software, and computer implemented methods for creating line item information from tabular data. One example method includes receiving event data values at a system. Column headers of columns in the event data values are identified. At least one column header is not included in standard line item terms used by the system. Column values of the columns in the event data values are identified. The identified column headers and the identified column values are processed using one or more models to map each column to a standard line item term used by the system. The processing includes using context determination and content recognition to identify standard line item terms. An event is created in the system, including the creation of line items from the identified column value. Each line item includes standard line item terms mapped to the columns.

Abstract: The present disclosure involves systems, software, and computer implemented methods for providing an asynchronous transport channel for OData data using web sockets. One example method includes upgrading a HTTP connection to a web socket connection in response to an upgrade request. The web socket connection is maintained, after the upgrading, for receipt of subsequent web socket requests. An OData request is received, from a consuming application, over the web socket connection. The OData request is translated to a format used by an OData runtime environment and sent to the OData runtime environment. A response to the OData request is received from the OData runtime environment. The response is incorporated into a web socket message. The web socket message is sent to the consuming application, over the web socket connection, in response to the OData request, with the web socket connection remaining open after the web socket message is sent.

Abstract: Methods, systems, and computer-readable storage media for receiving an index and a random number from a verifier, receiving, from a shared data storage, encrypted supply chain data (SCD) of a supply chain actor (SCA), re-encrypting the encrypted SCD to provide re-encrypted SCD using a public encryption key of a decryption party, executing homomorphic operations on the re-encrypted SCD to provide an encrypted blinded result using second random numbers, receiving an encrypted modified blinded result, generating an encrypted blinded verification result using a verification function, the encrypted blinded result, the encrypted modified blinded result, and the random number, and sending the encrypted blinded verification result for decryption to provide a blinded verification result to the verifier, the verifier determining a verification result based on the blinded verification result and the random number, and comparing the verification result to a product claim.

Abstract: Methods, systems, and computer-readable storage media for receiving a set of inference results generated by a ML model, the inference results including a set of query entities and a set of target entities, each query entity having one or more target entities matched thereto by the ML model, processing the set of inference results to generate a set of matched sub-sets of target entities by executing a search over target entities in the set of target entities based on constraints, for each problem in a set of problems, providing the problem as a tuple including an index value representative of a target entity in the set of target entities and a value associated with the query entity, the value including a constraint relative to the query entity, and executing at least one task in response to one or more matched sub-sets in the set of matched sub-sets.

Abstract: Disclosed herein are system, method, and computer program product embodiments for routing communication to applications. In an embodiment, first application can transmit a connection to a second application without knowing the exact route address of the second application. An agent can intercept the connection from a first application intended for a second application. The agent can identify the route address for the second application, determine the protocol the connection was intended to be transmitted, and forward the connection based on the route address and protocol to the second application.

Abstract: In an example embodiment, a self-supervised learning task is used for training commonsense-aware representations in a minimally supervised fashion and a pair level mutual-exclusive loss is used to enforce commonsense knowledge during representation learning. This helps to exploit the mutual-exclusive nature of the training samples of commonsense reasoning corpora. Given two pieces of input where the only difference between them are trigger pieces of data, it may be postulated that the pairwise pronoun disambiguation is mutually exclusive. This idea is formulated using a contrastive loss and then this is used to update the language model.

Abstract: Disclosed is a framework for providing microservices for database with embedded container service in a container cloud platform. The framework includes a container cloud manager for performing distributed management of resources in a data center. The container cloud manager includes an orchestrator with a flexible job scheduler and operator pool. The flexible job scheduler and container cloud master form a flexible job management platform for managing jobs of the data center. The flexible job management framework can isolate various types of jobs as well as mix various jobs for efficient usage of hosts or resources in the data center. The operator pool includes optional components of a database. The orchestrator facilitates quick customization and release of database products, such as a HANA database with optional services, as requested by a user in the container cloud platform.

Abstract: The layers of a neural network model are traversed in sequence one or more times while generating a plurality of relevance scores each time based on neuron weights and neuron biases of the neuron network model. Each relevance score of the plurality of relevance scores quantifies a relevance of a neuron in a lower layer of the sequence of layers to a higher layer of the sequence of layers. One or more relevance vectors can be populated from the plurality of relevance scores generated at the one or more times. Each of the relevance scores in each relevance vector quantifies a relevance of one of the input features to a task for which the neural network model is trained to perform. An explanation of a behavior of the neural network as a whole is generated based on the one or more relevance vectors.