Abstract: Techniques are described herein for generating, editing, and optimizing queries using neural networks. In some embodiments, the techniques include training a neural network using a set of performant database queries to automatically learn patterns between different sequences of tokens in performant queries. Once trained, the neural network may receive an incomplete query as input, where the incomplete query includes one or more query tokens. The trained neural network may then perform next token prediction to project a set of one or more additional query tokens that may follow the one or more query tokens in the incomplete query to form a completed, performant query.

Abstract: A device may receive a request to identify items that satisfy parameters of the request. The device may identify a plurality of items that satisfy the parameters. The device may generate a plurality of explanation sets. An explanation set of the plurality of explanation sets may relate to an item of the plurality of items. The explanation set may include at least one of: a positive explanation indicating that the item is positively associated with a first characteristic that relates to a first preference of a user, or a negative explanation indicating that the item is negatively associated with a second characteristic that relates to a second preference of the user. The device may select an item from the plurality of items based on the plurality of explanation sets. The device may provide information that includes an explanation set of the item selected.

Abstract: The data analysis server manages first data configuration information for managing a correspondence of the learned data model, a raw data bucket that stores the raw data for generating the learned data model, and a curated data bucket that stores the curated data for generating the learned data model and second data configuration information for managing correspondences of the learned data model and the first storage area, the curated data bucket and the second storage area, and the raw data bucket and the third storage area, and gives an instruction to acquire snapshots of the second storage area that stores the curated data for generating the learned data model and the third storage area that stores the raw data for generating the learned data model to the data storage via the storage management server when the learned data model is generated.

Abstract: The present technology provides an interface and system for generating a previous version of a directory structure that can be useful in the restoration of a directory or file. The system, methods and non-transitory computer-readable storage media include receiving a request for a previous version of a directory structure at an indicated date, querying a content management system for placeholder content items associated with the previous version of the directory structure at the indicated date, receiving the placeholder content items, and generating a read-only copy of at least a portion of the previous version of the directory structure at the indicated date based on the placeholder content items, wherein the read-only copy of the directory structure is separate from the directory structure at a current date.

Abstract: Data reduction services are implemented on one or more nodes of an IIoT data pipeline to intelligently determine an appropriate data reduction strategy based on characteristics of the incoming data. In one or more embodiments, data reduction components on the pipeline node or on an edge device define different data filtering rules or algorithms that are selectively applied to streaming time-series data based on a probability distribution of the data. The data pipeline node performs real-time distribution analysis on the streaming data to determine whether the data has a unimodal distribution, a multimodal distribution, or no mode, and selects one of the data filtering rules based on this determined probability distribution. In this way, the data is intelligently reduced in a manner that retains critical information within the reduced data set while achieving a high level of data reduction.

Abstract: Method and system for correlating multiple device entries within a domain is disclosed. The method may include receiving a first correlation key comprising a set of device properties associated with a first device from a first plugin. The method may further include comparing the first correlation key with each of a plurality of second correlation keys stored in a device database. The method may further include determining whether a second correlation key from the plurality of second correlation keys in the device database corresponds to the first device, based on the comparing. The method may further include selectively correlating the first device with a second device from the plurality of second devices mapped to the second correlation key in the device database, based on the determining.

Abstract: Transmission handling of analytics query response includes a search head, in a data intake and query system, receiving a query from an analytics system. The search head distributes at least a portion of the query to at least one indexer for processing the query. The at least one indexer transmits, bypassing the search head, and to the analytics system, events matching the query. The search head receives from the at least one indexer, data regarding the events, and sends the data regarding the events to the analytics system.

Abstract: Systems and methods are provided that allow federated data from various source systems to be accessed and analyzed through a data analysis platform. The federated data may be stored in different formats. The data analysis platform can receive the federated data in whatever format it has been stored at its respective source system. A script can be used to generate temporary representations (or temporary objects) for the federated data by transforming the federated data. Moreover, the temporary representations or temporary objects can be further transformed into a data analysis platform-specific format. A user of data analysis platform may access and/or manipulate either the temporary representations or objects as well as the data analysis platform-specific objects. Temporary objects can be transformed automatically into corresponding platform-specific objects when necessary to provide an enhanced capability or operation on the objects.

Abstract: The present disclosure relates generally to storing computer models, and more specifically to a platform for achieving replicability of a computer model (e.g., a trained machine-learning algorithm) by storing and providing access to data associated with the computer model using an immutable and decentralized ledger system (e.g., a blockchain ledger) and a distributed database. An exemplary computer-enabled method for storing a computer model, the method comprises: receiving data associated with the computer model; generating one or more asset files based on the data associated with the computer model; generating one or more hash values corresponding to the one or more asset files; generating one or more of location trackers corresponding to the one or more asset files; generating a ledger entry comprising the one or more hash values and the one or more location trackers; and adding the ledger entry to a blockchain ledger.

Abstract: One example method includes optimizing client-side deduplication. When backing up a client, a cadence and a change log resolution are determined. These values are evaluated alone or in combination with respect to various thresholds. Client-side deduplication is enabled or disabled based on whether any one or more of the thresholds are satisfied.

Abstract: There is a need for more effective and efficient detection of cross-data-column relationships. This need can be addressed by, for example, techniques for detecting cross-data-column data relationships that utilize at least one of feature-based similarity models and deep-learning-based similarity models. The cross-data-column data relationships may be displayed to an end-user using a cross-column relationship detection user interface.

Abstract: Disclosed are embodiments for replacing database table join keys with index keys. In one embodiment, a method is disclosed comprising: receiving, by a processor, annotation data, the annotation data comprising a set of rows; retrieving, by the processor, a root dataset, the root dataset stored in one or more files; generating, by the processor, a row identifier for each row in the set of rows, the row identifier storing a plurality of fields enabling alignment of a respective row in the annotation data to a corresponding row in the root dataset; generating, by the processor, an annotation dataset, the annotation dataset comprising the set of rows and corresponding row identifiers; and writing, by the processor, the annotation dataset to at least one file, the at least one file separate from the one or more files.

Abstract: A host system communicates to a production system that performs point-in-time copy operations to a backup system to generate a logical corruption protection (LCP) copy of a production copy that comprises a data set, wherein the point-in-time copy operations perform copying at a media level in which physical tracks are copied from the production system to the backup system, and wherein the point-in-time copy operations avoid using information on the data set or file structure that is included in a media on which the data set is stored. In response to a detection of a logical corruption in the data set, a recovery application executing in the host system recovers the data set from the LCP copy.

Abstract: A document review system is configured to quickly provide search results for large amounts of documents. Ontology templates are used to generate objects and links based on the documents. Users can view documents, and linked objects and object properties can be displayed. An index assisted search can be performed to determine the linked objects and object properties. A supplemental search can be performed on a second database that tracks modifications to the documents or objects. The modifications in the second database can be periodically written to the first database, and the index can be rebuilt. The user experience can be free of slow searches and slow index rebuilds.

Abstract: Techniques for schema mismatch detection and evolution are described. When data is being uploaded into a source table, schema of the data to be uploaded can be compared with the schema for the source table. If a schema mismatch is detected, the schema of the source table can be modified, and the upload can be continued without data loss.

Abstract: A system and method on one computer records actions on objects performed on a second computer. If an object is deleted from the second computer, the first computer can restore to a point in time the state of the object on the second computer, including assignments made between it and other objects, using the recorded actions.

Abstract: To manage expired backups in a storage system, a backup server retrieves multiple deletion logs which record invalid data included in one or more expired backups. When a deletion condition is met, the backup server identifies a first large object based on the multiple deletion logs, and sends a data migration request to an object-based storage system to instruct the object-based storage system to copy valid data of the first large object to a second large object. Thereafter, the backup server sends an object deletion request to the object-based storage system to instruct the object-based storage system to delete the first large object, thereby clearing up the expired backups.

Abstract: A media creation system is provided for creating photographs and videos that include assets sourced from different geographic locations. The media creation system may parsing media, transposition the media, and/or perform additional modifications to the media. A method for creating photographs, videos, and live streams of one or more assets sourced from different geographic locations using the audiovisual media composition system is also provided.

Abstract: Described herein is a mechanism for utilizing a neural network to identify and rank search results. A machine learning model is trained by converting training data comprising query-document entries into query term-document entries. The query term-document entries are utilized to train the machine learning model. A set of query terms are identified. The query terms can be derived from a query history. The trained machine learning model is used to calculate document ranking scores for the query terms and the resultant scores are stored in a pre-calculated term-document index. A query to search the document index is broken down into its constituent terms and an aggregate document ranking score is calculated from a weighted sum of the document ranking scores corresponding to the individual query terms. Because the term-document index can be pre-calculated, it can be downloaded to provide deep learning search capabilities in a computationally limited environment.

Abstract: This disclosure describes techniques for creating a universal schema with default fields that support sensor formats of different devices. In one example, the universal schema supports substantial equivalents between data fields in different sensor formats. Further, a sensor format may be configured to support inheritance and aggregation of sensor formats in prior devices. Accordingly, the mapping of sensor formats that supports inheritance and aggregation in the universal schema may provide several advantages such as capturing a mapping of substantive equivalents between the fields in different sensor formats.