A computer-implemented method for database management is provided. The method comprises: receiving, from a client device, first data to be stored in a database system that comprises first data storage configured to store a data table and a deletion history table; storing the first data in second data storage that is external to the database system and that is in communication with the database system via a network; obtaining a link that enables access, via the network, to the first data stored in the second data storage; storing the link in the data table; and performing a deletion operation of the first data, in response to a request from the client device to delete the first data from the database system, wherein the deletion operation comprises: deleting the link from the data table without deleting the first data from the second data storage; and storing the link in the deletion history table with a timestamp corresponding to a point in time when the link is deleted from the data table.

A processor-based method for locating data and metadata closely together in a storage system is provided. The method includes writing a first range of a file and a first metadata relating to attributes of the file into at least one segment controlled by a first authority of the file. The method includes delegating, by the first authority, a second authority for a second range of the file, and writing the second range of the file and second metadata relating to the attributes of the file into at least one segment controlled by the second authority.

Embodiments relate to systems and methods to optimize quantization of tensors of an AI model. According to one embodiment, a system receives an AI model having one or more layers. The system receives a number of input data for offline inferencing and applies offline inferencing to the AI model based on the input data to generate offline data distributions for the AI model. The system quantizes one or more tensors of the AI model based on the offline data distributions to generate a low-bit representation AI model, where each layer of the AI model includes the one or more tensors, where the one or more tensors include the one or more tensors. In one embodiment, the system applies online inferencing using the low-bit representation AI model to generate online data distributions for a feature map, and quantizes a feature map tensor based on the online data distributions.

A system and method to capture and interact with a comprehensive digital record of an individual's biographical history and produce a synthetic model of their personality. The captured biographical history is a detailed record of this individual's actions, interactions, and experiences over a period which may span decades of their lifetime. The biographical history is indexed by areas of data variability and neural network confidence variability to identify points of likely human interest. A synthetic personality model is generated as a representation of the individual's personality structure, biases, sentiments, and traits. The synthetic personality can be interacted with through a digital interface and demonstrates the interaction patterns, triggers, and habits of the original individual. The functioning and the performance of the system over an individual's lifespan are optimized through data synthesis and disposition.

Techniques are disclosure for a data enrichment system that enables declarative external data source importation and exportation. A user can specify via a user interface input for identifying different data sources from which to obtain input data. The data enrichment system is configured to import and export various types of sources storing resources such as URL-based resources and HDFS-based resources for high-speed bi-directional metadata and data interchange. Connection metadata (e.g., credentials, access paths, etc.) can be managed by the data enrichment system in a declarative format for managing and visualizing the connection metadata.

The present invention relates to a big data related script-based decision management system and method thereof that applies the Internet of Thing technology and adopts an open architecture for setting up a big data database. The big data database stores a plurality of data collected by a plurality of remote devices, wherein the data consists of a feature extracted from the operational statuses of a plurality of remote devices; an event that is formed by analyzing the feature; a script that stores the response strategy created based on the event; and a unit of analysis that produces a strategy result by comparing and analyzing the event and the script.

Methods and systems for encrypting and decrypting data comprising sending sensitive information to a first cryptographic processing system in a first cloud region for encryption with a first key encryption key generated by and stored by the first cryptographic processing system. The first encrypted sensitive information received from the first cryptographic processing system is stored in a first database. The sensitive information is also sent to a second cryptographic processing system in a second cloud region different from the first cloud region for encryption with a second key encryption key generated by and stored by the second cryptographic processing system. The second encrypted sensitive information received from the second cryptographic processing system is stored in a second database. If the first encrypted sensitive information cannot be decrypted by the first cryptographic processing system, the second encrypted sensitive information is sent to the second cryptographic processing system.

Examples described herein generally relate to database systems for storing and processing both small values that are smaller than size of a database column and large objects that exceed the size of the database column. A database management system (DBMS) determines that a value to be stored in a database is a large object having a size larger than a column of the database. The DBMS stores the value as a large object in an external storage associated with a token stored in the column of the database. The token includes information for processing the large object. A vector processing engine associated with the external storage processes the large object based on the information in the token in response to a database command from the DBMS on multiple records represented as a vector.

Examples described herein generally relate to database operations including encoding, within a length field for a first value to be stored in a column of a database as a token that includes information for processing a large object, an indicator indicating that the first value is of a token type, and storing, in the column of the database, the first value with the length field including the indicator.

Implementations described herein relate to methods, systems, and computer-readable media to write data records. In some implementations, a method may include calculating a data rate of a data stream that includes a plurality of data records and determining if the data rate of the data stream is less than an ingest threshold. The method may further include, if the data rate of the data stream is less than the ingest threshold, calculating a number of write requests per time unit based on the data stream; determining a storage capacity per storage bucket; determining a read interval for the data stream; based on the number of write requests per time unit, the storage capacity, and the read interval, selecting a size of time window per storage bucket; and writing the plurality of data records to a particular storage bucket.