A method includes receiving a plurality of product query arrays each including a plurality of individual product queries received during a single user search session. The method further includes inputting the plurality of product query arrays into the query rewrite model. Text of each of the plurality of individual product queries in each product query array is treated as a whole token. The method further includes receiving a product query from a user electronic device. The method further includes determining a query rewrite for the product query using the query rewrite model and determining search results for the product query using the query rewrite. The method further includes sending information for presenting the search results on a display of the user electronic device responsive to the product query.

Provided is a method including obtaining a corpus and an associated set of domain indicators. The method includes learning a set of vectors in an embedding space based on n-grams of the corpus. The method includes updating ontology graphs comprising a set of vertices and edges associating the set of vertices with each other. The method also includes determining a vector cluster using hierarchical clustering based on distances of the set of vectors with respect to each other in the embedding space and determining a hierarchy of the ontology graphs based on a set of domain indicators of a respective set of vertices corresponding to vectors of the vector cluster. The method also includes updating an index based on the ontology graphs.

A third-party service may be used to assist entities in responding to requests of users by determining a suggested response to a received communication. The third party service may receive a request from a first entity, such as via an application programming interface request, that includes a message in a conversation. A conversation feature vector may be computed by processing the message with a first neural network. A suggested respond to the message may be determined by processing the conversation feature vector with a second neural network. The third-party service may then return the suggested response for use in the conversation. The third-party service may similarly be used to assist other entities in responding to requests of users.

Adaptive data collections may include various type of data arrays, sets, bags, maps, and other data structures. A simple interface for each adaptive collection may provide access via a unified API to adaptive implementations of the collection. A single adaptive data collection may include multiple, different adaptive implementations. A system configured to implement adaptive data collections may include the ability to adaptively select between various implementations, either manually or automatically, and to map a given workload to differing hardware configurations. Additionally, hardware resource needs of different configurations may be predicted from a small number of workload measurements. Adaptive data collections may provide language interoperability, such as by leveraging runtime compilation to build adaptive data collections and to compile and optimize implementation code and user code together. Adaptive data collections may also provide language-independent such that implementation code may be written once and subsequently used from multiple programming languages.

A method includes receiving, by a logic layer processor, over a communication network, from a plurality of electronic resources, initial user personal identifiable information (PII) of a user of a plurality of users. The user PII includes a plurality of data elements. The plurality of data elements of the initial PII of the user are classified to populate a profile map data structure having a standardized predefined data schema of a plurality of vector elements so as to form a user-specific profile map data structure of the user. Additional user personal identifiable information (PII) of the user is iteratively received from the plurality of electronic resources. The additional user PII of the user is iteratively classified to update the user-specific profile map data structure of the user. A plurality of user-specific data management software functions is enabled based on the user-specific profile map data structure.

Time-series data indicating a temporal variation of an index, which indicates a usage state of each of resources that are used by multiple processes, is acquired, and an operation-data matrix including vectors is generated based on the time-series data such that each of the vectors indicates the time-series data at a predetermined time interval and includes as an element the index indicating the usage state of one of the resources at the predetermined time interval. A basis matrix including a predetermined number of basis vectors is generated by performing nonnegative matrix factorization on the operation-data matrix. Component values, which respectively correspond to the resources, indicated by each of the predetermined number of the basis vectors are extracted, and information on the extracted component values is output as usage states of the resources that are used by each of the multiple processes.

A method includes obtaining a query containing at least one field from which data is being queried, obtaining a dataset having a schema-free data exchange format having multiple fields of data at different physical positions in the dataset, and parsing the dataset by obtaining a structural index that maps logical locations of fields to physical locations of the fields of the dataset, accessing the structural index with logical locations of the fields that index to the physical locations, and providing data from the fields based on the physical locations responsive to the query.

The “free world replication protocol” makes use of client computing resources, wherein the clients are not part of the replicated key-value store, but instead reside in the “free world” outside of the dedicated resources of the nodes of the replicated key-value store. In the free world replication protocol, only a single “write” client is authorized to modify the key-value store at any time but any number of clients may be authorized to read data from the key-value store. The write client sends its transactions to multiple nodes in the replicated key-value store. As a result, the latency between the transaction being sent from the client and the transaction being received by the multiple nodes is reduced. A consensus protocol, driven by a master node, is used to periodically ensure consistency, but the data transactions themselves do not make use of a master node.

A computer-implemented method for content-agnostic referencing of a binary data file, the method comprising: determining a length of the binary data file, the length comprising the number of bits of the binary data file; for the determined length, generating all permutations of data of the determined length; locating an index within the generated permutations, wherein the index is the starting position of the binary data file within the generated permutations; and using the length and the index to indicate the binary data file.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating data defining a graph. In one aspect, a method comprises: sequentially generating a respective edge set for each node in the graph, wherein for each of a plurality of nodes after a first node, generating the edge set for the node comprises: receiving a context embedding for the node that summarizes a respective edge set for each node that precedes the node; generating, based on the context embedding for the node: (i) a respective edge set for the node, and (ii) a respective embedding of the edge set for the node; generating a context embedding for a next node in the ordering of the nodes using the embedding of the edge set for the node; and adding the set of edges defined by the edge set for the node to the graph.