Abstract: A social networking system maintains characteristics with its users, with various characteristics, such as age, specified by the users (i.e., “asserted characteristics”). The social networking system selects content for a user based at least in part on the characteristics associated with the user. To account for potential inaccuracies in an asserted age of a user, the social networking system clusters users based on ages of other users connected to users. The online system receives verified ages for users in a cluster from a trusted third party system that maintains more accurate characteristics for users than the social networking system. By comparing the asserted ages for users in the cluster to the verified ages for users in the cluster, the social networking system determines an accuracy of the asserted ages for users in the cluster. The accuracy may be used when selecting content for the users.

Abstract: In a production environment, an entry is logged in a log journal to represent a read operation on a record of a file. A problematic transaction in a batch job is selected and set of operations performed by the problematic transaction is intercepted. The set includes a combination of read, write, update, delete operations, and operations that use external resources. A window of entries is determined in the log journal where the entry is a beginning entry in the window and an ending entry is a last entry captured in the log journal before detecting the error. From the window, those entries are filtered that correspond to the record. A final value of the record is copied from the production environment to a development environment and rolled back in the development environment to an initial value stored in the beginning entry. The problematic transaction is performed in the development environment.

Abstract: A learning-to-rank method based on reinforcement learning, including obtaining, by a server, a historical search word, and obtaining M documents corresponding to the historical search word; ranking, by the server, the M documents to obtain a target document ranking list; obtaining, by the server, a ranking effect evaluation value of the target document ranking list; using, by the server, the historical search word, the M documents, the target document ranking list, and the ranking effect evaluation value as a training sample, and adding the training sample into a training sample set.

Abstract: A framework for implementing just-in-time decision support analytics within a business application deployment are provided. In one set of embodiments, the framework enables an application user to be presented with analytics that are in-context and in-place with respect to an operational workflow carried out by the user, where “in-context” means that the presented analytics are directly relevant to, and tailored for, the specific business context of the user's operational workflow, and “in-place” means that the analytics are displayed to the user within the same UI/application used to execute the operational workflow, and at the same time the workflow is actually executed. These characteristics advantageously ensure that the user is provided with the insights he/she needs to make confident business decisions and facilitates real-time decision making for time-critical operations.

Abstract: Machine learning based method for generating personalized query suggestions is described. Different users may have different search intent even when they are inputting the same search query. The technical problem of personalizing search query suggestions produced by a machine learning model is addressed by extending the sequence to sequence machine learning model framework to be able to take into consideration additional, personalized features of the user, such as, e.g., profile industry, language, geographic location, etc. This methodology includes an offline model training framework as well as an online serving framework.

Abstract: Techniques are presented herein for efficient query processing and data change propagation at a secondary database system. The techniques involve determining execution costs for executing a query at a primary DBMS and for executing the query at an offload DBMS. The cost for executing the query at the offload DBMS includes the cost of propagating changes to database objects required by the query to the offload DBMS. Based on the execution cost, the query is sent to either the primary DBMS or the offload DBMS.

Abstract: Systems, methods, and storage media for generating and/or recoverying an end-to-end hybrid cloud snapshot are disclosed. Exemplary implementations may: receive application specific triggering criteria for generating an end-to-end hybrid cloud snapshot; trigger a snapshot event based on the application specific triggering criteria and a machine learning model; accumulate snapshot portions from a plurality of nodes in a hybrid cloud network; and store, in a snapshot database, the end-to-end hybrid cloud snapshot comprising the snapshot portions from the plurality of nodes. Exemplary implementations may further include: generate a recovery model based on a snapshot database, wherein the snapshot database comprises a plurality of network snapshots associated with a network, receive parameters for a desired recovery state, and identify a recommended network snapshot from the plurality of network snapshots based on the recovery model and the parameters for the desired recovery state.

Abstract: This disclosure relates to method of extracting an information associated with design of formulated products and representing as a graph. A graph domain model of a plurality of vertices, and at least one formulation text as text file are received as an input. The information extraction is applied to identify at least one sentence and extract at least one subject-verb-object triple from every sentence of the at least one formulation text. A sentence including an ingredient listing and associated weights indicated by presence of weight numerals, and a sentence including at least one verb from the at least one subject-verb-object based on the graph domain model are classified. A representation of the recipe text is generated in terms of at least one action, ingredients on which the at least one action is performed, and condition. An insert query string is generated and executed to store the formulations as the graph.

Abstract: A method generates a data source. A user selects a first data set from a displayed object model of a database. The object model includes data sets linked visually by many-to-one relationships to form a tree. The user selects a data field in the first data set and specifies a filter condition. The system identifies a second data set in the tree, and joins the first and second data sets. When the second data set is related to the first data set via one or more many-to-one relationships, rows of the second data set are joined to rows of the first data set that satisfy the first filter condition. When the second data does not satisfy the condition, the second data set is joined to rows of the first data set that satisfy the first filter condition, and all rows of the second data set are included in the generation.

Abstract: Representative embodiments disclose mechanisms to assign search results associated with a user query to locations based in part on features associated with the user's Internet of Things (IoT) devices. This results in a search page that presents results in an order and prominence that improves user understanding and information accessibility. Features from IoT devices are used to train a machine learning model that receives as an input information from an IoT entity index, the query, and results from the search engine along with their static rankings. The resultant IoT positioning scores are used to assign an initial location on the search results page. A second machine learning model trained using IoT features is then used to assign a final location on the page. The system causes display of the page to the user.

Abstract: There is provided a method and a system for ranking subgraphs as potential explanations for a labelled edge type class. A first graph representing labelled digital items is obtained, where each labelled digital item is represented as an entity node connected via a labelled edge type to a property value node. The first graph is combined with a second graph representing structured relations in the labelled digital items to obtain a combined graph. Unlabelled digital items are received and matched to respective subgraphs in the combined graph. A machine learning model is used to embed the combined graph to generate graph vectors, and an expressivity score between matched subgraphs and respective labelled edge types based on the generated graph vectors. The matched subgraphs are ranked based on the expressivity score to obtain a ranked set of subgraphs as potential explanations for a respective labelled edge type class.

Abstract: The present disclosure describes a service which provides primary in-line deduplication. A streaming application program interface (API) may allow for streaming records into a storage system with high throughput and low latency. As part of this process, the API allows user to add identifiers as a field used for data deduplication. The deduplication service keeps a moving window of the identifiers in memory and does in-line deduplication by quickly determining whether data is a duplicate. Keeping only deduplication keys in memory reduces the cost of running the service. Moreover, the real-time nature of the moving window approach allows for storing deduplication information alongside the data and accessing it immediately on read. In this regard, read after write consistency is supported, and costs are reduced.

Abstract: Computer implemented methods for database hierarchy-independent data drilling are presented including: a. selecting one of at least two database structures each having data, where the at least two database structures include, at least two distinct paths that lead from a first of the at least two database structures to a second of the at least two database structures, where a path logically relates at least two data structures which are otherwise directly unrelated using data structures and data structures' relations, and a database information schema that provides information about the at least two database structures including information regarding relations between data structures; b. setting the selected database structure as a current data drilling state; and c. finding at least one related data structure corresponding with the selected database structure for the current data drilling state utilizing the database information schema.

Abstract: While connected to cloud storage, a computing device writes data and metadata to the cloud storage, indicates success of the write to an application of the computing device, and, after indicating success to the application, writes the data and metadata to local storage of the computing device. The data and metadata may be written to different areas of the local storage. The computing device may also determine that it has recovered from a crash or has connected to the cloud storage after operating disconnected and reconcile the local storage with the cloud storage. The reconciliation may be based at least on a comparison of the metadata stored in the area of the local storage with metadata received from the cloud storage. The cloud storage may store each item of data contiguously with its metadata as an expanded block.

Abstract: Systems and methods are described for linking records from different databases. A search may be performed for each record of a received record set for similar records based on having similar field values. Recommended records of the record set may be assigned with the identified similar records to sub-groups. Pairs of records may be formed for each record of the sub-group, and comparative and identifying features may be extracted from each field of the pairs of records. Then, a trained model may be applied to the differences to determine a similarity score. Cluster identifiers may be applied to records within each sub-group having similarity scores greater than a predetermined threshold. In response to a query for a requested record, all records having the same cluster identifier may be output on a graphical interface, allowing users to observe linked records for a person in the different databases.

Abstract: An information extraction unit 82 extracts a set including pieces of predetermined information from a past search history for each change of a configuration requirement. A learning unit 83 uses a collection of the sets as learning data to learn a calculation method for calculating a score indicating the degree of appropriateness of a concretizing rule included in the set including the pieces of predetermined information. A search unit 86 repeats extracting the set including the pieces of predetermined information for each pair including a concretizing rule applicable to a configuration requirement of interest and an application part of the concretizing rule in the configuration requirement of interest, calculating a score for the set, and newly selecting a configuration requirement as a configuration requirement of interest based on the score.

Abstract: A system and computer implemented method for execution of aggregation expressions on a distributed non-relational database system is provided. According to one aspect, an aggregation operation may be provided that permits more complex operations using separate collections. For instance, it may be desirable to create a report from one collection using information grouped according to information stored in another collection. Such a capability may be provided within other conventional database systems, however, in a non-relational database system such as NoSQL, the system is not capable of performing server-side joins, such a capability may not be performed without denormalizing the attributes into each object that references it, or by performing application-level joins which is not efficient and leads to unnecessarily complex code within the application that interfaces with the NoSQL database system.

Abstract: A method for deleting a row from a table in a database system comprises logically deleting the row in the first table in the database system by inserting a key of the row into a corresponding row of a dedicated table in the database system; querying the dedicated table during a query against the first table to identify the corresponding row in the dedicated table; and in response to identifying the corresponding row in the dedicated table, deleting the row from the first table and the corresponding row from the dedicated table as part of query processing during a subsequent query.

Abstract: A system, method and apparatus for storing metadata in a metadata store in a robust and efficient manner including receiving a request from a client to perform a data transaction, updating a key-value pair in a metadata store based on the request, entering the data transaction in a transaction log, updating a read cache with the key-value pair, and replicating the last transaction log entry in at least one other storage node in the metadata store.

Abstract: A solid-state drive (SSD) includes: a plurality of data blocks; a plurality of flash channels and a plurality of ways to access the plurality of data blocks; and an SSD controller that configures a block size of the plurality of data blocks. A data file is stored in the SSD with one or more key-values pairs, and each key-value pair has a block identifier as a key and a block data as a value. A size of the data file is equal to the block size or a multiple of the block size.