Abstract: Disclosed is a system and method for resolving conflicts that arise in changes made to data by multiple client devices. A data manipulation operation is represented as a changeset, which has instructions for performing an operation on the data. When two client devices perform operations that change the state of a data object, a changeset is generated at each of the two client devices. These changesets are merged at each of the two client devices based on merging rules that resolve any conflict between the operations. The rules ensure that the data at each of the multiple client devices will eventually be consistent.

Abstract: In embodiments, given a surface that has domains on it and a first location, a second location is generated in the surface. The second location belongs in the same domains as the first location, and does not belong in the same domains as the second location. An advantage can be that, in embodiments, the second location can be used in lieu of the first location. A use case can be where it is desired to protect the privacy of location data of an entity, such as its address.

Abstract: Embodiments presented herein disclose techniques for capturing a snapshot of a file system object (e.g., a file or a directory) that is associated with a write journal having outstanding data. A bridge process in a storage server receives a request to capture a snapshot of a file system object. The snapshot is a backup of a state of the file system object in a given point in time. Upon determining that the file system object has one or more outstanding updates recorded in a write journal, the bridge process generates a copy of the write journal. The bridge process captures the snapshot of the file system object. The bridge process also associates the copy of the write journal with the snapshot of the file system object.

Abstract: Multiple geographic grids are generated according to acquired location information of a service provider. Each geographic grid contains one or more service providers. The second region matching the target location information is a geographic grid for a region defined by the target location information. A presentation request for a search page is received. The presentation request includes target location information. A keyword is acquired that has a search popularity meeting a first condition in a first region matching the target location information, and the acquired keyword is used as a first keyword. The first keyword is provided for presentation in the search page.

Abstract: An information search method and apparatus, a device and a storage medium, relating to fields of intelligent search, intelligent recommendation, deep learning and the like. A specific implementation includes: determining feature information of a target user; acquiring at least one candidate search path to determine candidate search results in a database; determining a matching degree between the feature information of the target user and the at least one candidate search path using a matching model, and determining a target search path from the candidate search paths according to the matching degree; and determining a target search result matched with the target user using the target search path.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for retrieving and using contextual data from previous conversation sessions in conversational searches. In one aspect, a method includes receiving a first query for a first user session, determining that the first query refers to one or more tags in a first repository, the first repository associating respective identifiers to respective tags, each identifier representing a corresponding user session, determining one or more particular identifiers associated with the one or more tags in the first repository, retrieving particular contextual data associated with the determined particular identifiers in a second repository, the second repository associating respective identifiers to respective contextual data associated with corresponding user sessions represented by the respective identifiers, and performing an action responsive to the first query based on the retrieved particular contextual data.

Abstract: Apparatuses, methods, and systems are provided for making sequential recommendations using transition regularized non-negative matrix factorization. A non-application specific collaborative filtering based personalized recommender system can recommend a next logical item from a series of related items to a user. The recommender system can recommend a next desirable or series of next desirable new items to the user based on the historical sequence of all user-item preferences and a user's most recent interaction with an item. An asymmetric item-to-item transition matrix can capture aggregate sequential user-item interactions to design a loss function for matrix factorization that incorporates the transition information during decomposition into low-rank factor matrices.

Abstract: Facilitating outlier object detection in tiered storage systems is provided herein. A system can comprise a processor and a memory that stores executable instructions that, when executed by the processor, facilitate performance of operations. The operations can comprise determining respective parameters associated with objects of a group of objects of a tiered storage system. The respective parameters can comprise at least one of a size, an access percentage, or a cost. The operations also can comprise using the respective parameters associated with the objects of the group of objects as inputs and performing data clustering on the group of objects, resulting in at least one data cluster. Further, the operations can comprise selecting at least one object from the group of objects as at least one outlier object within the tiered storage system based on the at least one data cluster.

Abstract: A metadata management method, system and medium are provided. All directories and files in a file system are organized in a tree structure, and all directories and subfiles under a single directory are saved as a whole on a continuous address space of an underlying storage device in the order of creation in sequence. When metadata is updated, update operations on all subdirectories/subfiles under a directory are appended to the tail of the directory in the form of log in order, so that each metadata operation triggers only one IO request. In the process of metadata retrieval, parsing a subdirectory under a parent director requires reading all data on the continuous address space corresponding to the parent directory, and then searching for a match by means of multiple acceleration methods.

Abstract: According to certain aspects of the disclosure, a computer-implemented method may be used for information discovery recommendation. The method may include receiving a query for a requested data card and determining information contained on a set of data cards other than the requested data card. Additionally, categorizing the information into a plurality of dimensions of data and matching the dimensions of data with information contained on the requested data card. Additionally, applying a weighting value to each of the matched plurality of dimensions of data and determining a combined weight total for each of the data cards. Additionally, determining at least one recommended data card with the highest combined weight total and displaying a user interface indicating at least one recommended data card is available. Additionally, presenting the at least one recommended data card based on a user interaction with the user interface.

Abstract: From a set of natural language text documents, a concept tree is constructed. For a node in the concept tree a polarity of the subset represented by the node is scored. A second set of natural language text documents is added to the subset, the adding resulting in a modified subset of natural language text documents having a polarity score within a predefined neutral polarity score range. From the modified subset, a bin of sentences is selected according to a sentence selection parameter, a sentence in the bin of sentences being extracted from a selected document in the modified subset. A sentence having a factuality score below a threshold factuality score is removed from the bin of sentences. From the filtered bin of sentences a new natural language text document corresponding to the filtered bin of sentences is generated using a transformer deep learning narration generation model.

Abstract: The invention provides a comprehensive system, method and computer program product for data structure clustering based on variation in data attribute performance. In general, the invention is configured to link a sub-set of data structures out of a global set of data structures based on identifying stimulus type attributes that are structured to produce a functional performance parameter of the associated data structure, particularly in the field of hospital data analysis. Typically, the invention is configured to receive a request for construction of a peer data cluster for a first functional performance parameter of the first entity data structure. The invention may then determine at least one second entity data structure of a plurality of entity data structures that is a peer entity data structure, and linking the first entity data structure and the second entity data structure to form the peer data cluster.

Abstract: The present disclosure relates to systems and methods for updating static machine-learning models (e.g., a Doc2Vec model) without needing to retrain the models. More particularly, the present disclosure relates to systems and methods that can be used to add new data to a base model by training a client model using the new data, and transforming the vector space of the client model to align with the vector space of the base model. The base model can then be updated using the realigned client model. As such, the base model can be updated with the new data without needing to retrain the base model, which can be burdensome to processing resources, insecure, and time consuming.

Abstract: Embodiments are described for storing array snapshots of a file system in networked attached storage (NAS) device. A system and process determine an organization of the file system comprising directories and files in a defined hierarchy, and a size of each file and a number of files in each directory. One or more slicer processes slice the file system into a number of slices, wherein each slice contains at least one of: a defined number of files or files having a defined total size. Backup agents are deployed as proxies to back up the slices to the NAS device, and the slices are backed up as array snapshots in the NAS device.

Abstract: A resource-efficient technique is described for producing and utilizing a set of trained embeddings. With respect to its training phase, the technique receives a group of sparsely-expressed training examples of high dimensionality. The technique processes the training examples using a distributed training framework of computing devices. With respect to its inference stage, the technique draws on the embeddings produced by the training framework. But in one implementation, the inference-stage processing applies a different prediction function than that used by the training framework. One implementation of interference-stage processing involves determining a distance between a query embedding and a candidate item embedding, where each such embedding is obtained or derived from the trained embeddings produced by the training framework. Another manifestation of inference-stage processing involves adjusting click counts based on identified relations among items embeddings.

Abstract: A method for searching multiple documents on a computer system includes steps for inputting a search query and sending the search query to a system core where the query is passed to a search component for searching the documents. The system core in turn receives results from the search component indicating related documents to the query and passes to a summarization component a specified number of the results. The summarization component processes related documents corresponding to the specified number of results and removes duplicate results to produce a multi-document summary that can have different media formats including text, image, and video formats. The system core receives the summary from the summarization component. The multi-document summary is received from the system core and displayed. The results from the search component are also displayed.

Abstract: A method for generating a hierarchical edit script comprises generating, using a first linear sequence of nodes of a first hierarchical dataset and a second linear sequence of nodes of a second hierarchical dataset, a linear edit script defining a linear sequence of operations that can be applied to the first linear sequence to generate the second linear sequence, generating a first delta by matching one or more operations of the linear sequence of operations to the first hierarchical dataset, generating a second delta by matching one or more operations of the linear sequence of operations to the second hierarchical dataset, and generating the hierarchical edit script by merging the first delta and the second delta.

Abstract: The present application provides a system for use in a connected kitchen allowing a user to find and use traditional recipes as would be found in a recipe book but to have them automatically customized to their language, locality, dietary preferences and the appliances in their kitchen.

Abstract: An indication may be received, by a storage service, from an account, of at least a first object and a first threshold time duration, wherein the first object is to be automatically transferred from a lower-latency storage tier to a higher-latency storage tier based on object access history data and the first threshold time duration. A first determined time duration since a most recent access of the first object may be determined based on the object access history data. The first determined time duration may be compared to the first threshold time duration. It may be determined that the first determined time duration meets or exceeds the first threshold time duration. The first object may be transferred from the lower-latency storage tier to the higher-latency storage tier based on the first determined time duration meeting or exceeding the first threshold time duration.

Abstract: Scalable mining method, computer system and computer program product for temporally correlated events that works with input data that is non-transactional in nature and detects high correlation relationships between events with low support thresholds. The scalable mining method manages a large scale of input data by executing the mining process in a distributed manner in a clustered cloud environment via nodes.