Abstract: A database system comprising a database comprising a plurality of data storage nodes, the plurality of data storage nodes including a primary storage node and a plurality of secondary storage nodes. The database system may also include at least one processor configured to: perform at least one write command to a secondary storage node of the plurality of secondary storage nodes; and perform at least one backup command, at least partially in parallel with the at least one write command being performed to the secondary storage node, of at least some data from a snapshot of data stored in the secondary storage node. A method for performing at least one backup command in a database at least partially in parallel with the at least one write command being performed.

Abstract: A system groups multiple entities in a large distributed data store (DDS), such as directories and files, into a subset called a domain. The domain is treated as a unit for defining policies to detect and treat sensitive data. Sensitive data can be defined by enterprise or industry. Treatment of sensitive data may include quarantining, masking, and encrypting, of the data or the entity containing the data. Data in a domain can be copied as a unit, with or without the same structure, and with transformations such as masking or encryption, into parts of the same DDS or to a different DDS. Domains can be the unit of access control for organizations, and assigned tags useful for identifying their purpose, ownership, location, or other characteristics. Policies and operations, assigned at the domain level, may vary from domain to domain, but within a domain are uniform, except for specific exclusions.

Abstract: Techniques are provided for freeing and utilizing unused inodes. For example, an operation, targeting a first storage object of a first node having a replication relationship with a second storage object of a second node, is intercepted. A replication operation is created as a replication of the operation. The operation is implemented upon the first storage object and the replication operation is implemented upon the second storage object. A determination is made that the replication operation uses an inode no longer used by storage objects of the second node. The inode targeted by the replication operation is freed and utilized based upon the inode being a leaf inode. If the inode is a stream directory inode, then data streams of the stream directory inode are moved under a new private inode and the stream directory inode is released.

Abstract: Systems and methods for generating concise explanations of scored observations that strike good, and computationally efficient, trade-offs between rank-ordering performance and explainability of scored observations are disclosed. The systems and methods described herein for explaining scored observations are based on a framework of partial dependence functions (PDFs), multi-layered neural networks (MNNs), and Latent Explanations Neural Network Scoring (LENNS).

Abstract: A clustering machine can cluster descriptive vectors in a balanced manner. The clustering machine calculates distances between pairs of descriptive vectors and generates clusters of vectors arranged in a hierarchy. The clustering machine determines centroid vectors of the clusters, such that each cluster is represented by its corresponding centroid vector. The clustering machine calculates a sum of inter-cluster vector distances between pairs of centroid vectors, as well as a sum of intra-cluster vector distances between pairs of vectors in the clusters. The clustering machine calculates multiple scores of the hierarchy by varying a scalar and calculating a separate score for each scalar. The calculation of each score is based on the two sums previously calculated for the hierarchy. The clustering machine may select or otherwise identify a balanced subset of the hierarchy by finding an extremum in the calculated scores.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for providing content. One of the methods includes receiving a target image at an image platform; determining one or more images related to the target image; determining queries associated with the related images; deriving candidate keywords from the determined queries; filtering the candidate keywords to select one or more keywords; and providing the image to one or more users in response to respective incoming queries based on the one or more keywords.

Abstract: Methods and systems relating to a file distribution scheme in a computer network are disclosed that distributes files in an efficient manner that reduces, among other things, network traffic. In an embodiment of the invention, a method for updating a file is disclosed. In such a method, unique chunks in a first version of a digital file are identified. For a second version of the digital file, chunks that are the same as in the first version are identified. Recompilation information is generated and stored for these identified chunks. Also, for the second version of the digital file, chunks in the second version that are different from chunks in the first version are identified. Recompilation information is generated and stored for these identified chunks. With this information, the second version of the digital file is completely defined and can be efficiently stored.

Abstract: A recommendation method is provided. In the method, a candidate item to be recommended to a social network user is obtained. The social network user has at least two different types of social relationships. For at least one target social object in each of the at least two different types of social relationships of the social network user, attention of each of the at least one target social object in the respective type of social relationship to the candidate item is determined. According to the attention of each of the at least one target social object in the at least two different types of social relationships to the candidate item, a comprehensive attention of the target social objects of the at least two different types of social relationships to the candidate item is determined. According to the comprehensive attention, whether to recommend the candidate item to the social network user is determined.

Abstract: Disclosed are various embodiments for sharing documents among users of an enterprise as well as with users external to an enterprise. A document is identified and document components extracted from the document. A browser representation is generated that, when rendered or interpreted by a browser, causes the browser to generate a user interface that presents at least a portion of the document as the document would be viewed by a native viewer.

Abstract: In some embodiments, a data server generates an open data reference specifying a location for storing report data associated with a report. To generate the report data, the data server generates a data query based on a report definition that describes how the report data associated with the report is generated. The data server executes the data query on source data to obtain the report data and causes the report data to be stored on a storage device at the location specified by the open data reference. The report data can be retrieved via a network using the open data reference for integration within a third-party application. The data server further generates updated report data by executing, on the source data, an updated data query generated based on the report definition and has the report data stored on the storage device to be replaced with updated report data.

Abstract: In some implementations, systems and methods that are capable of customizing negative media searches using domain-specific search indexes are described. Data indicating a search query associated with a negative media search for an entity and a corpus of documents to be searched are obtained. Content from a particular collection of documents from among the corpus of documents is obtained and processed. Multiple scores for the entity are computed based on processing the content obtained from the collection of documents. The multiple scores are aggregated to compute a priority indicator that represents a likelihood that the collection of documents includes content that is descriptive of derogatory information.

Abstract: Embodiments presented herein describe techniques for deduplicating chunks of data across multiple clusters. A process executing in a storage system identifies one or more chunks in an incoming stream of data. For each chunk, a first fingerprint corresponding to the chunk is generated. The process determines whether the first fingerprint matches a second fingerprint listed in a corresponding entry in a deduplication map. Each entry of the deduplication map corresponds to a chunk stored in a location in one of the storage clusters. Upon determining that the first fingerprint matches the second fingerprint, the process writes, to a local persistent storage, a pointer referencing the location in that storage cluster.

Abstract: Techniques for ranking services and top N rank lists are disclosed. In some embodiments, a system, process, and/or computer program product for ranking services and top N rank lists includes receiving a set of network related event data, wherein the set of network related event data includes Domain Name System (DNS) related event data; aggregating the DNS related event data over a period of time and rank order by popularity; and generating a top N rank list for ranking popularity over the period of time for a set of domains using the aggregated DNS related event data and rank order by popularity.

Abstract: A database interface system uses individual application program interfaces to receive, convert, combine, and normalize if necessary, datasets received from disparate sources. The datasets may include quotes from various insurance providers. The categorized and normalized data may be further processed for presentation via a user interface that allows selective display of the full combined datasets or graphically selected subsets of the full datasets.

Abstract: The problem of distinct value estimation has many applications, but is particularly important in the field of database technology where such information is utilized by query planners to generate and optimize query plans. Introduced is a novel technique for estimating the number of distinct values in a given dataset without scanning all of the values in the dataset. In an example embodiment, the introduced technique includes gathering multiple intermediate probabilistic estimates based on varying samples of the dataset, 2) plotting the multiple intermediate probabilistic estimates against indications of sample size, 3) fitting a function to the plotted data points, and 4) determining an overall distinct value estimate by extrapolating the objective function to an estimated or known total number of values in the dataset.

Abstract: A set of both smart contract and off-chain tools is described that enable the management and organization of data so as to enable storage of that data in a distributed ledger according to relational database principles. A cross-distributed-ledger-platform specification plus reusable core components together create a system that may be implemented on distributed ledger platforms to enable storage and retrieval of data to/from the distributed ledger governed by relational principles. A realization of this system enables addition of system chaincode to Hyperledger® Fabric and uses schemas and data represented as JSON. In use, the user may create, update, and query data from code, a console, or a smart contract where every update is a distributed ledger transaction.

Abstract: A system for developing machine learning for use in the radiofrequency domain that produces a robust set of training data for machine learning from a small set of labelled training data that is bootstrapped with unlabeled electromagnetic environment data. A raw signal set is prepared from the labeled data and separately processed for any electromagnetic environment and interference signals as well as for a primary signal by applying the real electromagnetic environment data and then summed to generate a second data set that is larger than the first data set. Feature extraction is used to produce a bootstrapped labelled data set that is larger than the original labelled data set and that can be used as training data for machine language classification.

Abstract: A computer-implemented method for generating information associated with unknown verbal communications is disclosed. The computer-implemented method includes receiving a user indication of an unknown verbal communication. The computer-implemented method further includes displaying a potential unknown word list to the user, wherein the potential unknown word list includes words corresponding to verbal communications that occurred prior to receiving the user indication of the unknown verbal communication. The computer-implemented method further includes identifying that a word is selected from the potential unknown word list. The computer-implemented method further includes generating, in response to identifying that the word is selected from the potential unknown word list, additional information associated with the selected word.

Abstract: A processor-implemented method for generating a matrimony recommendation using a machine learning model is provided. The method includes (i) generating a user profile for (i) a first user or (ii) a second user, (ii) obtaining family information of (a) the first user from a first user device associated with the first user and (b) the second user from a second user device associated with the second user, (iii) generating a database using the family information associated with the first user and the second user, (iv) automatically generating a first family tree using the family information associated with the first user, and a second family tree using the family information associated with the second user based on a pre-determined tree template and (v) populating family members of (a) the first family tree or (b) the second family tree based on a response from the first user or the second user.

Abstract: Systems, methods, and non-transitory computer-readable media can determine respective engagement scores for one or more users based at least in part on their engagement with a plurality of content items associated with an entity in a content provider system. At least one user who is a top fan of the entity can be determined based at least in part on the determined engagement scores. The user can be associated with a visual badge in the content provider system to identify the user as a top fan of the entity, wherein the visual badge is made visible by the content provider system to at least the entity.