Abstract: Techniques for scaling record linkage via elimination of highly overlapped blocks are described. A method for scaling record linkage via elimination of highly overlapped blocks includes identifying a first plurality of blocks based at least on a plurality of records stored in a storage service of a provider network, identifying a plurality of sets of matching blocks from the first plurality of blocks, deleting the plurality of sets of matching blocks except for a first block from each set from the plurality of sets of matching blocks, and iteratively performing dynamic blocking based at least on the first block to generate subsequent pluralities of blocks until the subsequent pluralities of blocks are below a threshold size.

Abstract: In order to facilitate entity resolution, systems and methods include a processor receiving first records associated with one or more entities, and second records associated with the one or more entities. The processor generates candidate pairs based on a similarity between first entity data and second entity data. The processor generates features for each candidate pair based on similarity measures between the first entity record and the second entity record. The processor utilizes a scoring machine learning model to determine a match score for each candidate pair based on each feature. The processor determines clusters of candidate pairs based on the match score of each feature for each candidate pair. The processor merges records of candidate pairs of each cluster into a respective entity record. The processor determines an entity associated with each entity record and updates an entity database with the entity record.

Abstract: Systems and methods are disclosed for providing playlist analytics. A method includes logging first user interactions associated with a playlist of content items and generating metrics based on the logged first user interactions with the playlist. The playlist is associated with a default playback sequence of the content items. A first metric indicates a last content item played by a user from the playlist is different from a last content item in the default playback sequence associated with of the playlist. The method further includes reporting at least the first metric. The playlist is to be modified based on one or more of the plurality of metrics.

Abstract: Disclosed is a method for loading, by a manager, a profile into a memory of a secure element, SE. A profile allows the SE to identify itself to a service. The profile includes a profile element created by reference to a template. A template is identified by an OID identifier, which it keeps despite changes. The manager may transmit an audit request or update the operating system of the SE in order to obtain, in response, a piece of information representative of the version of the template stored by the SE. The manager then identifies any difference in version with the corresponding template it holds. In case of difference, the template of the SE or the template of the manager is updated so as to align their versions. Thus, a profile generated on the basis of this version of the template will be loaded, without malfunction, by the SE.

Abstract: The present disclosure describes a technique for performing an efficient deduplication of compressed source data. The techniques may reduce the required storage footprint required for deduplication of compressed data. In order to reduce the storage size required, the system may perform additional decompression/recompression processes by identifying particular compression algorithms used by a source storage system. Once the compression algorithm is identified, the system may initiate decompression and then perform fingerprint analysis of the segment in the file of the uncompressed data. When a recovery process is initiated, the system may recompress the deduplicated data using the same compression algorithm used by the source storage system. Accordingly, the data recovery process may be performed in manner in which the client device receives restored data as expected and in the original compression format.

Abstract: A media content item recommendation system recommends media content items based on one or more attributes of a seed playlist. The recommended media content items can be determined from a plurality of existing playlists that have been created over a period of time. Such existing playlists can be selected based on similarity to the seed playlist.

Abstract: There is provided a method to optimize memory usage in hash tables by organizing data storage in the hash table with arrays instead of linked lists where sizes of arrays can be predicted with high accuracy. The method has certain conditions and range of applicability, including where hash function of key provides uniform distribution of values. The method is more applicable if the number of entries is more than 60,000.

Abstract: A system for data space limitation includes and interface and a processor. The interface is configured to receive a structured data set. The processor is configured to determine a whole data space for the structured data set; remove duplicate pairs from the whole data space to generate a duplicate pair limited data space; remove hierarchical redundant pairs from the duplicate pair limited data space to generate a redundant pair limited data space; remove user settable combinations from the redundant pair limited data space to generate a user settable combinations limited data space; remove privacy preserving combinations from the user settable combination limited data space to generate a privacy preserving combinations limited data space; remove access prohibited combinations from the privacy preserving combinations limited data space to generate an access prohibited combination limited data space; and provide the access prohibited combination limited data space for combination calculation.

Abstract: In an embodiment, a method comprises creating and storing a parameter mapping that maps a plurality of resources to a plurality of HTTP URL parameters; receiving, at a client computer, a query comprising a request to access one or more resources of the plurality of resources; using the parameter mapping, determining one or more HTTP URL parameters of the plurality of HTTP URL parameters that are mapped to the one or more resources of the plurality of resources specified in the query; generating a HTTP URL by combining the one or more HTTP URL parameters; submitting a request to a server computer based on the HTTP URL that causes the server computer to retrieve the one or more resources of the plurality of resources.

Abstract: Devices often receive content items to be saved for later presentation to a user. In some instances, the content of the message is analyzed to retrieve supplemental information from external data sources, such as a map to a location discussed in a message. However, such supplementing often fails to reflect the personal context of the content item to the user, who may not easily recall the associated personal context when the content item is later presented. Instead, for a content item to be saved, a device may evaluate personal data sources to identify the personal context of the content item to the user, and add personal context annotations to a content item record of the saved content item. Later, during a presentation of the content item to the user, the personal context annotations may be included as a reflection of the personal context of the content item to the user.

Abstract: Described herein is a system and method for transforming contextually relevant items. A search query is received, and, for each of a plurality of electronically stored items (e.g., stored in a source platform), a probability that the stored item is relevant to the search query is calculated using a classification model. Stored items having a calculated probability greater than or equal to a threshold probability are identified. Contextually relevant portions of the identified stored items is determined using a contextual model. The determined contextually relevant portions of the identified stored items are extracted. The extracted contextually relevant portions of the identified stored items are transformed into a target platform. An output of the extracted contextually relevant portions of the identified stored items in the target platform is provided.

Abstract: Implementations of the present application provide a partition splitting method. Data files in a target source partition are divided into at least two data groups. A first data group from the at least two data groups is selected as a target splitting data group. A first splitting step is performed on the target splitting data group, including: splitting the target splitting data group into a plurality of data sub-groups based on a predetermined splitting rule, wherein each data sub-group of the plurality of data sub-groups corresponds to a new partition of at least two new partitions for the target source partition, and assigning each data sub-group to the new partition corresponding to the data sub-group. A second data group from the at least two data groups is selected as the target splitting data group, wherein the second data group is different from the first data group.

Abstract: An information processing unit includes: a diagnostic image input section that inputs the diagnostic image; an operation information obtaining section that obtains display operation history information representing an operation history of a user who controls displaying of the diagnostic image; a query image generation section that extracts a predetermined region of the input diagnostic image to generate a query image; a diagnosed image obtaining section that supplies the generated query image and the display operation history information to a diagnosed image search unit and obtains the diagnosed image obtained as a search result by the diagnosed image search unit; and a display control section that displays the diagnostic image and the obtained diagnosed image for comparison.

Abstract: An exemplary query parsing system accesses user input text representative of a search query. The system generates a plurality of candidate parses of the user input text. Each of the plurality of candidate parses includes a candidate what portion and a candidate where portion determined from the user input text. The system generates a score for each candidate parse in the plurality of candidate parses based on whether one or more of a plurality of predefined parsing features are found in the respective candidate parse and on a machine learning mode. The system selects, from the plurality of candidate parses based on the scores for the plurality of candidate parses, the candidate what portion and the candidate where portion of a particular candidate parse included in the plurality of candidate parses as search parameters for the search query. Corresponding methods and systems are also disclosed.

Abstract: A method for tag-based search includes capturing an image, extracting a tag from the image, identifying a location associated with the captured image, and querying stored content for information that matches the location and the tag. Local storage is checked for the information first, and remote storage may be checked subsequently. Any located information may be used to augment the image. Information located in the remote storage may be saved in the local storage until it reaches a certain age, until it fails to be accessed for a threshold period of time, or until the location moves outside a threshold radius associated with a location of the information located in the remote storage.

Abstract: The disclosed embodiments provide a system for processing data. During operation, the system obtains a set of partitions containing records spanning a time interval, wherein the records include a set of values for a metric and a set of dimensions associated with the set of values. For each dimensional subset associated with the set of dimensions, the system aggregates, based on the set of partitions, a corresponding subset of values for the metric into a distribution-based representation of the metric. The system then stores a dictionary encoding of the dimensional subset in association with the distribution-based representation of the metric. Finally, the system calculates, based on the distribution-based representation, one or more quantiles associated with the dimensional subset and outputs the quantile(s) for use in characterizing a distribution of the metric.

Abstract: Systems and methods for providing search results based on time obtain an item are discussed. In an example, a method can include receiving a search query, generating search results, calculating a time to obtain for each listing in the search results, and arranging the search results for display. The search query can include a current location of a mobile device. The search results can include a plurality of listings, with each listing including a location. The locations representing either a physical location proximate the current location or a delivery time to the current location. The search results can be arranged for display in various manners according to the calculated time to obtain for each result.

Abstract: Certain aspects of the present disclosure provide techniques for determining similarities between businesses. One example method generally includes receiving a similarity query and receiving transaction data associated with a plurality of businesses for comparing the plurality of businesses. The method further includes generating a set of vectors representing the plurality of businesses based on the transaction data and generating a set of compact vectors based on the vectors by applying a dimensionality reduction technique. The method further includes generating based on the set of compact vectors, a similarity index and determining a response to the similarity query using the similarity index.

Abstract: Provided are systems, methods, and devices for providing topical similarities in a knowledge database. A knowledge database containing a number of articles is accessed, and clickstream data for the knowledge database is obtained including aggregated counts of user clicks on links within articles in the knowledge database. The clickstream data is sorted by source article, with each source article corresponding to a number of target articles that the source article links to. A directed article graph is constructed based on the sorted clickstream data. An input search term is received from an input device, and then an unbounded graph search is performed on the directed graph based on the input search term. Finally, a ranking heuristic for a predefined number of articles in the knowledge database is provided corresponding to topical similarity of the articles to the input search term.

Abstract: A method of enhancing one or more fields of a master database record based on a consumption of one or more data points provided via a third-party data subscription service is disclosed. An aggregated view corresponding to a master database record is accessed. The aggregated view identifies correspondences between the one or more fields of the master database record and the one or more data points provided via the third-party data subscription service. Based on a determination that one or more consumption criteria are met, the one or more data points are consumed and stored in a central database. The one or more fields of the master database record are updated based on the one or more consumed data points. The enhanced master database record is provided via a network to one or more computerized processes associated with one or more consumers of the enhanced master database record.