Abstract: Methods, systems, and media for image searching are described. Images comprising one query image and a plurality of candidate images are received. For each candidate image, a first model similarity measure from an output of a first model configured for scene classification to perceive scenes in the images is determined. Further, for each candidate image of the plurality of candidate images, a second model similarity measure from the output of a second model configured for attribute classification to perceive attributes in the images is determined. For each candidate image of the plurality of candidate images, a similarity agglomerate index of a weighted aggregate of the first model similarity measure and the second model similarity measure is computed. The plurality of candidate images based on the respective similarity agglomerate index of each candidate image are ranked and a first ranked candidate images corresponding to the searched images are generated.

Abstract: Techniques are provided herein for selecting and transmitting snippets from a messaging application. A “snippet” refers to an audio segment of a song that is less than the whole of the song. A user may request to view various audio segments (e.g., by category, by search, etc.) corresponding to portions of respective songs via a user interface of the messaging application. In some embodiments, an audio segment can be selected and metadata associated with that particular audio segment may be transmitted to another computing device where the audio segment can be played (e.g., streamed). In this manner, these snippets can be employed by the user to enhance their chat or texting conversation.

Abstract: Techniques are disclosed relating to database query optimizers. In some embodiments, a query parser of a database system receives a query requesting data from a multi-tenant database. The query includes a first constraint that restricts a query optimizer's selection of a set of execution plans available to implement the query. The database system determines whether an override has been specified for a particular tenant associated with the received query, the override indicating that the first constraint is to be replaced with a second constraint for the particular tenant. In response to determining that the override has been specified for the particular tenant, the database system replaces the first constraint in the query with the second constraint and provides the query with the second constraint to the query optimizer for selection of an execution plan in accordance with the second constraint.

Abstract: A system comprises an interface, a processor, and a memory. The interface is configured to receive a query. The processor is configured to: determine a set of nodes for the query; determine whether a node of the set of nodes comprises a first engine node type or a second engine node type, wherein determining whether the node of the set of nodes comprises the first engine node type or the second engine node type is based at least in part on determining whether the node is able to be executed in a second engine; and generate a plan based at least in part on the set of nodes. The memory is coupled to the processor and is configured to provide the processor with instructions.

Abstract: The invention is a method, an apparatus and a computer program for the automatic processing of data identifiers (20, 22), wherein data (12) associated with data identifiers (20, 22) is transmitted to a cloud platform (16) and to an analysis unit (32) there having at least one functional unit (45-50), wherein each functional unit (45-50) performs an analysis of the data (12) and/or associated data identifier (20, 22) and outputs a confidence value as an analysis result, wherein the or each confidence value is mapped to an overall confidence value (60), wherein the overall confidence value encodes a correlation between two data identifiers (20, 22), wherein data identifiers (20, 22) determined as correlated are combined under a higher-order variable name (66), and wherein the higher-order variable name (66) can be used for a database query, which supplies data relating to all data identifiers (20, 22) combined under the variable name (66), wherein data identifiers (20, 22) having an overall confidence value (60

Abstract: The present invention relates to hash join acceleration. In an embodiment, a computer receives a request to join build data rows with probe data rows. Each data row is associated with a respective join key value. From multiple buckets of a hash table, a bucket is selected that contains entries of multiple respective join key values in respective locations within the bucket. Whether or not the bucket contains an entry of the join key value of a build data row is detected. While building, multiple locations within the bucket are concurrently inspected to find an empty location in which to store the entry of the join key value of the build data row. The bucket is concurrently probed to detect whether or not the entries in the bucket include an entry of a join key value of a probe data row. For the join request, a response is sent that is based on the concurrent probing of the bucket.

Abstract: The embodiments described herein enable the efficient storage and searching of large amounts of data to retrieve desired data with high accuracy and fast response time. The embodiments include for example parsing an input document containing input data into multiple sub-documents based on the structure of the input document and generating for each sub-document, a unique ID, pathway, and metadata. The embodiments also include storage of the ID and pathway for each sub-document in an index database and storage of the input document itself in a separate database. When a query for information from the input document is received, the ID and pathway of the sub document containing the queried information are identified and used to locate the queried information from the input document.

Abstract: the present disclosure relates to systems, methods, and computer-readable media for optimizing and implementing operator trees based on a received query. For example, systems disclosed herein may generate an operator tree based on a received query. The systems described herein may systematically analyze the impact of bitvector filters in optimizing a join order of the operator tree to generate an optimized operator tree. The systems described herein may further implement the bit-vector aware operator tree by providing the optimized operator tree to an execution engine for further processing.

Abstract: An example method of detecting an element using an autonomous vehicle includes the following operations: using a sensor on the autonomous vehicle to capture image data in a region of interest containing the element, where the image data represents components of the element; filtering the image data to produce filtered data having less of an amount of data than the image data; identifying the components of the element by analyzing the filtered data using a deterministic process; and detecting the element based on the components.

Abstract: Embodiments described herein provide techniques for in-line compiling of UDFs in other programming languages. These techniques enable faster programming iterations because it allows users to compile directly in the cloud processing system. Moreover, it allows the UDFs to tie into existing libraries. The compiled results are treated as sensitive and handled with appropriate security policies, as with any other user data in the system.

Abstract: A simulated dataset is queried for regression by validating a structured query language (SQL) statement, determining a pattern type of the SQL statement, reconstructing the SQL statement according to a predetermined process for the pattern type, creating a mutated SQL statement for querying a simulated dataset, and validating the mutated SQL statement. The simulated dataset is based on a confidential dataset having the confidential elements removed or replaced.

Abstract: Methods and systems predict parameters in a dataset of an identified piece of (“information technology”) IT equipment. An automated method identifies datasets IT equipment in a same category of IT equipment as a piece of IT equipment identified as having incomplete dataset information. Each dataset of IT equipment parameters is used to construct generalized linear models of different classes of IT equipment within the category of IT equipment. The class of the identified IT equipment is determined. A predicted equipment parameter of incomplete information of the identified piece of IT equipment is computed using the generalized linear model associated with the class. The predicted equipment parameter can be used to complete the dataset of the identified piece of IT equipment.

Abstract: A method and apparatus managing a set of processors for a set of queries is described. In an exemplary embodiment, a device receives a set of queries for a data warehouse, the set of queries including one or more queries to be processed by the data warehouse. The device further provisions a set of processors from a first plurality of processors, where the set of processors to process the set of queries, and a set of storage resources to store data for the set of queries. In addition, the device monitors a utilization of the set of processors as the set of processors processes the set of queries. The device additionally updates a number of the processors in the set of processors provisioned based on the utilization/Furthermore, the device processes the set of queries using the updated set of processors.

Abstract: Joining data using a disjunctive operator using a lookup table is described. An example computer-implemented method can include receiving a query with a set of conjunctive predicates and a set of disjunctive predicates. The method may also include generating a lookup table for each predicate in the sets of conjunctive predicates and disjunctive predicates. The method, for each row in a probe-side table, may also further include looking up a value associated with that row in each of the lookup tables and adding the row to a results set when there is a match. Additionally, the method may also include returning the results set.

Abstract: A method and apparatus managing a set of processors for a set of queries is described. In an exemplary embodiment, a device receives a set of queries for a data warehouse, the set of queries including one or more queries to be processed by the data warehouse. The device further provisions a set of processors from a first plurality of processors, where the set of processors to process the set of queries, and a set of storage resources to store data for the set of queries. In addition, the device monitors a utilization of the set of processors as the set of processors processes the set of queries. The device additionally updates a number of the processors in the set of processors provisioned based on the utilization. Furthermore, the device processes the set of queries using the updated set of processors.

Abstract: In a multi-tenant computing system, a cross-tenant user search system receives a user search input from a user using a client of a first tenant. The cross-tenant user search system accesses a policy that identifies one or more related tenants that are related to the first tenant. The policy also identifies directory records of users in the related tenant that can be searched by users of the first tenant. The cross-tenant user search system executes a search against a directory for the first tenant and executes a separate search against the directory records in the related tenants, identified in the policy. The search results from searching the directory for the first tenant and the related search results from searching the directory records in the tenant are aggregated into an aggregated set of search results which are then returned to the client.

Abstract: According to one embodiment, a document search apparatus includes the following elements. The query word extraction unit extracts, from an input sentence, a word for use in a search as a query word. The intention estimation unit estimates a search intention based on the input sentence. The query generator acquires, from a search expression list, a search expression corresponding to the estimated search intention, acquires, from an additional query word list, an additional query word corresponding to a set of the acquired search expression and the extracted query word, and generates a query based on the extracted query word, the acquired additional query word, and the acquired search expression. The search unit searches for a document that matches the query.

Abstract: Methods, systems and programming for providing query suggestions. In one example, a search query is received from a user. At least one rewritten search query is generated by replacing at least some text in the search query with a synonym of the at least some text. The synonym of the at least some text is obtained based on a plurality of previously submitted search queries. A plurality of query suggestions is obtained based on the search query and the at least one rewritten search query. At least a portion of the plurality of query suggestions is provided to the user.

Abstract: Aspects of the present disclosure relate to systems and methods that help a user manage a functional product. Every functional product, be it a service, device, or combination thereof, has one or more documents associated with it. These documents may include such documentation as: (1) Release Notes; (2) Configuration Guides; (3) command line interfaces (CLIs)/application program interfaces (APIs); (4) Data Sheets; (5) Installation Guides; (6) User Manuals; (7) Errata notices; and (8) other documentation. In embodiments, command extracted from the documents along with features available in the functional product and user history (if any) may be used to help suggest one or more commands for a user managing or using the functional product.

Abstract: Embodiments of the present disclosure relate generally to semantic indexing to improve search results of a large corpus. Some embodiments identify one or more closest matches between (i) a search semantic vector that corresponds to a search query, the search semantic vector based on a first machine-learned model that projects the search query into a semantic vector space, and (ii) a plurality of publication vectors corresponding to respective publications in the publication corpus, the plurality of publication vectors based on a second machine-learned model that projects the plurality of publication vectors into the semantic vector space.

Abstract: Systems and methods for preloading an amount of content based on user scrolling are disclosed. A body of content may be presented that takes up a certain amount of display space within a graphical user interface. Scroll information characterizing user scrolling within the graphical user interface may be obtained. A portion of the body of content outside a field of view of the graphical user interface may be determined based on the scroll information.

Abstract: A computing system receives as inputs data records stored in a database. The computing parses the data records into file pairs that each include a keyword file and record ID file and merge file pairs into a keyword file and record ID file, where the keyword file includes keywords in sorted order, and where the record ID file includes a list of record IDs for keywords in the keyword file. The computing system further creates an offset file which stores offset values for starting addresses of record ID lists in the record ID file, and generates an index of keywords by assigning unique identifiers to keywords in the keyword file. The computing system also provides a query interface that allows the database's data records to be searched using the generated index of keywords.

Abstract: In one embodiment, a method includes by a client computing device, sending a request for two or more categories of content for display in a user interface, wherein each of the categories of content are received from a different content source, and wherein each of the categories of content are displayed in a corresponding region of the user interface. The method further includes determining a space requirement for each of the categories of content, wherein the determining is based at least in part on information in the request and information stored on the client computing device. The method further includes allocating, based on the space requirements, space for each of the regions of the user interface, and displaying each of the categories of content in its corresponding region as the content is received.

Abstract: Methods and apparatus are described herein for classifying user reviews or portions thereof as being related to various entities, and for associating extracted descriptive segments of text contained in those user reviews or portions thereof with entities based on the classifications. In various implementations, one or more categories of observed user interest may be identified based on a corpus of user queries. One or more segments of text related to the one or more categories of observed user interest may be detected in one or more user reviews associated with a product. Based on the detecting, the product may be indexed on the one or more categories of observed user interest in a searchable database. In some implementations, the searchable database may be accessible to one or more remote client devices, and may be searchable by the one or more categories of observed user interest to provide search results to be rendered by the one or more remote client devices.

Abstract: A dynamic content controller is configured for communication with one or more data sources. The dynamic content controller comprises an analytics engine and a personalization engine coupled to the analytics engine. The analytics engine is configured to analyze cultural data collected from the one or more data sources during a current content browsing session. The personalization engine is configured to adapt content to be presented based at least in part on the analysis of the cultural data. The adaptation of the content to be presented is performed during the current content browsing session.

Abstract: A system and method for a collection of digital items that includes generating a collection of result items, wherein each result item corresponds to result data of at least one query input, and wherein at least one result item includes a context property; storing the collection of result items as a first instance; and augmenting the collection in a second instance through at least the context property.

Abstract: Techniques are provided for improving performance of parallel execution plans that include serial operations. A database server generates a parallel execution plan and includes one or more serial data flow objects in the parallel execution plan, each data flow object including one or more serial row sources. A parallel selector row source is associated with each serial data flow object included in the parallel execution plan. Each parallel selector row source coordinates with a query coordinator process the execution of the associated serial data flow object. In the event that that a plurality of serial data flow objects are included in a parallel execution plan, serial row sources included in the serial data flow objects may be executed concurrently by a separate parallel slave process, instead of sequential execution of each serial row source by the query coordinator process.

Abstract: Techniques are provided for a fully-automated process for tuning database query language statements that selects database query language statements for tuning, tunes the database query language statements and generates tuning recommendations, tests the tuning recommendations, and determines whether to implement the tuning recommendations based on the test results. The fully-automated tuning process may also automatically implement certain tuning recommendations and monitor the performance of the database query language statements for which tuning recommendations have been implemented.

Abstract: A method and device may accept a first database query plan including a plurality of rows and a second database query plan including a plurality of rows, compare each row in the first database query plan with a corresponding row in the second database query plan, and based on each comparison, calculate a difference between the compared rows. Each difference may be recorded in an output file.

Abstract: Embodiments of the present invention provide hardware-friendly indexing of databases. In particular, forward and reverse indexing are utilized to allow for easy traversal of primary key to foreign key relationships. A novel structure known as a hit list also allows for easy scanning of various indexes in hardware. Group indexing is provided for flexible support of complex group key definition, such as for date range indexing and text indexing. A Replicated Reordered Column (RRC) may also be added to the group index to convert random I/O pattern into sequential I/O of only needed column elements.

Abstract: Certain example embodiments described herein relate to techniques for user feedback driven dynamic query rewriting in complex event processing (CEP) environments. In certain example embodiments, the adjustment of already-initiated queries may be permitted during runtime in order to help continuously or substantially continuously meet a user's requirements. In certain example embodiments, parameters of a query may be adapted dynamically, e.g., without the need to redeploy it. Adaptable parameters in certain example embodiments may be identified and/or based on, for instance, an automatic determination from original query creation, user specifications, the way in which a user interacts with a user interface for visualizing the results of queries, etc.

Abstract: A hybrid context information matching approach may produce a customized product catalogue based on the user's context and the mobile device the user is using. A Knowledge Base (KB) and a KB manager, along with various processes perform specific collaborative tasks in order to achieve the overall goal of producing a customized product catalogue. In addition, the effort builds and/or updates the KB. Various contextual inputs are provided from both the user environment and data repositories. Hybrid matching is performed in order to determine optimal search results based on the contextual input provided.

Abstract: A technique for cataloging documents based on user activity includes assigning documents to a relevant document list based on activity of a user of a device. In this case, at least two of the documents are associated with different applications. The technique then provides the relevant document list to the user.

Abstract: A database system maintains a feature set of a modern database system while operating directly on raw data files. Systems may use an adaptive indexing mechanism that maintains positional information to provide efficient access to raw data files, a flexible caching structure, and techniques for selective parsing and selective tokenizing. In doing so, possible performance bottlenecks associated with repeated parsing, tokenizing, and expensive data type conversion costs can be overcome.

Abstract: Embodiments of the present invention provide a method, system and computer program product for pre-migration performance prediction of a database management system (DBMS). In an embodiment of the invention, a method for pre-migration performance prediction of a DBMS can include executing a calibration workload in a target DBMS to produce a conversion factor of cost of executing the calibration workload to temporal performance of executing the calibration workload. The method also can include subsequently submitting a sample workload from a database of a source DBMS for evaluation of cost of execution on an empty replica of the database in the target DBMS. Finally, the method can include predicting a temporal performance of the sample workload in the target DBMS as a product of the conversion factor and the cost of execution of the sample workload on the empty replica of the database in the target DBMS.

Abstract: Detection and use of relationships in tabular data is enhanced. A foreign-key-based relationship between two tables in memory is automatically discovered by a module. A view over table data is computationally augmented by adding an expandable hierarchical column. The expandable hierarchical column includes second table data that is related to rows of a first table by the discovered foreign-key-based relationship. A naming scheme for the added column is described. Sometimes the augmentation places an AddJoinColumn operator in a query expression tree; sometimes the tree is rewritten by moving a query operator or replacing a specified pattern of operators by a flat Join operation between two tables. Column expansion may be followed by reshaping data into a flat table, by getting a homogeneous query, by performing a join, or other operations.

Abstract: The present invention extends to methods, systems, and computer program products for optimizing queries of parallel databases. Queries can be partially optimized at an optimizer that is unaware of its use to optimize queries for parallel processing. The optimizer can produce a data structure (e.g., a SQL Server MEMO) that encapsulates a logical serial plan search space. The logical serial plan search space may not incorporate any notion of parallelism into the plan space itself. A parallel-aware optimizer can parallelize the logical serial plan search space by augmenting the data structure (e.g., transforming the SQL Server MEMO into a parallel MEMO). Augmentation can be with data movement operations that move data associated one or more compute nodes in a distributed architecture. Cost estimates can be calculated for the operations contained in the parallelized data structure. The parallel plan with the lowest estimated cost can be selected for the query.

Abstract: A search index for a collection of documents includes a plurality of keywords associated with the documents. Access to individual documents is detected based on searches employing the search index and keywords are recorded that are utilized in the searches and resulted in document access. The search index is modified to maintain the recorded keywords and remove keywords absent from the searches resulting in the document access.

Abstract: A technique for factoring uncertainty into cost-based query optimization includes: determining the degree of uncertainty involved in the cost estimates for the query, determining the degree of sensitivity the query has to that uncertainty, and determining if there is an access path that performs well across the range of possible conditions that could occur at execution time, reducing the risk of performance spikes and performance volatility. If such an access path exists, select that access path; if not, perform parametric query optimization or query re-optimization.

Abstract: Systems and method of the present invention provide for one or more server computers configured to receive one or more keywords topically relevant to a content of a web page, request from a search engine a first metric comprising a quantity of times the keywords have appeared in a search query with one or more question keywords during a time period and a second metric comprising a probability of receiving a high rank associated with the one or more keywords and the one or more question keywords, receive, from the search engine, the first metric and the second metric, calculate a keyword effectiveness index from the first metric and the second metric, and generate and transmit to a client computer one or more recommendations to include a high ranked suggested content on the web page according to the keyword effectiveness index.

Abstract: In an embodiment, a recursive query is received that comprises a first select statement with a seed select statement and a second select statement with a recursive reference, wherein the recursive query further identifies at least two columns in at least one table, wherein the at least two columns have parent-child relationships represented by nodes in a graph, wherein the graph represents the organization of values in rows in the at least one table. A target recursion depth is calculated for the graph based on a fanout of the graph. In an embodiment, the target recursion depth is calculated by summing the fanout at each recursion depth of the graph multiplied by a number of nodes at each recursion depth of the graph. An estimated number of rows that the recursive query will return is estimated based on the target recursion depth.

Abstract: In an embodiment, a hit percentage of an expected affinity for a first query is calculated, wherein the expected affinity comprises a first address range in a query plan cache, a hit percentage of a candidate affinity for the first query is calculated, wherein the candidate affinity comprises a second address range in a query plan cache, and if the hit percentage of the candidate affinity is greater than the hit percentage of the expected affinity by more than a threshold amount, query plans in the candidate affinity are swapped with query plans in the expected affinity.

Abstract: Presented is a system and method for evaluating relational database queries in a distributed system. An optimized query plan is received by a control node. The query plan is decoded to a SQL statement that is semantically equivalent to the query plan, even though the query plan contains elements that have no direct analogue in SQL. The decoded SQL is transmitted to a compute node for execution.

Abstract: A system joins predicate evaluated column bitmaps having varying lengths. The system includes a column unifier for querying column values with a predicate and generating an indicator bit for each of the column values that is then joined with the respective column value. The system also includes a bitmap generator for creating a column-major linear bitmap from the column values and indicator bits. The column unifier also determines an offset between adjacent indicator bits. The system also includes a converter for multiplying the column-major linear bitmap with a multiplier to shift the indicator bits into consecutive positions in the linear bitmap.

Abstract: A method for providing optimized data representation of relations for in-memory database query processing is disclosed. The method seeks to optimize the use of the available memory by encoding relations on which the in-memory database query processing is performed and by employing auxiliary structures to maintain performance. Relations are encoded based on data patterns in one or more attribute-columns of the relation and the encoding that is selected is suited to a particular type of data in the column. Members of a set of auxiliary structures are selected based on the benefit the structure can provide and the cost of the structure in terms of the amount of memory used. Encoding of the relations is performed in real-time while query processing occurs, using locks to eliminate conflicts between the query processing and encoding.

Abstract: A system and method for improving performance of join operations is disclosed. The method in one embodiment includes building a graph of vertices and edges, which represent the tables and join conditions for a particular database schema. The method then traverses the graph to find a set of candidate join operations. The method further includes selecting from the set of subset of join operations that makes the best use of a given amount of memory, which means finding a collection of denormalizations which when residing in memory provide the most benefit and the least cost for the given amount of memory. The method then deploys the denormalizations so that current queries in a query processing system can use them instead of actually performing the joins.

Abstract: A method, apparatus, and stored instructions are provided for transforming a query representation by unnesting a predicate condition that is based on whether or not a result exists for a subquery of the predicate condition. An initial query representation is received. The initial query representation represents an initial query that includes an EXISTS-equivalent predicate or a NOT-EXISTS-equivalent predicate and at least one other predicate in a disjunction. The initial query representation is transformed into a semantically equivalent transformed query representation that represents a transformed query. The transformed query includes, instead of the EXISTS-equivalent predicate or a NOT-EXISTS-equivalent predicate, a join operator that references the data object.

Abstract: A system includes a coordinator node and worker nodes in communication with the coordinator node. Each worker node stores data blocks. Each data block has data in a semi-structured format and each data block has an associated foreign table declaration specifying conversion of the data in the semi-structured format into a tabular format interpretable by a query language. A query processor executed by the coordinator node produces a distributed query plan in response to a query language query. The distributed query plan includes sub-queries. The sub-queries are executed by selected worker nodes of the worker nodes. The selected worker nodes use foreign table declarations to convert data in semi-structured formats into tabular formats of a distributed database to provide tabular data in response to the query language query.

Abstract: Systems, methods, and apparatuses, including computer program products, are provided for machine translation using information retrieval techniques. In general, in one implementation, a method is provided. The method includes providing a received input segment as a query to a search engine, the search engine searching an index of one or more collections of documents, receiving one or more candidate segments in response to the query, determining a similarity of each candidate segment to the received input segment, and for one or more candidate segments having a determined similarity that exceeds a threshold similarity, providing a translated target segment corresponding to the respective candidate segment.

Abstract: In an embodiment, a first query is received that specifies a first object identifier. If a first query plan exists that implements the first query, except that the first query plan does not comprise the first object identifier and instead comprises a parameter marker in place of the first object identifier, a first query execution plan is created from the first query plan, substituting the first object identifier in the first query execution plan for the parameter marker, and the first query execution plan is executed to read a first object identified by the first object identifier.