Microsoft Technology Licensing, LLC

Amazon Technologies, Inc.

Wei Yu

Qi Zhang

Istvan Cseri

Ramachandran Venkatesh

Beysim Sezgin

Shankar Pal

Jose A Blakeley

Milind Madhukar Joshi

Oliver Nicholas Seeliger

Gideon Schaller

Denis Y Altudov

Denis Churin

Dragan Tomic

Hyen Vui Chung

Nengwu Zhu

Stefano Stefani

Torsten Walther Grabs

Cesar Alejandro Galindo-Legaria

Conor J Cunningham

Bhavesh Anil Doshi

Shenghua Ye

Eric N Hanson

Michael Christopher Wenneman

Mikhail Bilenko

Ganapathy Subramaniam Krishnamoorthy

Campbell Bryce Fraser

Amir Jonatan Padovitz

Bruno H M Denuit

Balaji Rathakrishnan

Michael T Black

Peter Kukol

Andrew Scott Richardson

Adrian Sorin Baras

Frederic Morin

Joshua Stephen Ullom

Yubai Di

Cesar A Galindo-Legaría

Dima Suponau

Qihui Lee

F James Eisenhart

Dmytro Ivashchenko

Qihui Li

Nathan Manning

Chunyi Huang

Xinyu Liu

Siyuan Zheng

Xiaofeng Xie

Weiwen Xue

Bo Gao

Described is a query optimizer comprising a query tuner that performs actual execution of query fragments to obtain actual results during compilation time, and uses those actual results to select a query plan. The actual results may be combined with estimates for fragments that were not executed. The tree may be traversed in a top-down traversal, processing every node. Alternatively, the tree may be traversed in a bottom-up traversal, re-deriving data for higher nodes as each lower level is completed. A limit, such as a time limit or level limit, may be used to control how much time is taken to determine the execution plan.

Query execution plans by compilation-time execution

In large distributed computing environments, application execution may be distributed between a plurality of groups, the plurality of groups containing a set of host computer systems responsible for the execution of one or more operations of the application. Group membership may be determined by generating configuration information based at least in part on the plurality of groups. The configuration information may be provided to a plurality of host computer systems and each host computer system of the plurality of host computer systems may determine membership to a particular group of the plurality of groups based at least in part on the configuration information.

Membership self-discovery in distributed computing environments

Technologies are disclosed for providing a large scale data join service within a service provider network. A data set includes first and second sets of files that correspond to each other. Each file includes a first identifier (ID) and a second ID. The first set of files is partitioned based at least in part upon the first ID into a plurality of first subsets of files and the second set of files is partitioned based at least in part upon the first ID into a plurality of second subsets of files. Files within a first group of the plurality of first subsets and files within a second group of the plurality of second subsets are encoded into first and second bitsets, respectively, based at least in part upon the second IDs. An exclusive-or operation is performed on the first and second bitsets to find discrepancies between the data files.

Large scale data join service within a service provider network

A system is provided, including a display unit, a memory unit, and a processor. The processor is configured to calculate a mutual information value between a first chunk and a second chunk, and to add a new word to a language unit when a condition involving the mutual information value is satisfied. The new word is a combination of the first chunk and the second chunk. The processor is also configured to add the new word into an n-gram store. The n-gram store includes a plurality of n-grams and associated frequency or count information. The processor is also configured to alter the frequency or count information based on the new word.

Adaptive generation of out-of-dictionary personalized long words

Systems and methods to compress MQDF data are disclosed herein. A plurality of eigenvectors is identified. Each eigenvector in the plurality of eigenvectors can correspond to a pattern to be recognized. Each eigenvector in the plurality of eigenvectors can be split into sub-vectors. The sub-vectors can then be grouped into one or more groups according to a location of the sub-vectors within each of the eigenvectors. Each group can be associated with location data of the sub-vectors in the group. At least one group can be compressed according to a codebook. The codebook can be identifiable via the location data.

Compression of MQDF classifier using flexible sub-vector grouping

A memory footprint of an Modified Quadratic Discriminant Function (MQDF) pattern recognition classifier is reduced without resulting in unacceptable classification accuracy degradation. Covariance matrices for multiple classes are clustered into a smaller number of matrices where different classes share the same set of eigenvectors. According to another approach, different numbers of principal components are stored for different classes based on criteria such as class usage frequency, larger variation in writing, and the like, resulting in fewer principal components to be stored in memory.

Flexible MQDF classifier model compression

A two-phase process quickly and accurately identifies representations of the same items within a collection of item representations. In the first phase, referred to as a 'blocking phase,' frequency information indicating the frequency with which terms appear within the collection of item representations is used to quickly identify 'candidate pairs' (i.e., pairs of item representations that have a relatively high probability of matching). The blocking phase results in a reduced subset of the data for further analysis during the second phase. In the second phase, referred to as a &#x201c;matching phase,&#x201d; the candidate pairs are analyzed using fuzzy matching functions to accurately identify &#x201c;matching pairs&#x201d; (i.e., representations of the same items).

Large scale item representation matching

Modifying a portion of an XML document stored in a relational database includes converting operations on the document into an operator tree. The operator tree includes a first operator corresponding to a first portion of the schema for the modified data and a second operator corresponding to a parent portion of the schema for the modified data. A first and a second portion of the schema are extracted from the overall document schema. A first validation of the modified data using the first operator with the first portion of the schema is performed. This forms a partially validated XML document portion. Next, a second validation of the partially validated XML document portion with the second portion of the schema is performed using the second operator. The XML document constraints are verified and the modified data is merged into the XML document, thus avoiding a full validation of the XML document when certain XML Schema types are absent in the modified data.

Partial XML validation

Various embodiments of the present invention are directed to a database with an extensible type system for at least one user-defined type that comprises information for describing its structure as well as information for describing an inheritance relationship between itself and another user-defined type (either a subtype or a supertype/base type). For certain embodiments, the user-defined type is defined in something other than Sequential Query Language (SQL) statement such as, for example, a Common Language Runtime (CLR) statement, a statement in C, C++, and C# ('C-sharp'), and/or a visual basic statement. In any event, several embodiments are further directed to a system wherein the database is aware of the inheritance relationship between two user-defined types by an explicit registration of the user-defined types with said database.

Systems and methods for supporting inheritance for user-defined types

A suggested query execution plan is used as a constraint on a guided optimization process such that all execution plans that do not conform to the suggested execution plan are eliminated as possible alternatives. If the guided optimization process successfully generates a resulting execution plan, then the result will conform to the suggested execution plan. If the guided optimization process fails to generate a resulting execution plan, then it may be determined that the suggested execution plan is invalid.

System and method for forcing a query execution plan

Several embodiments of the present invention are directed to systems and methods for extending the UDT framework of an extended relational data store (ERDS) to include support for unordered collections (multisets) and ordered collection (lists). More specifically, several embodiments of the present invention use an UDT infrastructure, CLR generics, and a new UNNEST operator to create and utilize a special type abstraction for collections that is simultaneously a scalar and a relation. As a scalar, this collection type can be processed by all parts of the data store engine that understand scalars (including but not limited to the client stack) and, as a relation, this collection type is queriable like any other type of relation.

Systems and methods for the implementation of unordered and ordered collections in a data store

An extensible markup language (XML) expression included in a query is parsed to yield an abstract syntax tree. The abstract syntax tree is then transformed into a unified tree including XML algebra operations. The unified tree is then converted into a relational tree including enhanced relational algebra operations. The relational tree is then forwarded to a query processor for optimization and execution of the query at the query processor.

System and method for compiling an extensible markup language based query

Growing community of inventors

Issaquah, WA, United States of America

Wei Yu