Microsoft Technology Licensing, LLC

Drugstore. Com

Umesh Madan

Geary L Eppley

David Wortendyke

Aaron Stern

Devis Lucato

Samuel Edward Schillace

Mehul Y Shah

Bryan Dove

Robert May

Saurab Nog

Ali Reza Emami

Kishore Chaliparambil

Eric Gunnerson

Geary Eppley

Chi Kit Chan

Imran Mohiuddin

Luis Felipe Cabrera

Christopher G Kaler

Erik B Christensen

Kenneth David Wolf

John Peter Shewchuk

Stephen James Millet

Giovanni Moises Della-Libera

James Ernest Johnson

Tirunelveli R Vishwanath

Lance Olson

Richard Douglas Hill

Stefan R Batres

Sean P Nolan

Michael James Coulson

Stefan Harrington Pharies

John David Doty

Craig Andrew Critchley

Michael Marucheck

Uday Sooryakant Hegde

Vijay Gajjala

Elliot Lee Waingold

Krishnan Srinivasan

Matthew Thomas Tavis

Tomasz Janczuk

David Edwin Levin

Stephen T Swartz

Michael Steven Vernal

Shy Cohen

Max A Feingold

Richard Dievendorff

Jeffrey Dick Jones

Natasha Harish Jethanandani

Douglas M Purdy

Cesar Martin Ruiz-Meraz

David Owen Driver

Siddhartha Puri

Douglas A Walter

Jesus Ruiz-Scougall

Anastasios Kasiolas

Shawn Cantin Callegari

Kapil Gupta

Vipul A Modi

Ryan Thomas Sturgell

Evgeny Osovetsky

Alexander Martin DeJarnatt

Adrian Bonar

Sowmyanarayanan K Srinivasan

Mohammad Makarechian

Scott Christopher Seely

Jonathan T Wheeler

Gaurav Dinesh Kalmady

Erin Honeycutt

Chaitanya Dutt Upadhyay

Michael Thomas Dice

Nicholas Becker

Mauro Ottaviani

Yaniv Pessach

Brian Scott Krabach

Alfred McClung Lee, Iv

Alfred M Lee, Iv

Eugene Shvets

Yasser Shohoud

Daniel J Young

LeRoy F Miller

Alexander Chao

Abby Harrison

Gina Elizabeth Triolo

Mark Karle

Tara Eve Walker

Pompiliu Diplan

Anand Rjagopalan

Mohamed-Hany Essam Ramadan

Serge Sverdlov

Vaithialingam B Balayogan

Ceary L Eppley

Aspects of the present disclosure relate to systems and methods for creating a multi-dimensional entity (MDE) based on natural language (NL) input. A user may provide NL input into an application. One or more skills may be identified for the NL input, each of which has an associated prompt template. For example, a skill is associated with a computer-aided design and/or three-dimensional manufacturing application and/or file format, thereby enabling the generation of output associated with such applications and/or file formats. In examples, a skill chain may be generated that includes one or more skills with which to generate MDE output accordingly.

Multi-dimensional entity generation from natural language input

Methods, systems, and media for storing entries in and/or retrieving information from an embedding object memory are provided. In some examples, a content item is received that has content data. The content data associated with the content item may be provided to one or more semantic embedding models that generate semantic embeddings. From one or more of the semantic embedding models, one or semantic embeddings may be received. The one or more semantic embeddings may then be inserted into the embedding object memory. The semantic embeddings may be associated with respective indications corresponding to a reference to source data associated with the semantic embeddings. Further, the insertion may trigger a spatial storage operation to store a vector representation of the one or more semantic embeddings. A plurality of collections of stored embeddings may be received from the embedding object memory, based on a provided input, to determine an action.

Storing entries in and retrieving information from an embedding object memory

In accordance with examples of the present disclosure, a collaborative platform provides a digital collaboration assistant that continuously monitors and analyzes shared meeting contents (e.g., voice, text chat messages, shared links and documents, presentation materials, and the like) by participants during a collaborative meeting in near real-time, periodically updates a structure summary log of the meeting contents that are deemed important during the collaborative meeting, and interacts with the participants throughout the collaborative meeting in near real-time, for example, to answer questions or provide additional information.

Meeting insights with large language models

The described implementations relate to updating views in a database system. One implementation can identify an active view that retrieves records from a database and determine one or more physical tables in the database that are referenced by the active view. The implementation can track one or more columns in the physical tables that are used by the active view and identify a change to the physical tables in a transaction log of the database. If the change does not affect the one or more columns, the change can be treated as an ignorable change for the active view.

Data change tracking and event notification

In one example, a phrase analyzer may analyze a text input stream to identify phrases contained in the text input stream. The phrase analyzer may receive a specification, which includes dictionaries of phrases and synonyms, and a specification of the phrases, or sequences of phrases to be matched. The phrase analyzer may compare the input stream to the specification and may produce, as output, an identification of which phrases appear in the input stream, and where in the input stream those phrases occur.

Text analysis using phrase definitions and containers

In response to receiving a request for data stored according to a first version of a data type, a second version of the data type supported by the requester is determined, and the requested data is transformed from the first version to the second version to provide access to the data.

Data versioning through data transformations

At least one implementation described herein relates to evaluating queries structured according to a first format against one or more objects structured according to a second, different format without serializing the object. Typically, evaluating an XML (eXtensible Markup Language) query against a CLR (Common Language Runtime) object required serializing the CLR object and deriving an XML construct from the serialized data. The query is then be evaluated against the XML construct. In the described implementations, significant system resources are realized by creating an infoset model that maps properties of an object to an object infoset. The query is then evaluated using the infoset to locate object values and, thus, no serialization is required. Dynamically generated IL is used and re-used to efficiently perform subsequent evaluation steps on similar queries.

Evaluating queries against in-memory objects without serialization

Embodiments herein prevent or mitigate attacks on inverse query engines by providing safe mode routines that allow for the acceptance of third party messages and/or query expressions, as well as prevent trusted sources from accidental attacks. The mitigations fall into two categories: compile-time and runtime. Compile-time mitigations prevent query expressions from being accepted and compiled that are susceptible to known attacks. For example, the complexity of query expressions may be limited to functions with linear runtimes; constant memory usage; or ones that do not create large strings. Further, language constructs for the criteria in the query expression may not allow for nested predicates complexities. Runtime mitigations, on the other hand, monitor the data size and processing lengths of messages against the various query expressions. If these runtime quotas are exceeded, an exception or other violation indication may be thrown (e.g., abort), deeming the evaluation as under attack.

Safe mode for inverse query evaluations

A scalable event notification system is provided. The event notification system manages the set of topics available for notification via a topic manager network. The event notification system transmits event notifications to event consumers via a subscription manager network. The configuration and operation of the subscription manager network may to reconfigured in response to changing load and fault conditions to facilitate the timely delivery of event notifications. The notification to subscribers also includes mechanisms to provide individual privacy and non-tampering of notifications. Notification may dynamically be delivered to alternative locations and localized to accommodate different natural languages.

Large scale event notification system

Implementations described herein relate to evaluating multiple queries represented in an opcode tree having opcode nodes and branch nodes. Each opcode node comprises an opcode that is a part of an opcode block derived from compilation of a query, the opcode block consisting of a linear sequence of opcodes. Opcode nodes are arranged in the tree according to location of its corresponding opcode in the opcode block with opcodes that occur earlier in the opcode block being represented at a higher tree level than subsequent opcodes. Only one opcode node corresponds to a common opcode block segment that is included in multiple queries. Executing the opcode tree against an input evaluates each of the multiple queries represented in the opcode tree.

Systems and methods for parallel evaluation of multiple queries

At least one implementation described herein relates to using multiple filter engines to optimize query processing. A filter engine comprises a general matcher and at least one optimized matcher, a matcher being a filter engine in its own right. When the filter engine receives an input, the input is analyzed to determine if it can be handled by the optimized matcher. While the general matcher is fully compliant with a query language, the optimized matcher only handles a subset of the query language. Therefore, inputs that can be processed in the optimized matcher are compared against fewer filters, making the filtering process more efficient. The filter engine may also process only a portion of an input in the optimized matcher and another portion of the input in the general matcher, which reduces processing overhead.

Systems and methods for filter table optimization

The present invention extends to methods, systems, and computer program products for relaying messages through firewalls. A message relay negotiates first communication parameters with relay components inside a firewall and (potentially different) second communication parameters with applications outside the firewall. The message relay receives a message from an outside application in accordance with the second negotiated parameters and maps an electronic address in the received message to an inside application. The message relay transforms the message and sends the transformed message over an established connection through the firewall to a relay component in accordance with the first negotiated parameters. The relay component receives the transformed message and transforms the transformed message back into the message. The relay component delivers the message to the inside application in accordance with the second negotiated parameters.

Relaying messages through a firewall

An inverse query engine system having a cache integrated therein and methods for maintaining the cache are described herein. Entities storing filters in the inverse query engine cache thus do not have to maintain an individual cache to maintain the filters. The inverse query engine system cache is maintained within a bounded size by removing expired filters from the cache and trimming the cache to an optimal size when the cache reaches a maximum size. In the event that filter size cannot be determined efficiently, weights may be associated with filters stored in the cache, each weight identifying an estimate of a filter size. The weights are used to determine cache size and to maintain the cache size within particular bounds.

Inverse query engine systems with cache and methods for cache maintenance

Growing community of inventors

Bellevue, WA, United States of America

Umesh Madan