Microsoft Technology Licensing, LLC

Christopher John Brockett

William B Dolan

Jianfeng Gao

Christopher Brian Quirk

Michel Galley

Zhang Li

Margaret Ann Mitchell

Alessandro Sordoni

Eric Breck

Michael Auli

Yangfeng Ji

Michael Gamon

Gary Kacmarcik

Kaushik Ramaiah Narayanan

Susan Michele Hendrich

Michael Wilson Daniels

Jian-Yun Nie

Maria Isabel Carpenter

Alice Yingming Lai

Noelle Mallory Sophy

Juhi Amitkumar Naik

Xuedong D Huang

Yun-Cheng Ju

David Alexander Molnar

Lucretia Henrica Vanderwende

Robert Alexander Sim

Hisami Suzuki

Patrick Pantel

Domenic J Cipollone

Dengyong Zhou

Vikram Reddy Dendi

Pallavi Choudhury

Zhenghao Wang

Olivier Gauthier

Piali Choudhury

Si-Qing Chen

Jesse Alexander Freitas

Scott Laufer

Zeqiu Wu

Christopher Ian Charla

Warren Anthony Aldred

Benjamin David Van Durme

Gabriel Arien Desgarennes

Ryan H Volum

Svitlana Volkova

Mei Li

Nicole Beaudry

Thomas E Woolsey

Vahed Qazvinian

Stanley Kok

Sudha Rao

Hamid Palangi

Weixin Cai

Xinyu He

Yizhe Zhang

Aspects of the present disclosure relate to a personalized agent service that generates and evolves customized agents that can be instantiated in-game to play with users. Machine learning models are trained to control the agent's interactions with the game environment and the user during gameplay. A user may request that a personalized agent join the user's gameplay session. The user device sends a request for the personalized agent to a game platform. The game platform determines whether the user has a license to execute a second instance of the game. When the user has a license to execute a second instance of the game, the second instance of the game may be executed on the user device. Information received from a personalized agent service is used to instantiate a personalized agent in the second instance of the game.

Importing agent personalization instantiate a personalized agent in a user game session

Systems and method directed to assistive document generation are described. More specifically, similar documents share large portions of reusable text structures that can be used to generate an initial document thereby saving a user time. To generate the document, an indication to create the document may be received and based on the indication, a plurality of example documents and grounding content may be identified. Example documents may be existing documents that are similar to a target document of the writer. Grounding information may refer to content that is relevant, timely, and accurate when applied to the target document. The plurality of example documents and the grounding content may be received, and a document sketch based on the example documents and the grounding content may be generated and contains a plurality of predicted text sequences based on the example documents and the grounding content.

Automatic document sketching

Systems and methods are directed to generating content that is contextually relevant in a writing style of a user. In example embodiments, a plurality of logical inputs regarding a topic is received in bullet point format. A content generator generates draft content using machine learning (ML) models. The generating comprises identifying a writing style of the user by applying the plurality of logical inputs to a first ML model, determining a context and direction for the draft content using a second ML model, and based on the plurality of logical inputs, the identified writing style, and the context and direction, generating at least one paragraph of draft content in the writing style of the user that follows an outline associated with the bullet point format and comprises a same context and direction as the plurality of logical inputs. The draft content is then presented at a client device.

Automatic content generation

A method and system for providing replacement text segments for a given text segment may include receiving a request to provide the replacement text segment for the text segment in the document, examining a content characteristic of the document, and examining at least one of user-specific information, organization-specific information, or non-linguistic features of the document, before identifying at least one replacement text segment for the text segment, via a machine translation system, based on the content characteristic of the document and at least one of the user-specific information, the organization-specific information, or the non-linguistic features of the document. The method and system may include providing the identified replacement text segment for display to a user, receiving an input indicating a user's selection of the identified replacement text segment, and upon receiving the input, replacing the text segment in the document with the identified replacement text segment.

Method and system for intelligently suggesting paraphrases

A method for providing targeted rewrites can include receiving a selection of text in a file; generating a set of target rewrites of the selection of text, the set of target rewrites comprising: at least one phrase or sentence having semantic similarity to a phrase or sentence of the selection of text; and a style that corresponds to a particular target style, wherein a target style is a representative style for a genre, profession, or environment; and providing for selection one or more of the target rewrites of the set of target rewrites.

Targeted rewrites

Examples are generally directed towards context-sensitive generation of conversational responses. Context-message-response n-tuples are extracted from at least one source of conversational data to generate a set of training context-message-response n-tuples. A response generation engine is trained on the set of training context-message-response n-tuples. The trained response generation engine automatically generates a context-sensitive response based on a user generated input message and conversational context data. A digital assistant utilizes the trained response generation engine to generate context-sensitive, natural language responses that are pertinent to user queries.

Context-sensitive generation of conversational responses

Examples are generally directed towards automatic assessment of machine generated conversational responses. Context-message-response n-tuples are extracted from at least one source of conversational data to generate a set of multi-reference responses. A response in the set of multi-reference responses includes it context-message data pair and rating. The rating indicates a quality of the response relative to the context-message data pair. A response assessment engine generates a metric score for a machine-generated response based on an assessment metric and the set of multi-reference responses. The metric score indicates a quality of the machine-generated conversational response relative to a user-generated message and a context of the user-generated message. A response generation system of a computing device, such as a digital assistant, is optimized and adjusted based on the metric score to improve the accuracy, quality, and relevance of responses output to the user.

Metric for automatic assessment of conversational responses

Different advantageous embodiments provide a crowdsourcing method for modeling user intent in conversational interfaces. One or more stimuli are presented to a plurality of describers. One or more sets of describer data are captured from the plurality of describers using a data collection mechanism. The one or more sets of describer data are processed to generate one or more models. Each of the one or more models is associated with a specific stimulus from the one or more stimuli.

Crowdsourced, grounded language for intent modeling in conversational interfaces

The subject disclosure is directed towards developing a translation model for mapping search query terms to document-related data. By processing user logs comprising search histories into word-aligned query-document pairs, the translation model may be trained using data, such as probabilities, corresponding to the word-aligned query-document pairs. After incorporating the translation model into model data for a search engine, the translation model is used may used as features for producing relevance scores for current search queries and ranking documents/advertisements according to relevance.

Search query and document-related data translation

A processing system is described which generates stimulus information (SI) having one or more stimulus components (SCs) selected from an inventory of such components. The processing system then presents the SI to a group of human recipients, inviting those recipients to provide linguistic descriptions of the SI. The linguistic information that is received thereby has an implicit link to the SCs. Further, each linguistic component is associated with at least one feature of a target environment, such as a target computer system. Hence, the linguistic information also maps to the features of the target environment. These relationships allow applications to use the linguistic information to interact with the target environment in different ways. In one case, the processing system uses a challenge-response authentication task presentation to convey the stimulus information to the recipients.

Generating stimuli for use in soliciting grounded linguistic information

There is provided a computer-implemented method for user query reformulation. A graph is created to represent a relationship between previous user query terms. The graph may represent the previous user searches in n-grams that correspond to nodes. A random walk analysis is performed to determine probabilities that various n-grams corresponding to nodes of the graph could be used to effectively alter a user search term. The probabilities represent a quantification of relationships between nodes of the graph. A determination may be made regarding whether to reformulate the user query based on a relationship between a user search term in the user query and a graphed search term represented by a node of the graph. The determination takes into account a relationship between the user search term and the graphed search term.

User query reformulation using random walks

An 'Interactive Word Lattice' provides a user interface for interacting with and selecting user-modifiable paths through a lattice-based representation of alternative suggested text segments in response to a user's text segment input, such as phrases, sentences, paragraphs, entire documents, etc. More specifically, the user input is provided to a trained paraphrase generation model that returns a plurality of alternative text segments having the same or similar meaning as the original user input. An interactive graphical lattice-based representation of the alternative text segments is then presented to the user. One or more words of each alternative text segment represents a 'node' of the lattice, while each connection between nodes represents a lattice &#x201c;edge. Both nodes and edges are user modifiable. Each possible path through the lattice corresponds to a different alternative text segment. Users select a path through the lattice to select an alternative text to the original input.

User-modifiable word lattice display for editing documents and search queries

Conceptually related term identification technique embodiments are presented that involve identifying in the results of a search query, terms that are conceptually related to the search query terms. In one general embodiment, this is accomplished by first inputting the terms employed in a search query and the results of the search. Word tokens found in the search query terms are then identified, as are potential phrases that can be made from the identified word tokens. Conceptually related words and phrases are then identified in the search query results. These words and phrases correspond to the previously identified word tokens and potential phrases. The search query results are presented to a user on a display device in such a manner as to visually distinguish the conceptually related words and phrases from other words and phrases in the search query results.

Identifying conceptually related terms in search query results

A method is described herein that includes acts of receiving a selection of a first phrase in a first language and executing a random walk over a computer-implemented multipartite graph, wherein the multipartite-graph includes a first set of nodes that are representative of phrases in the first language, a second set of nodes that are representative of phrases in a second language, and edges between nodes that are representative of relationships between the respective phrases. The random walk includes traversals over edges of the graph between nodes. The method also includes the act of indicating that a second phrase in the first language is a paraphrase of the first phrase based at least in part upon the random walk.

Locating paraphrases through utilization of a multipartite graph

Growing community of inventors

Bellevue, WA, United States of America

Christopher John Brockett