Microsoft Technology Licensing, LLC

Amazon Technologies, Inc.

Imdb. Com, Inc.

Lars Hasso Liden

Paul J Walsh

Shahin Shayandeh

Swadheen Kumar Shukla

Matthew D Mazzola

Jianfeng Gao

Harsha Ramalingam

Michael A Carr

Samuel Ytzhak Donnelley

Thomas Park

Nishant Mittal

Graham T Rosser

Paul J Gardner

Steven Wayne Ickman

William L Portnoy

Thomas Matthew Laird-McConnell

Hai Liu

Daniel James Driscoll

Ashutosh Vishwas Kulkarni

Jinchao Li

Adrian Christopher Quan

Kati Amanda London

Eslam K Abdelreheem

Baolin Peng

David L Holcombe

Rachel Jane MacTaggart

Janice Von Itter

Garth A Henson

Systems and methods are provided for determining a response to a query in a dialog. An entity extractor extracts rules and conditions associated with the query and determines a particular task. The disclosed technology generates a transformer-based dialog embedding by pre-training a transformer using dialog corpora including a plurality of tasks. A task-specific classifier generates a first set of candidate responses based on rules and conditions associated with the task. The transformer-based dialog embedding generates a second set of candidate responses to the query. The classifier accommodates changes made to a task by an interactive dialog editor as machine teaching. A response generator generates a response based on the first and second sets of candidate responses using an optimization function. The disclosed technology leverages both a data-driven, generative model (a transformer) based on dialog corpora and a user-driven, task-specific rule-based classifier that accommodating updates in rules and conditions associated with a particular task.

Hybrid transformer-based dialog processor

This document relates to creating and/or updating a chatbot using a graphical user interface. For example, training dialogs for a chatbot can be displayed in a tree form on a graphical user interface. Based at least on interactions between a developer and the graphical user interface, the training dialogs can be modified in the tree form, and training dialogs can be updated based on the modifications provided on the tree form via the graphical user interface.

Visualization of training dialogs for a conversational bot

Methods and systems are disclosed for improving dialog management for task-oriented dialog systems. The disclosed dialog builder leverages machine teaching processing to improve development of dialog managers. In this way, the dialog builder combines the strengths of both rule-based and machine-learned approaches to allow dialog authors to: (1) import a dialog graph developed using popular dialog composers, (2) convert the dialog graph to text-based training dialogs, (3) continuously improve the trained dialogs based on log dialogs, and (4) generate a corrected dialog for retraining the machine learning.

Training a user-system dialog in a task-oriented dialog system

A computing system causes instructional media to be played on a device to a user. An instructor in the instructional media provides guidance as to how to perform an activity when the instructional media is played on the device. The computing system obtains user data pertaining to performance of the activity by the user. The computing system generates a user-customized portion of the instructional media based upon the user data and a computer-implemented model. The computing system causes the user-customized portion to be played on the device to the user, where the device emits audible words reproduced in a voice of the instructor, where the audible words are based upon the user data, and further where the device displays generated images of the instructor depicting the instructor speaking the audible words as the device emits the audible words.

Computationally customizing instructional content

Described herein are various technologies pertaining to assigning an emotional state to a computer-implemented entity. The emotional state is a function of data values in a data stream output by a data source, such as a sensor or a web service. The data values in the data stream are processed over time to generate a measure value. Values that are indicative of a dynamic state of a measure relative to an optimal measure state are computed and assigned to the measure based upon the measure value. A reaction of the computer-implemented entity is determined based upon these values. A customized personality for the computer-implemented entity maps reactions to respective emotional states, and an emotional state that is mapped to the reaction is assigned to the computer-implemented entity.

Assignation of emotional states to computer-implemented entities

Disclosed are various embodiments for presenting collections of items related to subjects in a performance to viewers. The collections are assumed to be identified prior to the performance by individuals with knowledge of subjects expected to appear in the performance. The subject's appearance may be detected based on a user input indicating the appearance or by a subject recognition algorithm. The collections may be presented to a customer in a display also rendering the performance or in a separate display. The customer may select the collections and/or items therein for performing an action with the item, such as, for example purchasing the item.

Medium, system, and method for identifying collections associated with subjects appearing in a broadcast

Techniques for providing friction-free transactions using geolocation and user identifiers are described herein. These techniques may ascertain a user's location based on a location of a mobile device. A transaction between the user and a merchant may be completed with zero or minimal input from the user based on the geolocation of the mobile device and the user identifiers. In some implementations, a transaction initiated earlier is completed when the mobile device arrives at the merchant. Additionally, a parent-child or similar relationship may be established between multiple devices. Security on the mobile device based may be provided by biometric identification and calculation of variance from regular movement patterns. Advertisements may be sent to the mobile device based on bids from merchants near to the mobile device. A relationship between the mobile device an a merchant may be bootstrapped from previous transactions that did not use the mobile device.

Transaction bootstrapping to create relationships

The techniques described herein are directed to cross-channel communication. The techniques can employ a central method, system, or mode ('Cross-Mode Manager' or 'CMM' herein), which is enabled to send and/or receive messages over a multiplicity of modes. Users can enable the CMM through a variety of methods described herein on modes which the user uses to communicates. Once enabled on at least one channel which the user uses to communicate, the presence of the CMM across a multiplicity of modes allows a user to create and/or join a cross-channel communication session maintained by the CMM and to communicate between the user's communication mode and any of the multiplicity of modes for which the CMM has a presence. In other words, the techniques permit users to communicate across disparate channels from within a channel of their choosing.

Cross-mode communiation

Various approaches described herein enable various types of users, such as finance managers, vendors, and vendor managers to access up-to-date information in a retail environment that can assist with obtaining and tracking various items to be offered through that environment. A budget planner can analyze information such as historical performance, projection, and vendor data to establish a budget plan for various providers for a particular period. The budget information is fed to an assortment planner, which can use similar and other types of information to allocate that budget across various types and styles of item for each provider. A commitment tracker enables a user to commit to at least some portion of the allocation for various items, and track how well the orders are filled. Information at each level is fed back into the system to be almost instantly available and to adjust allocations for current or future periods.

Budget planner for softlines

Disclosed are various embodiments for relating items to objects detected in media. An item recognition application obtains media information associated with a broadcast being consumed form the client device. The item recognition application identifies objects in the media based on user input of the location of the object on the client device display and/or heuristic detection of the object. The user may indicate the object's location through an input device. The object may be heuristically detected based on an algorithm that differentiates the object from the background. The item recognition application identifies items related to the object. The item recognition application then performs an action with the item.

Relating items to objects detected in media

Methods, computer systems, and computer-storage media are provided for connecting devices. Shared spaces may be created among one or more users and/or devices. Based on various factors including locations, histories among users, time of day, etc., spaces may be created among users such that content may be shared within the shared space. The spaces may be created using multiple technologies such that one user may connect to the space via a first identifier format while a second user may connect to the space via a second identifier format. Once created, content may be shared with one or more users of the space. Depending on privacy and/or security settings of the space, any user may invite other users to join the space. The content may be available to any participant once the space has terminated.

Creating connections and shared spaces

Techniques for providing friction-free transactions using geolocation and user identifiers are described herein. These techniques may ascertain a user's location based on a location of a mobile device. A transaction between the user and a merchant may be completed with zero or minimal input from the user based on the geolocation of the mobile device and the user identifiers. In some implementations, a transaction initiated earlier is completed when the mobile device arrives at the merchant. Additionally, a parent-child or similar relationship may be established between multiple devices. Security on the mobile device based may be provided by biometric identification and calculation of variance from regular movement patterns. Advertisements may be sent to the mobile device based on bids from merchants near to the mobile device. Points may be accumulated through transactions with merchants and later redeemed for free or discounted goods and/or services.

Transaction tracking and incentives

An outsourcing environment is described herein by which an outsourcing entity may delegate document-transformation tasks to at least one worker entity, while preventing the worker entity from gaining knowledge of sensitive items that may be contained within a non-obfuscated original document (NOD). In one example, the environment may transform the NOD into an obfuscated original document (OOD) by removing sensitive items from the NOD. The worker entity may perform formatting and/or other document-transformation tasks on the OOD, without gaining knowledge of the sensitive items in the NOD, to produce an obfuscated transformed document (OTD). The environment may then allow for the outsourcing entity to view a content-restored version of the OTD.

Growing community of inventors

Seattle, WA, United States of America

Lars Hasso Liden