International Business Machines Corporation

Google Inc.

Oipenn, Inc.

Mark Edward Epstein

Pedro Jose Moreno Mengibar

Michael Alan Picheny

Fadi Biadsy

Raimo Bakis

Robert T Ward

William Stuart Meisel

Martin Jansche

Miroslav Novak

Pedro J Mengibar

Dimitri Kanevsky

David Nahamoo

James Robert Lewis

Ciprian Ioan Chelba

Martin Doston Carberry

Glenn T Puchtel

Susan M Wingate

Ridley M Whitaker

Yuqing Gao

Ponani S Gopalakrishnan

Jan Sedivy

Salim Estephan Roukos

Martin Franz

Ruhi Sarikaya

Peter F Brown

Kevin Brian Smith

Stephen A Della Pietra

Steven V De Gennaro

Hakan Erdogan

Jerome L Quinn

Vladislav Schogol

Peter V Desouza

Joshua David Koppelman

Sharon B Jones

Jean-Christophe Marcadet

Lucy Vasserman

Stephan Herman Maes

A computer-implemented method can include identifying a first set of text samples that include a particular potentially offensive term. Labels can be obtained for the first set of text samples that indicate whether the particular potentially offensive term is used in an offensive manner. A classifier can be trained based at least on the first set of text samples and the labels, the classifier being configured to use one or more signals associated with a text sample to generate a label that indicates whether a potentially offensive term in the text sample is used in an offensive manner in the text sample. The method can further include providing, to the classifier, a first text sample that includes the particular potentially offensive term, and in response, obtaining, from the classifier, a label that indicates whether the particular potentially offensive term is used in an offensive manner in the first text sample.

Classification of offensive words

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for building language models. One of the methods includes identifying a first group of one or more users associated with a user in a social network. The method includes identifying first linguistic information associated with the first group. The method includes generating a first language model based on the first linguistic information. The method includes identifying a second group of one or more users associated with the user. The method includes identifying second linguistic information associated with the second group. The method includes generating a second language model based on the second linguistic information. The method includes associating the first language model and the second language model with the user.

Building language models for a user in a social network from linguistic information

This document describes, among other things, a computer-implemented method. The method can include obtaining a plurality of text samples that each include one or more terms belonging to a first class of terms. The plurality of text samples can be classified into a plurality of groups of text samples. Each group of text samples can correspond to a different sub-class of terms. For each of the groups of text samples, a sub-class context model can be generated based on the text samples in the respective group of text samples. Particular ones of the sub-class context models that are determined to be similar can be merged to generate a hierarchical set of context models. Further, the method can include selecting particular ones of the context models and generating a class-based language model based on the selected context models.

Clustering classes in language modeling

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for acoustic model generation. One of the methods includes identifying one or more demographic characteristics for a user of a social networking site. The method includes receiving speech data from the user, the speech data associated with a user device. The method includes storing the speech data associated with demographic characteristics of the user and the user device.

Using social networks to improve acoustic models

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating language models. In some implementations, data is accessed that indicates a set of classes corresponding to a concept. A first language model is generated in which a first class represents the concept. A second language model is generated in which second classes represent the concept. Output of the first language model and the second language model is obtained, and the outputs are evaluated. A class from the set of classes is selected based on evaluating the output of the first language model and the output of the second language model. In some implementations, the first class and the second class are selected from a parse tree or other data that indicates relationships among the classes in the set of classes.

Generating language models

Methods and systems for facilitating development of voice-enabled applications are provided. The method may comprise receiving, at a computing device, a plurality of actions associated with a given application, parameters associated with each respective action, and example instructions responsive to respective actions. The method may also comprise determining candidate instructions based on the actions, parameters, and example instructions. Each candidate instruction may comprise one or more grammars recognizable by a voice interface for the given application. The method may further comprise the computing device receiving respective acceptance information for each candidate instruction, and comparing at least a portion of the respective acceptance information with a stored acceptance information log comprising predetermined acceptance information so as to determine a correlation. Still further, the method may comprise determining a set of instructions comprising one or more of the candidate instructions based on the determined correlation.

Methods and systems for determining instructions for applications that are recognizable by a voice interface

Some implementations include a computer-implemented method. The method can include providing a training set of text samples to a semantic parser that associates text samples with actions. The method can include obtaining, for each of one or more of the text samples of the training set, data that indicates one or more domains that the semantic parser has associated with the text sample. For each of one or more domains, a subset of the text samples of the training set can be generated that the semantic parser has associated with the domain. Using the subset of text samples associated with the domain, a language model can be generated for one or more of the domain. Speech recognition can be performed on an utterance using the one or more language models that are generated for the one or more of the domains.

Language modeling in speech recognition

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training recognition canonical representations corresponding to named-entity phrases in a second natural language based on translating a set of allowable expressions with canonical representations from a first natural language, which may be generated by expanding a context-free grammar for the allowable expressions for the first natural language.

Bootstrapping named entity canonicalizers from English using alignment models

A method includes accessing data specifying a set of actions, each action defining a user device operation and for each action: accessing a corresponding set of command sentences for the action, determining first n-grams in the set of command sentences that are semantically relevant for the action, determining second n-grams in the set of command sentences that are semantically irrelevant for the action, generating a training set of command sentences from the corresponding set of command sentences, the generating the training set of command sentences including removing each second n-gram from each sentence in the corresponding set of command sentences for the action, and generating a command model from the training set of command sentences configured to generate an action score for the action for an input sentence based on: first n-grams for the action, and second n-grams for the action that are also second n-grams for all other actions.

Increasing semantic coverage with semantically irrelevant insertions

A method iteratively processes data for a set of actions, including: for each action: accessing a corresponding set of command sentences for the action, determining first n-grams that are semantically relevant for the action and second n-grams that are semantically irrelevant for the action, and identifying, from a log of command sentences that includes command sentences not included in the corresponding set of command sentences, candidate command sentences that include one first n-gram and a third n-gram that has not yet been determined to be a first n-gram or a second n-gram; for each candidate command sentence, determining each third n-gram that is semantically relevant for an action to be a first n-gram, and determining each third n-gram that is semantically irrelevant for an action to be a second n-gram, and adjusting the corresponding set of command sentences for each action based on the first n-grams and the second n-grams.

Mining data for natural language system

The present disclosure provides methods operable by computing device having one or more applications configured to perform functions based on a received verbal input. The method may comprise receiving a verbal input, obtaining one or more textual phrases corresponding to the received verbal input, and providing the one or more textual phrases to an appropriate application on the computing device. The method may further comprise accumulating data on the one or more textual phrases. The data comprises at least a count of a number of times a particular textual phrase is obtained based on a given received verbal input. Based on the count exceeding a threshold, the method may further comprise providing a query corresponding to the textual phrase, where the query is usable to search an advertisement database for one or more advertisements relating to the textual phrase.

Determining advertisements based on verbal inputs to applications on a computing device

A method, a system, and an apparatus biasing a speech recognizer based on prompt context. The present invention is capable of analyzing the words used in the prompt given to the user. Then, a set of words the user is likely to say in response to the prompt is determined. The word set may be determined using a technology used by the speech recognition system, such as n-grams, grammars, or both. The speech recognition system boosts the probabilities of the analyzed words in the word set by a preconfigured amount. The preconfigured amount is selected based on collected data.

Biasing a speech recognizer based on prompt context

A system and method for speech recognition includes generating a set of likely hypotheses in recognizing speech, rescoring the likely hypotheses by using semantic content by employing semantic structured language models, and scoring parse trees to identify a best sentence according to the sentence's parse tree by employing the semantic structured language models to clarify the recognized speech.

Semantic language modeling and confidence measurement

A system and method for servicing natural language requests with a plurality of remote host systems. The system utilizes a computer program that comprises: (1) an input system for inputting an NL command; (2) a translation system that extracts a request from the NL command and stores the request in a host-independent format; and (3) a routing system for servicing the request, wherein the routing system comprises a mechanism for selecting a host, for converting the request into a host dependent directive, and for forwarding the directive to the selected host. The system may further include a speech recognition system, a local data source for servicing the NL command, templates for converting the request into the host dependent directive, a heuristic for selecting the host, and an output system for obtaining and outputting the response.

Growing community of inventors

Katonah, NY, United States of America

Mark Edward Epstein