Abstract: In one aspect, at least one processor may receive a first user input and identify one or more first keywords from the first user input to determine one or more previously recorded outputs corresponding to the first keywords. The previously recorded outputs may correspond to a virtual character, such as a celebrity. Further, the at least one processor may select one of the determined previously recorded outputs and output the selected previously recorded output corresponding to the virtual character. The at least one processor may also identify one or more second keywords from at least the first user input, temporarily store the second keywords in memory in association with at least one portion of the user input, receive a second user input, and determine another previously recorded output based at least in part on the temporarily stored keywords associated with the first user input.

Abstract: An approach is provided in which a knowledge manager analyzes multiple document phrases using a natural language processing model and generates multiple interpretations based upon the analysis. The knowledge manager identifies misinterpretation patterns by comparing the multiple interpretations with multiple corrections that include corrections to the multiple interpretations. In turn, the knowledge manager generates interpretation rules based upon the identified patterns and applies the interpretation rules to the natural language processing model.

Abstract: According to certain aspects, the present invention provides a social platform for developing informed communities. These communities are either hyper local or super focused within a specific geographical location; the social information created within the system of the invention is organized, mined and delivered to other community peers seeking information about important and time critical daily matters. Since these communities are organized around common interests, geographical location and/or goals, the participants will benefit from the information shared by their community peers. In embodiments, users of the invention's communities can ASK, SAY and stay connected with information on daily matters in their local communities that are organized around family, living, neighborhoods, schools, activities, special interests or causes.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for natural language processing. One of the methods includes receiving a first voice query; generating a first recognition output; receiving a second voice query; determining from a recognition of the second voice query that the second voice query triggers a correction request; using the first recognition output and the second recognition to determine a plurality of candidate corrections; scoring each candidate correction; and generating a corrected recognition output for a particular candidate correction having a score that satisfies a threshold value.

Abstract: A system, methods, and user interface for extracting information from unstructured data sources and presenting such information in a structured or semi-structured format for better information search and utilization, and can be applied to replace the conventional methods of displaying search results. The methods identify terms representing topics and related comments in various types of text contents including documents and Web pages, and extract such terms and present them in a form of a topic-comment or object-properties hierarchy, including a heading+list format and heading+cloud or group format. Methods and interface object are provided to make a file object a non-terminal node in a computer file system, with information extracted from the file content displayed as deeper levels of the file system hierarchy. Methods for displaying information extracted from unstructured document contents in terms of class-members and topic-attributes are also disclosed.

Abstract: In some implementations, data that indicates multiple candidate transcriptions for an utterance is received. For each of the candidate transcriptions, data relating to use of the candidate transcription as a search query is received, a score that is based on the received data is provided to a trained classifier, and a classifier output for the candidate transcription is received. One or more of the candidate transcriptions may be selected based on the classifier outputs.

Abstract: A computer processor determines a first span of a communication, wherein a span includes content associated with one or more dialog statements. If the content of the first span contains one or more topic change indicators which are identified by at least one detector of a learning model, the computer processor, in response, generates scores for each of the one or more indicators. The computer processor aggregates scores of the one or more indicators of the first span, which may be weighted, to produce an aggregate score. The computer processor compares the aggregate score to a threshold value, wherein the threshold value is determined during training of the learning model, and the computer processor, in response to the aggregate score crossing the threshold value, determines a topic change has occurred within the first span.

Abstract: Presented are improvements for speech recognition systems used to control devices. Features include two-stage confirmation, two-stage limited speech recognition mode, and two-stage wake-up for speech driven applications and systems. A headset computer device includes such staged confirmation operation.

Abstract: Mechanisms are provided for validating a candidate answer to an input question. A candidate answer to an input question is identified using a natural language processing of the input question and a corpus of information from which the candidate answer is identified. A validator is selected to apply to the candidate answer based on a characteristic of a correct answer for the input question. The validator is applied to the candidate answer to evaluate whether or not criteria of the validator are met by the candidate answer. Validation information is generated based the evaluation of whether or not criteria of the validator are met by the candidate answer. The validation information is stored in a validation status object associated with the input question.

Abstract: A topical search computer system identifies topics from various definitional (i.e., data) sources. The system generates a catalog of different topics from the data sources. Topics with similar names are differentiated by the system based on the context in which each topic is used. The context for a topic is represented by a context vector, which describes the co-occurrence relationships between the topic and other topics derived from the data sources. Because the system has computed a context for each topic, the system can provide improved search results responsive to user queries for information.

Abstract: A system includes a library of elements to create a model describing a waveform parameter technique. A constraint checker verifies that the created model is correct by construction by needing no verification after compilation. The constraint checker also implements a valid programmable device build according to the waveform parameter technique. A placement decision module receives the verified models produces a placement decision for placing the waveform parameter technique based on the verified model in a programmable device. A synthesis tool receives the verified model from the placement decision module and synthesizes the waveform parameter technique based on the verified model. A link/loader receives the placement decision from the placement decision module and receives the synthesized technique from the synthesis tool. The link/loader also places the waveform parameter technique in the programmable device according to the placement decision.

Abstract: Automatic trading environments with their high degree of automation have become the backbone of modern financial markets. The ability to process orders and manage risk in these systems while maintaining a low latency between participants is crucial for the safety and liquidity of these markets. The disclosed system describes a four valued Monte Carlo simulation for the stochastic modeling of risk and syntactic pattern matching techniques to facilitate the design of these systems. The system is a self-compiling, machine independent system capable of dividing, scaling and communicating multiple-asset instruments efficiently in a parallel environment. The system also allows for the integration of computerized financial heuristics on financial instruments and user interfaces for creating trading strategies to monitor and hedge risk over a trading desk for financial institutions.

Abstract: According to an embodiment, a text-to-speech device includes a receiver to receive an input text containing a peculiar expression; a normalizer to normalize the input text based on a normalization rule in which the peculiar expression, a normal expression of the peculiar expression, and an expression style of the peculiar expression are associated, to generate normalized texts; a selector to perform language processing of each normalized text, and select a normalized text based on result of the language processing; a generator generate a series of phonetic parameters representing phonetic expression of the selected normalized text; a modifier modifies a phonetic parameter in the normalized text corresponding to the peculiar expression in the input text based on a phonetic parameter modification method according to the normalization rule of the peculiar expression; and a output unit to output a phonetic sound synthesized using the series of phonetic parameters including the modified phonetic parameter.

Abstract: A computer-implemented system, method and program product generates answers to questions in an input query text string. The method includes determining, by a programmed processor unit, a lexical answer type (LAT) string associated with an input query; automatically obtaining a candidate answer string to the input query from a data corpus; mapping the query LAT string to a first type string in a structured resource; mapping the candidate answer string to a second type string in the structured resource; and determining if the first type string and the second type string are disjointed; and scoring the candidate answer string based on the determination of the types being disjointed wherein the structured resource includes a semantic database providing ontological content.

Abstract: This disclosure relates to contextual determination of related media content. A referral source identification component determines a referral source associated with a request for media content, and a determination component determines a set of context data for the request based in part on the referral source. A contextual suggestion component identifies or selects a set of additional content based in part on a subset of the context data, and suggests or recommends a subset of the additional content to a user based on a set of recommendation criteria.

Abstract: Methods and systems to determine consumer locations based on navigational voice cues are disclosed. An example method includes identifying, with a processor, a route point designated by a navigational voice cue output by a navigation system providing directions to an operator of a vehicle. The example method also includes determining, with the processor, at least one of a location or an anticipated location of an occupant of the vehicle based on the route point.

Abstract: This disclosure generally relates to a method and system for counting words spoken by a user. For example, an audio input representative of spoken speech may be received and divided into a series of windows. The windows may be analyzed to determine whether or not one or more of the windows meet one or more speech criteria. The number of times a window meets one or more speech criteria may be counted and modified by a modifier factor to determine a word count. The word count may be output in a graphical user interface.

Abstract: A dialog system that includes a dialog manager to manage a conversation between the dialog system and a user, and to associate the conversation with a complex activity, and a plan engine to execute a plan script in connection with the complex activity, the plan script including a set of atomic dialog activities and logic to control a data and sequence flow of the atomic dialog activities, the set of atomic dialog activities being sub-activities of the complex activity, the complex activity being specified via a declarative activity specification language that connects the atomic dialog activities with a process.

Abstract: A glossary management device includes a read circuit, a storage circuit, an acquisition circuit, an analysis circuit, a term matching circuit, and a registration circuit. The storage circuit has a storage area for a glossary. The acquisition circuit acquires text data of a document if reading of the document is executed by a user. The analysis circuit performs analysis of the text data acquired by the acquisition circuit to identify a language of the document and parts of speech of text segments in the text data and extracts one or more text segments from the document based on the analysis. The term matching circuit performs matching for each of the extracted text segments against a public dictionary. The registration circuit adds to the glossary, each extracted text segment that does not match any entry term in the public dictionary.

Abstract: A method for translation using a translation tree structure in a portable terminal includes inputting, by a user speaking in a first language, one or more first language words to a first portable terminal; translating the inputted first language words into one or more second language words using a database of the first portable terminal; displaying the translated second language words according to the translation tree structure; selecting one of words of the second language words displayed in the translation tree structure; and transmitting the selected second language words to a server and a second portable terminal which uses the second language.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for obtaining, for each of multiple words or sub-words, audio data corresponding to multiple users speaking the word or sub-word; training, for each of the multiple words or sub-words, a pre-computed hotword model for the word or sub-word based on the audio data for the word or sub-word; receiving a candidate hotword from a computing device; identifying one or more pre-computed hotword models that correspond to the candidate hotword; and providing the identified, pre-computed hotword models to the computing device.

Abstract: A computer-implemented system, method and program product generates answers to questions in an input query text string. The method includes determining, by a programmed processor unit, a lexical answer type (LAT) string associated with an input query; automatically obtaining a candidate answer string to the input query from a data corpus; mapping the query LAT string to a first type string in a structured resource; mapping the candidate answer string to a second type string in the structured resource; and determining if the first type string and the second type string are disjointed; and scoring the candidate answer string based on the determination of the types being disjointed wherein the structured resource includes a semantic database providing ontological content.

Abstract: A system for verifying factual assertions in natural language. Receiving a text input. Identifying, a verifiable factual statement in the text. Forming a query based on the verifiable factual assertion. Searching a corpus based on the query. Determining the veracity of verifiable factual assertion based on the search results.

Abstract: A method for accessing offline voicemail messages within a mobile messaging application may be provided. First, a voice mail message may be received and the voicemail message may be transcribed to text. Next, the voicemail message and the text transcription may be stored. The recipient may then be presented with a list of voicemail messages and the voicemail message may be retrieved in response to the recipient. The recipient may read or listen to the voicemail message or both. The recipient may also annotate the voicemail message.

Abstract: A database keyword search technique that relies on a domain based storage infrastructure is disclosed. In operation, a keyword search string is processed to generate a set of search string permutations. Each string permutation specifies a different ordering of one or more portions of the search string. A domain based search process is then executed asynchronously for each string permutation. Each execution generates a search result set that identifies rows in the database that include data relevant to the string permutation. The results in each result set are scored and ranked based in part on the similarity between the string permutation and the search string provided by the user. The rankings determine which of the results are to be presented to the user.

Abstract: An approach is provided in which a knowledge manager computes a set of unreasonable answer probabilities based upon a corresponding set of answers. Each of the set of unreasonable answer probabilities indicate a likelihood that its corresponding answer is an unreasonable answer to a question received by the knowledge manager. The knowledge manager determines that one or more unreasonable answers are included in the set of answers based upon determining that their corresponding unreasonable answer probabilities are above a threshold. In turn, the knowledge manager removes the one or more unreasonable answers from the set of answers and provides a modified set of answers to a user.

Abstract: A method of recognizing speech is provided that includes generating a decoding network that includes a primary sub-network and a classification sub-network. The primary sub-network includes a classification node corresponding to the classification sub-network. The classification sub-network corresponds to a group of uncommon words. A speech input is received and decoded by instantiating a token in the primary sub-network and passing the token through the primary network. When the token reaches the classification node, the method includes transferring the token to the classification sub-network and passing the token through the classification sub-network. When the token reaches an accept node of the classification sub-network, the method includes returning a result of the token passing through the classification sub-network to the primary sub-network. The result includes one or more words in the group of uncommon words. A string corresponding to the speech input is output that includes the one or more words.

Abstract: Building a runtime dialog is provided. An initial answer to an initial question is received from a user. A state of a conversation thread is created based on the initial answer. A next question to ask is automatically selected by a processor based on the updated state of the conversation thread. The next question is presented to the user. A next answer to the next question is received. The state of the conversation thread is updated based on the next answer. The selecting, presenting, receiving, and updating are repeated until a criterion is satisfied.

Abstract: According to one embodiment, a speech of a first language is recognized using a speech recognition dictionary to recognize the first language and a second language, and a source sentence of the first language is generated. The source sentence is translated into a second language, and a translation sentence of the second language is generated. An unknown word included in the translation sentence is detected. The unknown word is not stored in the speech recognition dictionary. A first pronunciation candidate of the unknown word is estimated, from a representation of the unknown word. A second pronunciation candidate of the unknown word is estimated from a pronunciation of an original word included in the source sentence corresponding to the unknown word. The unknown word, the first pronunciation candidate and the second pronunciation candidate, are registered into the speech recognition dictionary correspondingly.

Abstract: The agent server of the present invention includes (i) a conversation monitoring section for monitoring, based on a conversation log saved in a family message board server, comments made by members belonging to a community, (ii) a status information collecting section for collecting status information of the community which information corresponds to (i) a specific member or a combination of specific members and (ii) a specific keyword or a combination of specific keywords which have been detected based on the conversation log, and (iii) a comment making section for outputting, to the server, a comment message which conforms to an event of the community which event is indicated by status information.

Abstract: A method of extracting information from text within a natural language understanding system can include processing a text input through at least one statistical model for each of a plurality of features to be extracted from the text input. For each feature, at least one value can be determined, at least in part, using the statistical model associated with the feature. One value for each feature can be combined to create a complex information target. The complex information target can be output.

Abstract: The present specification relates to a speech recognition apparatus and method capable of accurately recognizing the speech of a user in an easy and convenient manner without the user having to operate a speech recognition start button or the like. The speech recognition apparatus according to embodiments of the present specification comprises: a camera for capturing a user image; a microphone; a control unit for detecting a preset user gesture from the user image, and, if a nonlexical word is detected from the speech signal which is input through the microphone from the point in time at which the user gesture was detected, determining the speech signal detected after the detected nonlexical word as an effective speech signal; and a speech recognition unit for recognizing the effective speech signal.

Abstract: A method and device for matching speech to text are disclosed, the method including: receiving a speech input, the mentioned speech input carrying input speech information; obtaining initial text corresponding to the input speech information, and respective pinyin of the initial text; generating at least one approximate pinyin for the initial text based on predetermined pronunciation similarity information; and from a preset mapping relationship table, obtaining additional text corresponding to the respective pinyin of the initial text or to the at least one approximate pinyin of the initial text, wherein the preset mapping relationship table includes a respective record for each word in a word database, including respective pinyin and at least one respective approximate pinyin for said each word, and a respective mapping relation between said respective pinyin, said at least one respective approximate pinyin, and said each word.

Abstract: A post-processing speech system includes a natural language-based speech recognition system that compares a spoken utterance to a natural language vocabulary that includes words used to generate a natural language speech recognition result. A master conversation module engine compares the natural language speech recognition result to domain specific words and phrases. A voting engine selects a word or a phrase from the domain specific words and phrases that is transmitted to an application control system. The application control system transmits one or more control signals that are used to control an internal or an external device or an internal or an external process.

Abstract: A speech recognition platform configured to receive an audio signal that includes speech from a user and perform automatic speech recognition (ASR) on the audio signal to identify ASR results. The platform may identify: (i) a domain of a voice command within the speech based on the ASR results and based on context information associated with the speech or the user, and (ii) an intent of the voice command. In response to identifying the intent, the platform may perform a corresponding action, such as streaming audio to the device, setting a reminder for the user, purchasing an item on behalf of the user, making a reservation for the user or launching an application for the user. The speech recognition platform, in combination with the device, may therefore facilitate efficient interactions between the user and a voice-controlled device.

Abstract: The disclosed embodiments relate to a tax-information assembly technique, which extracts tax information and associated context information from income-tax documents, where these income-tax documents are associated with an income-tax agency, and some of the income-tax documents include the same tax information in different document formats. During this technique, semantic and structural heuristics are used to identify tax phrases in the extracted tax information. Moreover, additional tax phrases in the extracted tax information are identified using a statistical identification technique. Next, relationships between the tax phrases and the additional tax phrases are determined, and the context information is used to consolidate the tax phrases and the additional tax phrases into a tax-information data structure.

Abstract: A resource provider is able to manage instance data associated with virtual compute instances running in the resource provider environment. For example, when provisioning a compute instance, the resource provider can obtain data associated with the compute instance and can store this data, for example, in a data store. The resource provider can act as a centralized repository of such data for some or all instances that are running in the resource provider environment. Entities (e.g., users or other compute instances running in the resource provider environment) can query the resource provider to perform various operations (e.g., read, modify, duplicate) on the data being managed by the resource provider for the various compute instances. Any changes to the data for a compute instance, for example, by a user, the compute instance, or a different compute instance, can be saved by the resource provider and propagated to the compute instance.

Abstract: A device control system includes an ambient sound obtainer that obtains ambient sound around the plurality of devices at a first time, an operation state obtainer that obtains the operation states of the plurality of devices at the first time from the plurality of devices through a network, and a database manager that stores the operation states at the first time obtained by the operation state obtainer and the ambient sound obtained by the ambient sound obtainer at the first time in a database while associating the operation states and the ambient sound with each other.

Abstract: Systems for improving or generating a spoken language understanding system using a multitask learning method for intent or call-type classification. The multitask learning method aims at training tasks in parallel while using a shared representation. A computing device automatically re-uses the existing labeled data from various applications, which are similar but may have different call-types, intents or intent distributions to improve the performance. An automated intent mapping algorithm operates across applications. In one aspect, active learning is employed to selectively sample the data to be re-used.

Abstract: Systems and methods for acquiring information associated with a user by using NLP techniques are disclosed. One or more phrases are classified in one or more categories at least partly on the basis of a period for which a product has been used by the user, the user's experience with the product, preferences of the user, or needs of the user by applying one or more natural language processing (NLP) techniques. The one or more phrases are extractable from an electronic publication at least partly on the basis of on a predefined set of verbs, a predefined set of domain-specific terms, and terms indicative of temporal information. One or more terms from the classified phrases are extracted, in which the one or more terms are indicative of the information about the user.

Abstract: A computer-implemented system and method for efficiently reducing transcription error during a call is provided. A group of similar questionable utterances is monitored. Each questionable utterance is assigned a transcribed value and is selected from a different call. A sample of the similar questionable utterances is selected from the group and provided to a human transcriber. A further transcribed value for each of the similar questionable utterances in the sample is received from the human transcriber. The further transcribed values for the similar questionable utterances in the sample are compared. Each of the remaining similar questionable utterances in the group are provided to the human transcriber when the compared further transcribed values of the similar questionable utterances in the sample are different.

Abstract: Described is a system and technique to supplement speech commands with gestures. A user interface may be improved by providing the ability for a user to intuitively provide speech commands with the aid of gestures. By providing gestures contemporaneously with a speech command, the user may delimit the commencement and end of a command thereby allowing the system to provide an immediate response. In addition, gestures may be detected in order to determine a source of a provided speech command, and accordingly, user specific actions may be performed based on the identity of the source.

Abstract: Methods and systems are provided for creating a calendar event using context. A natural language expression including at least one of words, terms, and phrases of text may be received at a calendar event creation module from an application. The calendar event creation module may identify one or more slots in the text of the natural language expression related to the calendar event using a first grammar module and a second grammar module. The one or more slots identified by the first grammar module and the second grammar module that indicate a calendar event may be compared to determine whether there is a match between the one or more identified slots. If a match is found, at least one calendar event using the one or more slots identified by the first grammar module and the second grammar module may be created.

Abstract: Methods, apparatus, and computer-readable media are described herein related to a user interface (UI) that can be implemented on a head-mountable device (HMD). The UI can include a voice-navigable UI. The voice-navigable UI can include a voice navigable menu that includes one or more menu items comprising an original menu item and an added command menu item. The original menu item can be associated with one or more original commands, and the added menu item can be associated with one or more added commands, including a first added command. The interface can also present a first visible menu that includes at least a portion of the voice navigable menu. Responsive to a first utterance comprising the first added command, the interface can invoke the first added command. In some embodiments, the interface can display a second visible menu, wherein the first added command can be displayed above other menu items.

Abstract: Causal topic mining can include incorporating non-text time series data with a number of articles based on a time relationship and analyzing the incorporated non-text time series data and the number of articles at a particular time to determine a causal relationship.

Abstract: Techniques for providing a prompt for real-time cognitive assistance. A method includes analyzing input from at least one environmental sensor to identify context information pertaining to a user situation, identifying a likely subsequent cognitive task of the user in the user situation based on the context information and use of a learned model, determining an action with respect to information to be suggested to the user via a corresponding prompt, wherein the determining is based on the likely subsequent cognitive task, the context information and information learned from at least one previous user situation, computing a confidence value to represent a level of certainty in the action, and providing the prompt to the user if the action has a confidence value greater than a threshold value corresponding to the action.

Abstract: Techniques for training a tokenizer (or word segmenter) are provided. In one technique, a tokenizer tokenizes a token string to identify individual tokens or words. A language model is generated based on the identified tokens or words. A vocabulary about an entity, such as a person or company, is identified. The vocabulary may be online data that refers to the entity, such as a news article or a profile page of a member of a social network. Some of the tokens in the vocabulary may be weighted higher than others. The language model accepts the weighted vocabulary as input and generates pseudo sentences. Alternatively, regular expressions are used to generate the pseudo sentences. The pseudo sentences are used to train the tokenizer.

Abstract: A method and apparatus are provided for automatically acquiring grammar fragments for recognizing and understanding fluently spoken language. Grammar fragments representing a set of syntactically and semantically similar phrases may be generated using three probability distributions: of succeeding words, of preceding words, and of associated call-types. The similarity between phrases may be measured by applying Kullback-Leibler distance to these tree probability distributions. Phrases being close in all three distances may be clustered into a grammar fragment.

Abstract: Automatically adjusting confidence scoring functionality is described for a speech recognition engine. Operation of the speech recognition system is revised so as to change an associated receiver operating characteristic (ROC) curve describing performance of the speech recognition system with respect to rates of false acceptance (FA) versus correct acceptance (CA). Then a confidence scoring functionality related to recognition reliability for a given input utterance is automatically adjusted such that where the ROC curve is better for a given operating point after revising the operation of the speech recognition system, the adjusting reflects a double gain constraint to maintain FA and CA rates at least as good as before revising operation of the speech recognition system.

Abstract: Systems and techniques for extracting data from unstructured documents are described. One such method involves assigning one or more labels to one or more nodes in a first object model of a first web page; comparing a second object model of a second web page to the first object model; if the first object model matches the second object model to a determined degree, extracting from the second web page data associated with nodes in the second object model that match labeled nodes in the first object model; and providing the extracted data for storage in a structured database in a manner associated with the labels.