Abstract: Systems and methods for automated custom training of a scoring model are disclosed herein. The method include: receiving a plurality of responses received from a plurality of students in response to providing of a prompt; identifying an evaluation model relevant to the provided prompt, which evaluation model can be a machine learning model trained to output a score relevant to at least portions of a response; generating a training indicator that provides a graphical depiction of the degree to which the identified evaluation model is trained; determining a training status of the model; receiving at least one evaluation input when the model is identified as insufficiently trained; updating training of the evaluation model based on the at least one received evaluation input; and controlling the training indicator to reflect the degree to which the evaluation model is trained subsequent to the updating of the training of the evaluation model.

Abstract: Described are techniques for predicting when data associated with a user input is likely to be selected for deletion. The system may use a trained model to assist with such predictions. The trained model can be configured based on deletions associated with a user profile. An example process can including receiving user input data corresponding to the user profile, and processing the user input data to determine a user command. Based on characteristic data of the user command, the trained model can be used to determine that data corresponding to the user command is likely to be selected for deletion. The trained model can be iteratively updated based on additional user commands, including previously received user commands to delete user input data.

Abstract: Systems and methods are presented herein for recording portions of a media asset relevant to recording criteria. A media application receives input indicating the recording criteria and identifying a first keyword. The media application accesses a data structure to identify a first node associated with the first keyword. The data structure includes the first node and a plurality of nodes connected to the first node via a plurality of paths. The media application receiving audio component data for a portion of the media asset extracts a term from the audio component data, and identifies a second node in the data structure that is associated with the extracted term. The media application calculates a path score for the portion of the media asset based on a path size in the data structure between the first node and the second node. When the score is high enough, the portion of the media asset is recorded.

Abstract: A de-coupled computing infrastructure is described that is adapted to provide domain specific contextual engines based on conversational flow. The computing infrastructure further includes, in some embodiments, a mechanism for directing conversational flow in respect of a backend natural language processing engine. The computing infrastructure is adapted to control or manage conversational flows using a plurality of natural language processing agents.

Abstract: A system and a method for locating populations of content-specific data portions. The method includes determining a current population of data portions to be searched based on at least one prioritization criterion; accessing the current population of data portions; examining at least one data portion of the current population of data portions and extracting content-specific data; comparing the content-specific data to at least one suspect criterion; determining whether the current population meets at least one population criterion by analyzing the content-specific data; determining at least one next population of data portions to be searched based on proximity to the current population; and determining the at least one next population of data portions to be searched based on the at least one prioritization criterion.

Abstract: Systems and methods are presented herein for recording portions of a media asset relevant to recording criteria. A media application receives input indicating the recording criteria and identifying a first keyword. The media application accesses a data structure to identify a first node associated with the first keyword. The data structure includes the first node and a plurality of nodes connected to the first node via a plurality of paths. The media application receiving audio component data for a portion of the media asset extracts a term from the audio component data, and identifies a second node in the data structure that is associated with the extracted term. The media application calculates a path score for the portion of the media asset based on a path size in the data structure between the first node and the second node. When the score is high enough, the portion of the media asset is recorded.

Abstract: A method includes receiving a search query and identifying an entity record in an entity data store using the search query. The entity data store includes a plurality of entity records. Each of the plurality of entity records includes an entity name, an entity type, and entity information. The entity type indicates a category in which the entity name and the entity information belongs. The method further includes generating a reformulated query by inserting one or more terms from the identified entity record into the search query. Additionally, the method includes performing a search for software applications using the reformulated query and generating a list of software applications identified during the search.

Abstract: A chatbot answer database can be generated from a question and answer document. A question/answer processor can receive a document that includes questions and answers. A set of entities can be determined from the answers in the document. A set of triplets can be constructed from the answers, and stored in a chatbot database. For each answer, the question/answer processor can determine text for slots of the triplet based on the entities extracted from the answer. The triplet can be stored as a tree structure in a database of tree structures associated with the document. A chatbot can receive questions, and provide answers based on the database of tree structures associated with the document.

Abstract: A processor causes a robot to execute any one of a first, second, and third action as an initial action. The initial action is executed for communication with a target person according to a captured image and a captured sound. When a sound is acquired by a microphone after execution of the current action, the processor causes the robot to execute an action one-level higher than the current action. The current action includes the initial action. When the sound is not acquired, the processor determines whether a time elapsed from the execution of the current action is shorter than a threshold. When the time is shorter than the threshold, the processor causes the robot to continue the current action. When the time is equal to or longer than the threshold, the processor causes the robot to execute an action one-level lower than the current action.

Abstract: A dialog manager receives text data corresponding to a dialog with a user. Entities represented in the text data are identified. Context data relating to the dialog is maintained, which may include prior dialog, prior API calls, user profile information, or other data. Using the text data and the context data, an N-best list of one or more dialog models is selected to process the text data. After processing the text data, the outputs of the N-best models are ranked and a top-scoring output is selected. The top-scoring output may be an API call and/or an audio prompt.

Abstract: A system and method are presented for a multiclass approach for confidence modeling in automatic speech recognition systems. A confidence model may be trained offline using supervised learning. A decoding module is utilized within the system that generates features for audio files in audio data. The features are used to generate a hypothesized segment of speech which is compared to a known segment of speech using edit distances. Comparisons are labeled from one of a plurality of output classes. The labels correspond to the degree to which speech is converted to text correctly or not. The trained confidence models can be applied in a variety of systems, including interactive voice response systems, keyword spotters, and open-ended dialog systems.

Abstract: In speech processing systems personalization is added in the Natural Language Understanding (NLU) processor by incorporating external knowledge sources of user information to improve entity recognition performance of the speech processing system. Personalization in the NLU is effected by incorporating one or more dictionaries of entries, or gazetteers, with information personal to a respective user, that provide the user's information to permit disambiguation of semantic interpretation for input utterances to improve quality of speech processing results.

Abstract: Aspects of the present disclosure describe techniques for identifying and recovering out-of-vocabulary words in transcripts of a voice data recording using word recognition models and word sub-unit recognition models. An example method generally includes receiving a voice data recording for transcription into a textual representation of the voice data recording. The voice data recording is transcribed into the textual representation using a word recognition model. An unknown word is identified in the textual representation, and the unknown word is reconstructed based on recognition of sub-units of the unknown word generated by a sub-unit recognition model. The textual representation of the voice data recording is modified by replacing the unknown word with the reconstruction of the unknown word, and the modified textual representation is output.

Abstract: An approach for enhancing user communication with a chatbot is provided. In an embodiment, a communication from a user using a chatbot application is received. One or more intents related to the user communication are identified, with each intent having a respective confidence score. Further, any number of attributes related to the user communication are extracted. If multiple viable intents are found (e.g., intents having a confidence score exceeding a predefined threshold), then relationships between the extracted attributes are derived to establish a context. Based on the context, a disambiguation question is generated to clarify an actual intent of the user.

Abstract: Systems and processes for operating an intelligent automated assistant are provided. In one example process, a speech input is received from a user. In response to determining that the speech input corresponds to a user intent of obtaining information associated with a user experience of the user, one or more parameters referencing a user experience of the user are identified. Metadata associated with the referenced user experience is obtained from an experiential data structure. Based on the metadata, one or more media items associated with the referenced are retrieved based on the metadata. The one or more media items associated with the referenced user experience are output together.

Abstract: Systems and methods for identifying content corresponding to a language are provided. Language spoken by a first user based on verbal input received from the first user is automatically determined with voice recognition circuitry. A database of content sources is cross-referenced to identify a content source associated with a language field value that corresponds to the determined language spoken by the first user. The language field in the database identifies the language that the associated content source transmits content to a plurality of users. A representation of the identified content source is generated for display to the first user.

Abstract: A character input device includes a speech input unit that receives an input speech, a controller that detects a substitution voice included in the input speech received by the speech input unit, and generates a predictive suggestion including a part replaced with the substitution voice and corresponding to the input speech, and a suggestion output unit that outputs the predictive suggestion.

Abstract: A system comprises an apparatus having a first voice I/O device; and a voice interface apparatus having a second voice I/O device, and connected to the apparatus by audio connection via short-range wireless communication, wherein the apparatus includes a voice I/O unit that performs voice input and output by using the first voice I/O device or the second voice I/O device; an interaction unit that performs voice interaction with a user; and a process unit that performs a process other than the voice interaction, by using the voice I/O, and the voice I/O unit switches a device used for the voice input and output to the first voice input/output device in a case where the process unit is brought into a first state in which the voice input and output is required when the voice interaction with the user is performed by using the second voice I/O device.

Abstract: Systems and methods are presented herein for recording portions of a media asset relevant to recording criteria. A media application receives input indicating the recording criteria and identifying a first keyword. The media application accesses a data structure to identify a first node associated with the first keyword. The data structure includes the first node and a plurality of nodes connected to the first node via a plurality of paths. The media application receiving audio component data for a portion of the media asset extracts a term from the audio component data, and identifies a second node in the data structure that is associated with the extracted term. The media application calculates a path score for the portion of the media asset based on a path size in the data structure between the first node and the second node. When the score is high enough, the portion of the media asset is recorded.

Abstract: An example electronic device is disclosed. The electronic device includes a communication circuit for communicating with a server storing information on a plurality of clips and storing keywords by the plurality of clips; an output device; an input device; and a processor which, when a voice uttered by a user is received via the input device, transmits a clip request signal to the server, based on a keyword included in the received uttered voice and information on content outputted by the output device, and outputs a short clip via the output device, based on information on the short clip received from the server in response to the request signal.

Abstract: A terminal includes a memory configured to store voice data and a processor configured to measure reliability of learnable data stored in the memory, to classify the learnable data into learning data or adaptive data according to the measured reliability, to generate a learning model by performing unsupervised learning with respect to the learning data, to generate an adaptive model using the adaptive data, and to evaluate recognition performance of each of the learning model and the adaptive model.

Abstract: A method and electronic device that enables quick presentation of user briefs on the electronic device includes: receiving, at the electronic device, an informational brief (IB) request (IBR) input that includes an identifier of a topic and a trigger that causes the electronic device to open an IB content screen that temporarily presents specific information corresponding to the topic. In response to receipt of the IB input, the method further includes: retrieving the IB content screen, with content that includes at least one of the specific information; presenting the IB content screen on the electronic device; monitoring an elapsed presentation time for the IB content screen on the electronic device; comparing the elapsed presentation time against a time limit allocated for presenting the IB content screen; and in response to expiration of the time limit, closing the IB content screen to return the device to its previous operating state.

Abstract: A non-transitory computer-readable storage medium storing a program that causes a computer to execute a process, the computer controlling an utterance of a robot, the process including detecting an utterance of a person by using a microphone, obtaining, in response to the detecting, pieces of response information for the utterance of the person based on first information indicating a content of the utterance of the person, obtaining second information relating to at least one of the person and a motion of the person other than the utterance of the person, selecting specified response information among the pieces of response information based on the second information, and transmitting, to the robot, an instruction that causes the robot to execute a response in accordance with the specified response information.

Abstract: Generating, updating, and using a knowledge graph. Concepts in a knowledge graph can have relations to one another. These relations may be expressed as confidence values. A training data set may be split into two portions, with the first portion used to update confidence values for existing relations between concept pairs, using the knowledge graph. These confidence values can be used, together with the second portion used to update confidence values for known phrases that express known relations. These confidence values, in turn, can be used, together with the first portion, to increase the accuracy of the original confidence scores with respect to existing relations. The process may be iteratively employed, with each iteration increasing the accuracy of confidence scores.

Abstract: Identifying data quality along a data flow. A method includes identifying quality metadata for two or more datasets. The quality metadata defines one or more of quality of a data source, accuracy of a dataset, completeness of a dataset, freshness of a dataset, or relevance of a dataset. At least some of the metadata is based on results of operations along a data flow. Based on the metadata, the method includes creating one or more quality indexes for the datasets. The one or more quality indexes include a characterization of quality of two or more datasets.

Abstract: The present disclosure provides an interface intelligent interaction control method, apparatus and system, and a storage medium, wherein the method comprises: receiving user-input speech information, and obtaining a speech recognition result; determining scenario elements associated with the speech recognition result; generating an entry corresponding to each scenario element and sending the speech recognition result and the entry to a cloud server; receiving an entry which is best matched with the speech recognition result, returned by the cloud server and selected from the received entries; performing an interface operation corresponding to the best-matched entry. The solution of the present disclosure can be applied to improve flexibility and accuracy of the speech control.

Abstract: A text generating device includes an acquirer that acquires the utterance of the user relative to content provided in real time, a retriever that retrieves from a microblog server data relating to the content, a sentence generator that generates a sentence relating to the content and the acquired utterance of the user, based on the data retrieved by the retriever 140, and an voice outputter and/or a display for using the generated sentence to reply to the user.

Abstract: A network resource access system for providing access by a user to network resources over a communications network, the system comprising: a resource registry including stored resource records associated with each of the network resources and a stored user profile containing a list of network resources such that the network resources have a ranking relative to each other based at least in part on user behavior with respect to usage of each of the network resources, the user profile associated with the user such that the list of network resources contains the network resources previously accessed by the user; and a resource service for receiving an access query from a network terminal identifying the user and associated with submission of application data for processing by a selected network resource from the list, the resource service further configured for accessing the user profile to identify a suggested network resource from the list in view of the relative ranking and for sending identification of the su

Abstract: Methods, systems, devices, and media for creating a plan through multimodal search inputs are provided. A first search request comprises a first input received via a first input mode and a second input received via a different second input mode. The second input identifies a geographic area. First search results are displayed based on the first search request and corresponding to the geographic area. Each of the first search results is associated with a geographic location. A selection of one of the first search results is received and added to a plan. A second search request is received after the selection, and second search results are displayed in response to the second search request. The second search results are based on the second search request and correspond to the geographic location of the selected one of the first search results.

Abstract: Techniques are disclosed for client-side analysis of audio samples to identify one or more characteristics associated with captured audio. The client-side analysis may then allow a user device, e.g., a smart phone, laptop computer, in-car infotainment system, and so on, to provide the one or more identified characteristics as configuration data to a voice recognition service at or shortly after connection with the same. In turn, the voice recognition service may load one or more recognition components, e.g., language models and/or application modules/engines, based on the received configuration data. Thus, latency may be reduced based on the voice recognition engine having “hints” that allow components to be loaded without necessarily having to process audio samples first. The reduction of latency may reduce processing time relative to other approaches to voice recognitions systems that exclusively perform server-side context recognition/classification.

Abstract: A mobile terminal including a touch screen; a microphone configured to receive voice information from a user; and a controller configured to analyze the voice information using a voice recognition algorithm, extract a term predicted to be unfamiliar to the user from the analyzed voice information based on a pre-stored knowledge database, search for information on the extracted term based on a context of the analyzed voice information, and display the searched information to the touch screen.

Abstract: Systems and methods for detecting, classifying, and correcting acoustic (waveform) events are provided. In one example embodiment, a computer-implemented method includes obtaining, by a computing system, audio data from a source. The method includes accessing, by the computing system, data indicative of a machine-learned acoustic detection model. The method includes inputting, by the computing system, the audio data from the source into the machine-learned acoustic detection model. The method includes obtaining, by the computing system, an output from the machine-learned acoustic detection model. The output is indicative of an acoustic event associated with the source. The method includes providing, by the computing system, data indicative of a notification to a user device. The notification indicates the acoustic event and response(s) for selection by a user.

Abstract: A method, computer program product, and computing system for receiving a presentation file including one or more audio portions and one or more textual portions. An audio transcript of the one or more audio portions of the presentation file may be generated. A textual transcript of the one or more textual portions of the presentation file may be generated. One or more rich portions of the presentation file may be determined based upon, at least in part, a comparison of the audio transcript and the textual transcript. At least the one or more rich portions of the presentation file may be presented.

Abstract: Systems and methods for identifying content corresponding to a language are provided. Language spoken by a first user based on verbal input received from the first user is automatically determined with voice recognition circuitry. A database of content sources is cross-referenced to identify a content source associated with a language field value that corresponds to the determined language spoken by the first user. The language field in the database identifies the language that the associated content source transmits content to a plurality of users. A representation of the identified content source is generated for display to the first user.

Abstract: An optimized fact checking system analyzes and determines the factual accuracy of information and/or characterizes the information by comparing the information with source information. The optimized fact checking system automatically monitors information, processes the information, fact checks the information in an optimized manner and/or provides a status of the information. In some embodiments, the optimized fact checking system generates, aggregates, and/or summarizes content.

Abstract: Systems and methods are presented herein for recording portions of a media asset relevant to recording criteria. A media application receives input indicating the recording criteria and identifying a first keyword. The media application accesses a data structure to identify a first node associated with the first keyword. The data structure includes the first node and a plurality of nodes connected to the first node via a plurality of paths. The media application receiving audio component data for a portion of the media asset extracts a term from the audio component data, and identifies a second node in the data structure that is associated with the extracted term. The media application calculates a path score for the portion of the media asset based on a path size in the data structure between the first node and the second node. When the score is high enough, the portion of the media asset is recorded.

Abstract: A computer-implemented method includes receiving at a computer server system, from a computing device that is remote from the server system, a string of text that comprises a search query. The method also includes identifying one or more search results that are responsive to the search query, parsing a document that is a target of one of the one or more results, identifying geographical address information from the parsing, generating a specific geographical indicator corresponding to the one search result, and transmitting for use by the computing device, data for automatically generating a navigational application having a destination at the specific geographical indicator.

Abstract: The present disclosure provides a method and a device for processing a speech based on artificial intelligence. The method includes: grading a current frame included in a speech packet to be decoded by using an acoustic model to obtain a grading result; identifying whether the current frame is a quasi-silent frame according the grading result; and skipping the current frame and not decoding the current frame if the current frame is the quasi-silent frame. In the present disclosure, before the current frame included in the speech pocket to be decoded is decoded, it is identified whether to decode the current frame according to the grading result obtained with the acoustic model. When there is no need to decode the current frame, the current frame is skipped. Thus, a redundancy decoding may be avoided, a speed of decoding is improved and recognition of the speech packet to be decoded is expedited.

Abstract: A computer-implemented method includes identifying a first set of utterances from a plurality of utterances. The plurality of utterances is associated with a conversation and transmitted via a plurality of audio signals. The computer-implemented method further includes mining the first set of utterances for a first context. The computer-implemented method further includes determining that the first context associated with the first set of utterances is not relevant to a second context associated with the conversation. The computer-implemented method further includes dynamically muting, for at least a first period of time, a first audio signal in the plurality of audio signals corresponding to the first set of utterances. A corresponding computer system and computer program product are also disclosed.

Abstract: Systems and processes for operating an automated assistant are disclosed. In one example process, an electronic device provides an audio output via a speaker of the electronic device. While providing the audio output, the electronic device receives, via a microphone of the electronic device, a natural language speech input. The electronic device derives a representation of user intent based on the natural language speech input and the audio output, identifies a task based on the derived user intent; and performs the identified task.

Abstract: A computer-implemented method includes identifying a first set of utterances from a plurality of utterances. The plurality of utterances is associated with a conversation and transmitted via a plurality of audio signals. The computer-implemented method further includes mining the first set of utterances for a first context. The computer-implemented method further includes determining that the first context associated with the first set of utterances is not relevant to a second context associated with the conversation. The computer-implemented method further includes dynamically muting, for at least a first period of time, a first audio signal in the plurality of audio signals corresponding to the first set of utterances. A corresponding computer system and computer program product are also disclosed.

Abstract: Disclosed are embodiments of method and system to predict health condition of a human subject. The method comprises receiving historical human-subject related data including records corresponding to multiple data views. The method estimates one or more latent variables based on: a first value indicative of count of records in a cluster, a second value indicative of count of records, and a third value indicative of a parameter utilizable to predict a fourth value. The fourth value corresponds to selection probability of a D-vine pair copula family, of a D-vine mixture model, utilizable to model a cluster. The method generates the D-vine mixture model based on the estimated one or more latent variables. The method further comprises receiving multi-view data of a second human subject and predicting health condition of the second human subject based on the multi-view data using a classifier trained based on the estimated latent variables.

Abstract: A method is disclosed for triggering upon signal events occurring in frequency domain signals. The method includes repeatedly sampling a time-varying signal and generating a plurality of digital frequency domain spectrums based on the samples of the time-varying signal. A frequency domain bitmap for the time-varying signal is repeatedly updated via application of the digital frequency domain spectrums. The method further includes selecting a portion of the frequency domain bitmap, determining a signal occupancy in the selected portion, and triggering a capture of the time-varying signal based on and in response to the occupancy determination for the selected portion of the bitmap.

Abstract: Disclosed are embodiments of methods and systems for predicting a health condition of a first human subject. The method comprises receiving a measure of one or more physiological parameters associated with the first human subject. The method estimates one or more latent variables based on a first count indicative of a number of the plurality of d-vines, a second count indicative of a number of the one or more records, a first value that is representative of a number of the one or more records clustered into a d-vine from the plurality of d-vines, and a second value that is representative of a parameter utilizable to predict a third value. The method generates the plurality of d-vines based on the estimated one or more latent variables. The method predicts health condition of the first human subject by utilizing a trained classifier based on the estimated one or more latent variables.

Abstract: Disclosed are embodiments of methods and systems for predicting a health condition of a first human subject. The method comprises extracting a historical data including physiological parameters of second human subjects. Thereafter, a first distribution of a first physiological parameter is determined based on a marginal cumulative distribution of a rank transformed historical data. Further, a second distribution of a second physiological parameter is determined based on the first distribution and a first conditional cumulative distribution of the rank transformed historical data. Further, a latent variable is determined based on the first and the second distributions. Thereafter, one or more parameters of at least one bivariate distribution, corresponding to a D-vine copula, are estimated based on the latent variable. Further, a classifier is trained based on the D-vine copula. The classifier is utilizable to predict the health condition of the first human subject based on his/her physiological parameters.

Abstract: According to one embodiment, an apparatus includes a storage unit, a first acquisition unit, a second acquisition unit, an analyzer, and a recognition unit. The storage unit stores first situation information about a situation assumed in advance, a first representation representing a meaning of a sentence assumed, intention information representing an intention to be estimated, and a first value representing a degree of application of the first representation to the first situation information and the intention information. The first acquisition unit acquires a natural sentence. The second acquisition unit acquires second situation information about a situation when acquiring the natural sentence. The analyzer analyzes the natural sentence and generates a second representation representing a meaning of the natural sentence. The recognition unit obtains an estimated value based on the first value associated with the first situation information and the first representation.

Abstract: Offer listings can be classified as accessory offers or product offers using a classification operation performed on a corpus of offers. Data from the classification operation can be used to classify received queries as either product or accessory, and to classify results as products or accessories for purposes of presenting a relevant list of results to a user.

Abstract: An embodiment of a speech endpoint detector apparatus may include a speech detector to detect a presence of speech in an electronic speech signal, a pause duration measurer communicatively coupled to the speech detector to measure a duration of a pause following a period of detected speech, an end of utterance detector communicatively coupled to the pause duration measurer to detect if the pause measured following the period of detected speech is greater than a pause threshold corresponding to an end of an utterance, and a pause threshold adjuster to adaptively adjust the pause threshold corresponding to an end of an utterance based on stored pause information. Other embodiments are disclosed and claimed.

Abstract: An in-vehicle terminal sends a spoken voice as a voice signal to a relay server, and the relay server includes a voice recognition unit which converts the received voice signal into a string, and a control unit which searches for information stored in a main database or a temporary storage database by using the string and sends a search result to the in-vehicle terminal, and, upon searching for information stored in the main database, stores the search result in the temporary storage database. Upon receiving a voice signal, when the search result is stored in the temporary storage database, the control unit searches for information stored in the temporary storage database by using the string converted from the received voice signal, and, when the search result is not stored in the temporary storage database, the control unit searches for information stored in the main database by using the string.

Abstract: Example medical device data platforms are disclosed herein. In an example, the platform may include at least one integration device to access information originating from a plurality of implantable medical devices manufactured by a plurality of manufacturers and implanted in a plurality of patients. The system may also include an information processor to process the accessed information to generate at least one of patient-oriented information and provider-oriented information. The system may also include at least one communication device providing at least one of a patient portal and a provider portal to provide the patient-oriented information and the provider-oriented information, respectively.