Abstract: A method, an apparatus, a device, a storage medium and a program product of performing a text matching are provided, which relate to a field of a computer technology, and in particular to natural language processing and deep learning technologies. The method includes: determining a word set and a plurality of semantic units from a text set, the word set is associated with a first predetermined attribute, and the text set contains a plurality of first texts indicating an object information and a plurality of second texts indicating an object demand information; generating a graph; and generating a final feature representation associated with the text set and the word set based on the graph and a graph convolution model, so as to perform the text matching.

Abstract: The present application discloses a voice control method for applets and a device, and a storage medium, relating to a voice technology in computer technologies, specific implementations being: a voice control instruction for a target applet in an intelligent voice device from a user is acquired through a voice collecting device; voice recognition and intention analysis are performed by a voice interaction system on the voice control instruction to acquire intention information, and the intention information is transmitted to a target applet; the intention information is received and converted into a control instruction executable by a thread of the target applet by the target applet, and the control instruction is executed by the thread of the target applet.

Abstract: An information providing method includes: generating first information indicating that a friendly gathering is occurring in a home when (i) a threshold amount of time or longer has elapsed from a start time of food preparation by a user and (ii) the volume of sound in a dining space is a first threshold volume or greater; obtaining, from a second information processing apparatus connected to a first information processing apparatus, information indicating first request content over a network; and when content of the first information is included in the first request content, outputting, to the second information processing apparatus, second information including information for identifying the user or the home, using the first information generated.

Abstract: Systems, devices, and methods are disclosed in relation to a system for natural language based message categorization designed to identify text from a particular topic from a potentially inexhaustible set of potential topics. In one of many possible implementations, a vector space model is first used to translate text into a vector representation. This vector is used to determine if the text can be recreated by swapping words and phrases from a training corpus of documents. This is done by determining if the vector is within the conical span of the vector representations of the text in the training corpus of documents. Span composition is evaluated by a two vector boolean comparison, enabling great computational complexity and short-circuiting enabling fast real-time topic determination.

Abstract: An embodiment for identifying and replacing logically neutral phrases in natural language queries may include receiving a natural language query. The embodiment may also identify one or more logically neutral or non-logically neutral anchors in the natural language query. The embodiment may also identify boundaries containing one or more logically neutral phrases. The embodiment may further include detecting semantic and logical relations between verbal phrases and functional language between and adjacent to the one or more logically neutral and non-logically neutral anchors to reintroduce non-logically neutral language back into a non-logically neutral portion of the natural language query. The embodiment may also include generating a modified natural language query by automatically removing the boundaries and optionally replacing the one or more logically neutral phrases in the natural language query.

Abstract: The present embodiments relate to on demand generation of curated audio content. Responsive to initiation of an electronic communication by a client, audio relating to a client speaking to the operator can be processed to derive a series of insights relating to features of the audio. Previously recorded audio relating to the operator can be combined with a text-based script to generate audio content. The audio content can be modified using the series of derived insights. The modified audio content can include a series of disclaimers that are played back to the client. Generation of the modified audio content can be dynamically generated responsive to detecting a trigger that allows for on-demand generation of the modified audio content.

Abstract: Techniques for performing centralized unsuperivised learning in a multi-domain system are described. A user may request labeled data for an ML task, where the request includes a prompt for obtaining relevant explicit user feedback. The system may use the prompt to collect explicit user feedback for relevant runtime user inputs. After a duration of time (in the user's request for labeled data) has elapsed, the system determines whether collected user feedback indicates processing of the user input was defective and, if so, determines a cause of the defective processing. The system then uses one or more label generators to generate labeled data using the collected user feedback, whether the processing was defective, and the determined defect cause.

Abstract: A method and device for compressing audio signals forming, over time, a succession of sample frames, in each of N channels of an ambisonic representation of order higher than 0. The method includes: forming, based on the channels and for a current frame, a matrix of inter-channel covariance, and searching for eigenvectors of the covariance matrix with a view to obtaining a matrix of eigenvectors; testing the matrix of eigenvectors to verify that it represents a rotation in an N-dimensional space, and if not, correcting the matrix of eigenvectors until a rotation matrix is obtained, for the current frame; and applying the rotation matrix to the signals of the N channels before separate-channel encoding of the signals.

Abstract: Phrase alternative data structures are generated from the lattice output of an audio input to Automatic Speech Recognition (ASR) system. A user interface is supported for users to view phrase alternatives to selected portions of an audio transcript of the audio input, search the transcript based on query phrases, or edit the transcript based on phrase alternatives.

Abstract: Systems, devices, and methods of the present invention involve discourse trees. In an example, a method involves generating a discourse tree. The method includes identifying, from the discourse tree, a central entity that is associated with a rhetorical relation of type elaboration and corresponds to a topic node that identifies a central entity of the text. The method includes determining a subset of elementary discourse units of the discourse tree that are associated with the central entity. The method includes forming generalized phrases from the subset of elementary discourse units. The method includes forming tuples from the generalized phrases, where a tuple is an ordered set of words in normal form. The method involves responsive to successfully converting an elementary discourse unit associated with an identified tuple into a logical representation, updating the ontology with an entity from the identified tuple.

Abstract: Described herein are systems and methods to enable generation of high-quality summaries of documents that have questions and answers. To help summarize such documents, parsing methods are disclosed that account for different document formats. Methods are disclosed to anonymize personal and identifying information present in the documents. An ontology is defined to categorize dialog acts for the questions and answers present. Classifiers are disclosed based on this ontology. Methods are also disclosed to transform a question-answer pair to a canonical form. Based on the dialog acts for the question and answer combination, transformation methods were developed that build upon each of traditional NLP, and deep learning, techniques.

Abstract: A neural network training method is provided. The method includes obtaining an audio data stream, performing, for different audio data of each time frame in the audio data stream, feature extraction in each layer of a neural network, to obtain a depth feature outputted by a corresponding time frame, fusing, for a given label in labeling data, an inter-class confusion measurement index and an intra-class distance penalty value relative to the given label in a set loss function for the audio data stream through the depth feature, and updating a parameter in the neural network by using a loss function value obtained through fusion.

Abstract: A conversation engine performs conversations with users using chatbots customized for performing a set of tasks that can be performed using an online system. The conversation engine loads a chatbot configuration that specifies the behavior of a chatbot including the tasks that can be performed by the chatbot, the types of entities relevant to each task, and so on. The conversation may be voice based and use natural language. The conversation engine may load different chatbot configurations to implement different chatbots. The conversation engine receives a conversation engine configuration that specifies the behavior of the conversation engine across chatbots. The system may be a multi-tenant system that allows customization of the chatbots for each tenant.

Abstract: An image scanning device may comprise a microphone for receiving a voice, an image scanning section for scanning an image, and a control section for instructing the image scanning section to start scanning after a delay time elapses after the voice. The image scanning device may be able to scan a copy sheet, without being affected by an air flow caused by a voice when instructed to scan the copy sheet with a voice instruction.

Abstract: A computer-implemented method, system and computer program product for improving understandability of text by a user. A final word vector for each word in a sentence of a document is computed, such as by averaging a first word vector and a second word vector for that word. Furthermore, elements of a user portrait are vectorized. A distance is then computed between a vector for each word in the sentence and a vectorized element in the user's portrait which is summed to form an evaluation result for the element. An evaluation result is also formed for every other element in the user's portrait by performing such a computation step. A “final evaluation result” is then generated corresponding to the evaluation results for every element in the user's portrait. The document is then transformed in response to the final evaluation result indicating a lack of understanding of the sentence by the user.

Abstract: Systems and methods are described herein for inserting dialogue into query responses by generating and using dialogue metadata in conjunction with response templates. Metadata for each portion of dialogue of a plurality of portions of dialogue from a number of content items is stored, including information regarding the source content item, a transcript of the dialogue, and grammatical information. Upon receiving a query related to a content item, a type of response is first determined. Based on the type of response, and using the dialogue metadata, a portion of dialogue is identified for insertion into the response. The identified portion of dialogue is retrieved and inserted at an appropriate position within the response. The response is then generated for output.

Abstract: A method for estimating response relevance with respect to a received query includes receiving a set of user feedback items, a set of historical feedback data, and a set of context data, creating a user profile model according to the set of historical feedback data, wherein the user profile model indicates a weighting attribute based on the set of historical feedback data, weighting the set of user feedback items according to the created user profile model, creating a response relevance estimation model based on the weighted set of user feedback items, the received set of context data, and the received set of historical feedback data, and ranking one or more responses according to the created response relevance estimation model. The method may further include adjusting the user profile model and the response relevance estimation model responsive to receiving additional data.

Abstract: Methods for analyzing elements of a first text collection, and devices and methods for searching for information in a set of digital data, in particular in a computer network.

Abstract: In a method for improving speech analysis between devices, a processor receives a speech input comprising audio from a speech recognition platform. A processor segments the speech input into input vectors. A processor maps the input vectors to a profile. A processor calculates affinity coefficients between each input vector and the profile. A processor aggregates the input vectors and affinity coefficients in a user profile. A processor implements the user profile in a speech recognition program.

Abstract: In general, according to one embodiment, a named-entity extraction apparatus according to an embodiment includes: an unit extracting, using an extraction dictionary, named entities and relations from extraction document data; a unit designating character strings corresponding to the named entities extracted among character strings in learning document data; a generator generating, by applying a relation extraction rule, a learning document in which relations between named entities belonging to the categories of the relation extraction rule among the named entities designated; and a unit learning the extraction dictionary based on the learning document.