31 results about "Verbal expression" patented technology

An enhanced system for speech interpretation is provided. The system may include receiving a user verbalization and generating one or more preliminary interpretations of the verbalization by identifying one or more phonemes in the verbalization. An acoustic grammar may be used to map the phonemes to syllables or words, and the acoustic grammar may include one or more linking elements to reduce a search space associated with the grammar. The preliminary interpretations may be subject to various post-processing techniques to sharpen accuracy of the preliminary interpretation. A heuristic model may assign weights to various parameters based on a context, a user profile, or other domain knowledge. A probable interpretation may be identified based on a confidence score for each of a set of candidate interpretations generated by the heuristic model. The model may be augmented or updated based on various information associated with the interpretation of the verbalization.

The systems and techniques disclosed here feature a device including an audio analyzer connected to microphone(s) monitoring a conversation between one or more individuals. The audio analyzer can identify auditory expressions from audio captured by the microphone, where the auditory expressions can include linguistic expressions and nonlinguistic expressions. The audio analyzer can analyze the linguistic expressions to identify a topic and a time period when the topic was discussed. A video analyzer can be connected to camera(s) monitoring the conversation. The video analyzer can identify non-auditory expressions by the individuals from video captured by the camera. A sentiment analyzer can analyze auditory expressions during the time period when the topic was discussed and non-auditory expressions during the time period when the topic was discussed to determine sentiments to the topic. A processor can analyze the determined sentiments to the topic to determine a group sentiment to the topic.

An information analysis device includes a correlation value calculation unit which specifies the number of appearances of links between one linguistic expression and other linguistic expression and an appearance time of each link based on link information. The correlation value calculation unit calculates a correlation value between the one linguistic expression and the other linguistic expression according to a degree that the link continuously appears by using the specified number of appearances of the link and the appearance time of each link.

The invention provides a training method and device of a pre-training language model, computer equipment and a medium. The method comprises the following steps: obtaining a training data combination for expressing the same semantics by using different languages; inputting the training data combination into a pre-training language model with different language understanding capabilities to enable the pre-training language model to pre-train the training data combination to obtain an output data combination corresponding to the training data combination, and calculating a loss value of the training data combination according to the output data combination; and updating model parameters of the pre-training language model by using the loss value so as to improve the similarity of the output data combination. According to the method, the downstream task processing performance of the natural language processing model in a low-resource language scene is improved, and the type range of downstream tasks applicable to the pre-training language model in a multi-language scene is widened.

An information analysis device (1) uses a plurality of linguistic expressions as an analysis target, includes a link information generating unit (3) and a correlation value calculation unit (4). The link information generating unit (3) extracts time information included in each of a plurality of electronic documents including at least any one of the plurality of linguistic expressions and a relationship between the electronic documents in the plurality of electronic documents from the plurality of electronic documents, detects a link between one linguistic expression and another linguistic expression in the plurality of linguistic expressions and an appearance time of the link based on the extracted time information and the relationship between the electronic documents, and generates link information specifying the extracted link and the appearance time of the link. The correlation value calculation unit (4) specifies the number of appearances of links between the one linguistic expression and the other linguistic expression and an appearance time of each link based on the link information, and calculates a correlation value between the one linguistic expression and the other linguistic expression according to a degree that the link continuously appears by using the specified number of appearances of the link and the appearance time of each link.

The invention discloses a language training auxiliary system and method based on the residual memory of senile dementia, which establishes a personalized patient database that may be well known in thelong-term memory of AD patients and provides five training modules of assisting naming, assisting memorizing poems and patter, asking questions and answering, guiding singing and filling in the blanks. The AD patient personalized database is used as the input of the intelligent training scheduling module to generate training content metadata. The training content metadata is processed by the above five training modules to generate specific training content details, and interactive training is carried out with AD patients through the interactive module. The invention uses the remaining memoryto regain the patter, poems, songs and the like preserved in long-term memory and with good connection function, so as to carry out language expression training on patients and train language expression ability without increasing language work burden. Thus, the mental pressure and nervousness caused by the language barrier are reduced.

The invention is suitable for the technical field of voice recognition, and provides a punctuation mark adding method and system, a mobile terminal and a storage medium, and the method comprises the steps: obtaining to-be-recognized voice for voice recognition, so as to obtain a target text; extracting and labeling feature words from the target text, and matching labeling results of the feature words with language expression habits; when the labeling result is successfully matched with the language expression habit, judging a correction condition of the extracted target text, performing punctuation mark correction on the target text according to a judgment result, and outputting the target text; and when the labeling result is not successfully matched with the language expression habit, adding punctuation marks to the target text according to the language expression habit, and outputting the target text. According to the method, the punctuation marks of the corresponding text are subjected to auxiliary correction on the basis of pause detection, the marking result of the feature words and language expression habits, so that the phenomenon of low accuracy caused by adding the punctuation marks only according to a 3-gram mode is prevented.

A processor-implemented machine translation method includes translating a source sentence expressed in a source language into a target language to determine a target sentence, determining, based on a reliability of the target sentence, whether the target sentence is appropriate as a translation result of the source sentence, and re-determining a target sentence corresponding to the source sentence in response to a result of the determining being that the target sentence is inappropriate as the translation result.

The invention belongs to the technical field of face recognition and emotion calculation technology, and specifically relates to a method for calculating an emotion index based on human face action unit detection. The specific steps include: establishing a data set to be trained, forming a database of human face action units; Unit action intensity detection; calculation of emotional index; the invention realizes the emotional process expressed through facial expressions, can convey detailed information more accurately and objectively, and can make up for the profound information that cannot be described and conveyed by language, especially for language barriers and Situational applications that cannot normally obtain language information provide a very feasible solution.

Disclosed herein are systems, methods and computer readable storage media for determining tags or labels from natural language expressions expressing a preference or choice, determining attributes from natural language expressions and other data, and predicting preferences from natural language expressions and other data.

Embodiments of the invention provide a semantic comprehension method and apparatus, electronic equipment and a storage medium. The method comprises the steps of determining a to-be-comprehended text;inputting the text to be understood into a semantic understanding model to obtain a semantic understanding result output by the semantic understanding model, wherein the semantic comprehension model is obtained by training based on a sample text, a sample semantic comprehension result corresponding to the sample text and a semantic description text, the semantic understanding model is used for carrying out semantic understanding on the text to be understood based on representation features obtained by carrying out feature extraction on the semantic description text. According to the method anddevice, the electronic equipment and the storage medium provided by the embodiment of the invention, semantic understanding is carried out through the semantic description text, and rich informationrelated to semantic understanding and carried in the semantic description text can be fully utilized, so that the accuracy of semantic understanding under diversified language expression forms is improved.

The invention provides a semantic recognition method and device, electronic equipment and a computer readable storage medium. The method comprises the steps: receiving an intention statement inputtedby a user for a target scene; respectively calculating the similarity between the intention sentence and a plurality of preset reference sentences by adopting a preset similarity algorithm; and determining a preset reference sentence matched with the intention sentence according to the similarity, and obtaining a semantic attribute of the intention sentence according to the preset reference sentence matched with the intention sentence. According to the method, the similarity between the intention sentence and the reference sentence is calculated through the preset algorithm, so that the semantic attribute of the intention sentence is determined, the problems of overlarge sentence library scale and excessive resource occupation caused by various language expression combination forms and a large amount of information in the prior art are solved, and the effect of reducing resource occupation is achieved.

The invention discloses an invalid information determination method and device, computer equipment and a storage medium, and belongs to the technical field of the Internet. The method comprises the steps of extracting context features of a target feedback text, wherein the context features are used for representing a language expression environment of the target feedback text; extracting grammatical features of the target feedback text, wherein the grammatical features are used for representing grammatical attributes of the target feedback text; extracting semantic features of the target feedback text, wherein the semantic features are used for representing semantic attributes of the target feedback text; and determining whether the target feedback text is invalid information or not according to the context feature, the syntax feature and the semantic feature. According to the technical scheme provided by the embodiment of the invention, the labor cost can be reduced, and the user feedback information analysis efficiency is improved.

The invention discloses a morpheme-based language training method and control equipment. The method includes the steps: testing basic language ability to obtain a basic test score; formulating a corresponding training plan according to the basic test score; displaying or playing a training course according to the training plan, and acquiring a training condition; and reformulating a language ability test according to the training situation, acquiring a secondary test score, and comparing the basic test score with the secondary test score to form a result analysis report. According to the invention, the corresponding training plan is formulated according to the basic test score, then, the training course is displayed or played according to the training plan, the training situation is acquired, and the language ability test is reformulated according to the training situation to obtain the result analysis report by which a difference of language expression ability and understanding ability of a user before and after language training may be grasped to realize intelligent language training without manpower, so that language training operation of the user is simple, and labor cost is reduced.

The invention discloses a language recognition and training system for patients with language disorder and postoperative language expression difficulty, and belongs to the technical field of language recognition and training. The intelligent analysis system comprises a facial information input unit, a real-time data input unit, an expression duration statistical unit for performing statistics on facial expression duration of a patient, and a data feedback module for identifying facial information, and the data feedback module comprises a mouth shape information identification feedback unit for identifying mouth shape information and a vibration identification distinguishing unit for analyzing the pronunciation vibration of the patient. According to the language practice recognition system for the language disorder patients and the postoperative utterance expression difficulty patients, data captured by a real-time data input unit is transmitted to a data feedback module, and a mouth shape information recognition feedback unit in the module is combined with a facial information auxiliary feedback unit to carry out analysis based on a cloud database; therefore, the general and semantic meanings of the patient can be confirmed, and further communication or training of medical care and rehabilitation personnel can be facilitated.

Methods and systems for providing alternative query suggestions are provided. For example, a speech recognition component may receive a spoken natural language expression and convert it into a text query. A spoken natural language expression may include one or more words, terms and / or phrases. Phonetically confusable segments of the text query can be identified by a classifier component. The classifier component can determine at least one alternative query based on at least identifying a phonetically confusing segment of the text query. The classifier may further determine whether to suggest at least one alternative query based on whether the at least one alternative query is meaningful and / or useful. When it is determined to suggest at least one alternative query, the at least one alternative query may be provided to the user interface display and displayed on the user interface display.

The invention aims to provide a method, device and system for natural language content title disambiguation. Specifically, the method comprises the steps of outputting device type information of a device; determining a natural language command of the user from the semantic item information, and converting the natural language command to obtain actual semantic item information corresponding to thecontent title, and outputting the content corresponding to the content title according to the actual meaning item information, so that the real meaning of the user can be understood when the meaning expressed by the natural language of the user is converted into the meaning which is explicit in expression and the referred content is uncertain, the accuracy of user graph recognition is improved, and the voice interaction experience of the user is also improved.

The invention provides a special visual chart lightbox. The special visual chart lightbox comprises a display panel, wherein a visual chart is arranged on the panel; background lights are arranged inthe panel and include a red background light and a green background light; a control unit is arranged in the lightbox and is used for receiving a control signal and controlling lighting colors of thebackground lights. When the special visual chart lightbox is matched with a specific examination method, the subjective feeling of an examinee can be quantized with numbers, the examination results can be precisely compared, and misjudgment on clear degrees of different background sighting marks caused by subjective feeling and individual verbal expression can be avoided.

The systems and techniques disclosed here feature a device including an audio analyzer connected to microphone(s) monitoring a conversation between one or more individuals. The audio analyzer can identify auditory expressions from audio captured by the microphone, where the auditory expressions can include linguistic expressions and nonlinguistic expressions. The audio analyzer can analyze the linguistic expressions to identify a topic and a time period when the topic was discussed. A video analyzer can be connected to camera(s) monitoring the conversation. The video analyzer can identify non-auditory expressions by the individuals from video captured by the camera. A sentiment analyzer can analyze auditory expressions during the time period when the topic was discussed and non-auditory expressions during the time period when the topic was discussed to determine sentiments to the topic. A processor can analyze the determined sentiments to the topic to determine a group sentiment to the topic.

The embodiment of the present application discloses a text processing method and related devices, at least related to natural language processing and machine learning in artificial intelligence, and data parallel computing in cloud computing technology, etc., for the first text to be recognized, the second Text processing is performed on a text, and the first text is modified into a second text by modifying at least one content of the first text. Wherein, any content modification corresponds to a modification pair. In order to be able to determine the language expression error type and error cause corresponding to the content modification, not only based on the information provided by the modification pair, but also introduce the first text and the second text to complete the complete context information of the content modification before and after the modification. Therefore, in the modification On the basis of the correctness, by combining the first text and the second text, a relatively complete information basis is provided for determining the type of language expression error and the cause of the error in the revised pair, and the accurate identification of the type of text expression error and the specific error are realized. cause.

The present application discloses a language expression evaluation and assistance method and device, wherein the method includes: the server extracts the voice features of each chatting user, user features, and the user's voice satisfaction with different voice features from the user voice chat sample data in advance Degree, and use machine learning method, according to the result of the extraction, generate a learning model for voice expression evaluation; when the server receives the voice data to be evaluated and the user identification of the corresponding chat object sent by the terminal device, The voice data to be evaluated and the user identifier are used, and the learning model is used to estimate the satisfaction degree of the chat object to the voice data to be evaluated, and the estimated result is sent to the terminal device. By adopting the present invention, the user's language expression can be intelligently evaluated, which is beneficial for the user to optimize the language expression.

The invention relates to a language learning system and device based on a virtual scene. The system comprises a scheduling management module and a scene editing module;the system also comprises a correlation analysis module; the correlation analysis module analyzes and provides at least one correlation pattern and / or picture based on second language information inputted by a user; the correlation analysis module displays the correlation pattern and / or picture selected by the user on a user side; a teacher of the user side knows the language information expressed by the student in the second language based on the correlation pattern and / or picture. According to the language learning system and device based on the virtual scene of the invention, the correlation analysis module is arranged to display the related correlation pattern, and the teacher can be promoted to know the expression content of the student, so as to provide effective language guidance, and thus, language learning efficiency is improved.

A new technology of prediction of manipulability in response even to an instruction with missing information in an object manipulation task to have a robot manipulate some kind of object is provided. An instruction understanding system includes an obtaining engine configured to obtain a linguistic expression of a name of an object to be manipulated and a linguistic expression of a situation where the object corresponding to the name is placed in a real environment and a classifier configured to receive input of the linguistic expression of the name and the linguistic expression of the situation and output manipulability of the object corresponding to the name in the real environment.

The present invention relates a virtual-scene-based language-learning system, at least comprising a scheduling and managing module and a scene-editing module, and the system further comprises an association-analyzing module, the scheduling and managing module are connected to the scene-editing module and the association-analyzing module, respectively, in a wired or wireless manner, the association-analyzing module analyzes based on second-language information input by a user and provides at least one associated image and / or picture, and the association-analyzing module displays the associated images and / or picture selected by the user on a client, so that a teacher at the client is able to understand the language information expressed in the second language by the student based on the associated image and / or picture. The present invention displays associated image by setting up an association-analyzing module, which can promote teachers' understanding of students' expression content, thereby providing effective language guidance and improving language learning efficiency.

The invention aims to provide a method, device and system for natural language content title disambiguation. Specifically, the method comprises the steps of outputting device type information of a device; determining a natural language command of the user from the semantic item information, and converting the natural language command to obtain actual semantic item information corresponding to thecontent title, and outputting the content corresponding to the content title according to the actual meaning item information, so that the real meaning of the user can be understood when the meaning expressed by the natural language of the user is converted into the meaning which is explicit in expression and the referred content is uncertain, the accuracy of user graph recognition is improved, and the voice interaction experience of the user is also improved.

The embodiment of the invention discloses a text processing method and a related device, at least relates to natural language processing and machine learning in artificial intelligence, data parallelcomputing in a cloud computing technology and the like. Text processing is carried out on a to-be-recognized first text, at least one content of the first text is modified. the first text is modifiedinto the second text, wherein any content modification corresponds to one modification pair. In order to determine the language expression error type and the error reason corresponding to the contentmodification, the first text and the second text are introduced to perfect the complete context information of the content modification before and after the modification according to the information provided by the modification pair, so that on the basis of the modification pair, by combining the first text and the second text, complete information bases are provided for determining language expression error types of modification pairs and error causes of the language expression error types, and accurate recognition of the text expression error types and specific error causes are achieved.

The invention provides a disease early warning method and device based on behavior recognition, equipment and a storage medium, and relates to the technical field of audio and video processing. The method can warn guardians when people with limited language expression and disturbance of consciousness suffer from diseases, and avoids missing of disease diagnosis opportunities due to the fact that the guardians cannot clearly express the diseases. The method comprises the steps: acquiring object behavior images and background noise audios of objects from real-time audios and videos, wherein the objects comprise infants of 0-2 years old, language disabled people or Alzheimer's disease patients; calculating a first adaptation degree between the object behavior image and a preset image set, wherein the preset image set comprises a limb behavior image subset, an abnormal posture image subset and an abnormal behavior image subset; calculating a second adaptation degree between the background noise audio and preset abnormal audio information; when the first adaptation degree is larger than a first preset threshold value or the second adaptation degree is larger than a second preset threshold value, carrying early warning of the probability that the object is sick on a user.

The invention aims to provide a method, device and system for natural language content title disambiguation. Specifically, the method comprises the steps of outputting device type information of a device; determining a natural language command of the user from the semantic item information, and converting the natural language command to obtain actual semantic item information corresponding to thecontent title, and outputting the content corresponding to the content title according to the actual meaning item information, so that the real meaning of the user can be understood when the meaning expressed by the natural language of the user is converted into the meaning which is explicit in expression and the referred content is uncertain, the accuracy of user graph recognition is improved, and the voice interaction experience of the user is also improved.