Patents
Literature
Hiro is an intelligent assistant for R&D personnel, combined with Patent DNA, to facilitate innovative research.
Hiro

30 results about "Expressive communication ability" patented technology

Expression ability dimension evaluation method and device for intelligent interview

The invention discloses an expression ability dimension evaluation method and device for intelligent interview, and the method comprises the steps: collecting an interview video, extracting a video frame and audio data, firstly inputting the video frame into a CNN-based hand detection model to obtain a bottom-layer hand posture feature, and inputting the audio data into an audio processing module based on natural language processing to obtain a bottom-layer voice expression feature; respectively inputting the two underlying features into two LSTM codes to form time sequence features; extracting abstract high-level semantic features, including hand expression, expression fluency, infectivity, orderliness and conciseness information, from the two low-level features through a priori knowledge rule; standardizing the time sequence features and the high-level semantic features and then inputtingthe standardized time sequence features and the standardized high-level semantic features into a full-connection neural network, and predicting the expression ability level. The system uses distributed function calculation service construction to form an expression ability evaluation system. According to the method, the efficiency problems of subjective deviation and time cost of expression ability scoring in interviews are solved, and rapid and accurate expression ability evaluation of a large number of interview videos is realized.
Owner:南京智能情资创新科技研究院有限公司

Emotional dialogue generation method and system based on interactive fusion

The invention discloses an emotional dialogue generation method and system based on interactive fusion. The method comprises the following steps: S100: receiving a dialogue statement text; S200, an emotion classifier identifying an emotion category contained in the dialogue statement text and representing the emotion category as a vector; S300, encoding the dialogue statement text into a context vector by an encoder; S400, a fine-grained interaction module fusing the emotion category representation vector and the context vector of the dialogue statement text to generate an interactive vector; S500, the decoder performing decoding by using the interactive vector to generate a reply; and S600, outputting a reply. The problems that an existing man-machine conversation system is weak in emotion perception ability and insufficient in emotion expression ability can be solved.
Owner:STATE GRID E COMMERCE CO LTD +1

Emotion recognition method based on context interaction relationship

ActiveCN113076905ASolve the problem of ignoring the rest of the branchesProof of importanceBiometric pattern recognitionNeural architecturesPattern recognitionFace detection
The invention discloses an emotion recognition method based on a context interaction relationship. The method comprises the following steps: performing face detection and human body recognition on an expression data set to obtain a bounding box of a face and a body; preprocessing the pictures by using bounding boxes of the human face and the body, and generating space masks for the bounding boxes to obtain three types of pictures of the human face, the body and the scene; inputting the preprocessed images into three pre-trained branch networks to extract features, wherein a context interaction module is inserted into a second layer and a fourth layer of the network, and features of other branches in the interaction module are weighted and fused to each branch; and performing expression classification in combination with the face emotion features, the body emotion features and the scene emotion features to form an emotion recognition model based on a context interaction relationship. According to the method, the feature expression ability of the context is improved, the noise existing in the context is inhibited, the problems of emotion uncertainty and noise during independent extraction of context features are solved, and the emotion recognition accuracy is higher.
Owner:SOUTH CHINA UNIV OF TECH

Language training auxiliary system and method based on residual memory of senile dementia

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.
Owner:FUJIAN UNIV OF TRADITIONAL CHINESE MEDICINE

Speech and discourse evaluation integrated little speech practice system for Chinese

PendingCN112668883AAccurately point out the problems existing in the two aspects of speech and discourseSpeech analysisNatural language data processingSpoken languageImitation learning
The invention provides a speech and discourse evaluation integrated little speaking practice system for Chinese. The system comprises an ocean cavity calabash speech evaluation module which is used for carrying out the multi-dimensional speech quality evaluation of speech inputted by a user; the discourse evaluation module is used for carrying out voice recognition on the voice input by the user, allowing the user to edit and modify, and carrying out multi-dimensional discourse quality evaluation on the text modified by the user; the model essay learning module is used for providing reference model essays and mandarin audios of the reference model essays for the questions selected by the user for the user to listen and read to realize imitation learning; receiving a record of the user according to the selected question; then evaluation is carried out through the ocean cavity calabash voice evaluation module and the discourse evaluation module; displaying the evaluation report; the reference model essay of the selected question and the mandarin audio thereof are acquired according to the model essay learning module, so that the standardized spoken language expression ability of students is improved, and the individual spoken language learning requirements of Chinese learners are met.
Owner:HUAQIAO UNIVERSITY

Video interview evaluation method and device based on multi-modal analysis and storage medium

The invention provides a multi-modal analysis-based video interview evaluation method. The method comprises the following steps: receiving a short video of an interviewer applying to a specific post during interview; extracting audio data and video data in the short video; extracting the Mel-frequency cepstrum coefficient of the audio data as the input of the expression ability evaluation model based on voice analysis to obtain an expression ability evaluation score based on voice analysis; analyzing the video data to serve as input of an occupational matching model of the personal basic attributes, an image occupational degree evaluation model and a personality evaluation model of action analysis, and respectively obtaining an occupational matching degree score based on the personal basic attributes, an image occupational degree score and a personality evaluation score based on action analysis; and obtaining a comprehensive post matching degree score through summation and summation. According to the invention, rapid screening in massive video resumes can be realized, and the recruitment efficiency is improved.
Owner:无锡爱视智能科技有限责任公司

English language learning evaluation device

InactiveCN111899603AImprove oral expression skillsStrengthen oral English practiceTeaching apparatusInput/output processes for data processingSpoken languageDisplay device
The invention discloses an English language learning evaluation device, and relates to the field of evaluation equipment; the English language learning evaluation device comprises a terminal operationmodule, a voice acquisition unit, a processor module, a voice evaluation module, a voice broadcast unit, a wireless transmission module, a storage module, a storage battery, a touch screen displayerand an evaluation assembly. The processor module comprises a voice analysis unit, a voice comparison unit and a controller; a microphone and a loudspeaker are arranged in the touch screen displayer; the evaluation assembly comprises a support. The upper end of the bracket is fixedly connected with a storage box; spoken English voices of students are collected through the voice collection unit in the microphone; after the collected voice is analyzed and compared through the voice analysis unit and the voice comparison unit in the processor module, the similarity degree is judged and scored, andthen broadcasting is performed through the loudspeaker, so that students can know own spoken language evaluation conditions, the students can conveniently and specifically strengthen spoken English practice, and the spoken language expression ability of the students is improved.
Owner:SICHUAN INFORMATION TECH COLLEGE

Method and device for evaluating language reading ability by using non-lexical body symbols

The invention provides a method and equipment for evaluating language reading ability by using non-lexical body symbols, and the method comprises the steps: obtaining video data of the non-lexical body symbols of a target object through a multi-vision machine vision technology, converting the video data of the non-lexical body symbols into three-dimensional data files, and carrying out the classified storage of the three-dimensional data files; identifying the specific features of the non-lexical body symbols of the target object, and classifying and labeling the specific features of the non-lexical body symbols of the target object; inputting the specific features of the non-lexical body symbols of the target object into a data mapping rule model for classification, deconstruction and analysis; and mapping the non-word body symbol evaluation criteria to obtain the language reading ability and expression ability of the target object. The method has the beneficial effects that the language reading ability and expression ability of the children can be analyzed and evaluated from the perspective of behaviors, so that a standard for screening the early behaviors of the language reading disorder of the children is established.
Owner:凌云美嘉(西安)智能科技有限公司

Text translation model training method and device and storage medium

The invention provides a training method and device for a text translation model and a storage medium, and the training method comprises the steps: receiving a text training sample set containing multiple pairs of natural language texts from a client, carrying out the comparison learning based on the word level, and combining the word frequency information of the natural language texts, the model parameters of the text translation model are optimized, a final text translation model is obtained through multiple rounds of training until the loss function of the text translation model converges, and the text translation model is used for translating one natural language text into another natural language text. Since the training process can optimize the model parameters of the text translation model based on the word frequency information, semantic space representation of words with different word frequencies is optimized, the semantic expression ability of the model to the input text is enhanced, and the performance of the machine translation model is improved.
Owner:ALIBABA (CHINA) CO LTD

A method of generating topics for rapid expressive force test

The invention relates to the technical field of machine learning, in particular to a method for generating topics for a rapid expression ability test, and aims to solve the problems that the artificial proposition efficiency of the existing expression ability test is low, human factor deviation exists, and a large amount of manpower and time costs are often consumed for preparing test questions once. In order to solve the problem, the invention provides a quick topic generation technology which can be used in an expression ability test, and by combining a currently popular web crawler technology and basic characteristics of an expression ability evaluation language material, effective information of a text is quickly extracted by utilizing a natural language processing technology; expression evaluation questions which are formally used are effectively helped to be generated by experts.
Owner:上海擅择教育科技有限公司

Speech emotion recognition method

The invention relates to a speech emotion recognition method. The method comprises the following steps: firstly, carrying out fragmentation operation on an input audio signal, then extracting features capable of expressing emotion information through an audio feature extraction method, and then carrying out calculation and preliminary splicing fusion on each feature by using a function; and inputting the fused features into the constructed one-dimensional convolutional neural network model for training and recognition. The defects of singleness and limited expression ability of a single feature are made up by adopting multi-feature fusion, and a method of re-fusing multiple expression emotion features can obtain speech emotion information from different angles and levels and perform more comprehensive description on the speech emotion information so that a system obtains a higher recognition rate, meanwhile, the system robustness is improved; and the emotion change can be mastered more accurately through audio signal fragmentation recognition, and the problem that the emotion change is ignored due to the fact that only the tag with the maximum probability is returned in whole audio recognition is avoided.
Owner:青岛洞听智能科技有限公司

Chinese speech recognition method based on pinyin constraint joint learning

The invention relates to a Chinese speech recognition method based on pinyin constraint joint learning, and belongs to the technical field of natural language processing. According to the method, firstly, pinyin texts corresponding to voices and texts are collected from a public Chinese corpus set, secondly, speech features are encoded through a shared encoder, then pinyin speech recognition is used as an auxiliary task, and then pinyin is used as a decoding constraint in the decoding process. Pinyin speech recognition and Chinese speech recognition are combined for learning based on a sharedencoder, inductive bias closer to speech is introduced, and the expression ability of the encoder for Chinese speech is enhanced. According to the Chinese speech recognition method based on pinyin constraint joint learning provided by the invention, the word error rate of Chinese recognition is reduced, and powerful support is provided for subsequent work such as pinyin fusion and pinyin error correction in the Chinese speech recognition process; and the problem that the end-to-end model is difficult to converge in Chinese character recognition is relieved.
Owner:KUNMING UNIV OF SCI & TECH

Aspect-level text sentiment analysis method based on heterogeneous graph neural network

The invention discloses an aspect-level text sentiment analysis method based on a heterogeneous graph neural network, and belongs to the field of language processing. The method comprises the following steps: constructing a word-sentence-evaluation aspect three-level graph structure network according to a co-occurrence relationship between words and sentences in a text and evaluation aspects contained in the sentences; then obtaining an initial embedded vector representation of each node; training model parameters by using a graph attention network, continuously updating embedded vector representation of nodes in the graph network according to a connection relation of the nodes in the graph network through a multi-head attention mechanism, and finally predicting aspect-level emotional tendency of the text; and according to the finally obtained embedding vector representation of the sentence node and the evaluation aspect node, calculating the correlation between the sentence node and the evaluation aspect node by using a self-attention mechanism, thereby obtaining the predicted text aspect level emotional tendency. According to the method, the expression ability and generalization ability of the model are effectively improved.
Owner:XI AN JIAOTONG UNIV

Data processing method, device, equipment and storage medium for evaluating test drive experience

ActiveCN110797050BGuaranteed accuracyAccurately reflect the experience effectSpeech analysisEvaluation resultEmotion assessment
The present invention provides a data processing method, device, device, and storage medium for evaluating test drive experience. The method includes: acquiring voice information in the car during the test drive, and seat information corresponding to the voice information, The seat information is used to characterize the seat of the person who generated the voice information; identifying the emotion in the voice information to obtain initial emotional evaluation data; according to the initial emotional evaluation data of each voice information and the seat information, determine The final emotional assessment data for this test drive experience. In the present invention, since the accuracy of the evaluation result is not affected by people's ability to remember and understand, and is not easily affected by the ability to express, the accuracy of the evaluation can be guaranteed to a certain extent. The present invention also fully considers the specific conditions of people in different seats, so as to more accurately reflect the overall experience effect when multiple people participate in the test drive.
Owner:上海能塔智能科技有限公司

Pre-natal beta-cryptoxanthin benefits children

Beta-cryptoxanthin administered to a pregnant woman has numerous benefits. It can decrease anxiety, and lessen the risk of her developing anxiety during her pregnancy. Further it can impart various benefits to her child: increased cognition, increased receptive language skills, fine motor skills, and gross motor skills. Also lycopene administered to the pregnant woman can increase expressive language in the child. Pre-natal supplements are also included in this invention.
Owner:AGENCY FOR SCI TECH & RES +1

A language recognition method and system based on an adaptive central anchor

The invention discloses a language recognition method and system based on an adaptive central anchor, which includes constructing a language data set; performing enhanced processing on speech data in the language data set; extracting features of the enhanced speech data to generate a feature data set; Constructing a deep neural backbone network, and using a supervised learning method to train the deep neural backbone network based on the classification loss function; based on the metric loss function, using an adaptive central anchor method to further train the deep neural backbone network; based on the trained deep neural backbone network Carry out language recognition; this method expands the data set through data enhancement, and also increases the robustness of the data set; uses a multi-scale fusion residual neural network to enhance the feature expression ability of the backbone network, and uses classification loss and measurement successively The loss further improves the accuracy of recognition.
Owner:北京快鱼电子股份公司

A key frame extraction method of gesture images based on deep learning

The invention discloses a method for extracting key frames of gesture images based on deep learning. First read the input gesture video, convert the input gesture video into a video frame image; use the Mobilenet-SSD target detection model to detect gestures in the video frame image, and segment the detected gestures; use VGG16 training model training The image is segmented by hand gestures to obtain the corresponding abstract features, and the spatial gradient is calculated. According to the gradient difference between two adjacent frames of pictures, an appropriate threshold is set to determine the key frame. The present invention proposes to use the Mobilenet-SSD target detection model to detect and segment the hand area, remove the noise in the background area, and use VGG-16 to accurately extract the abstract features of the hand, which not only greatly enhances the expressive ability of the picture, but also reduces the amount of parameters. The complexity of the model is reduced, which is suitable for this kind of video key frame extraction with small changes.
Owner:康旭科技有限公司

Compression method and platform of pre-trained language model based on knowledge distillation

The invention discloses a compression method and platform of a pre-trained language model based on knowledge distillation. The method first designs a universal knowledge distillation strategy for feature transfer, and in the process of distilling the knowledge of the teacher model to the student model, the The feature map of each layer of the student model is close to the characteristics of the teacher, focusing on the expressive ability of small samples in the middle layer of the teacher model, and using these features to guide the student model; then using the self-attention distribution of the teacher model to detect the semantic and The syntactic ability builds a cross-knowledge distillation method based on self-attention; finally, in order to improve the learning quality of the learning model in the early stage of training and the generalization ability in the later stage of training, a linear migration strategy based on the Bernoulli probability distribution is designed and gradually completed from Knowledge transfer with teacher-to-student feature maps and self-attention distributions. Through the present invention, the multi-task-oriented pre-training language model is automatically compressed, and the compression efficiency of the language model is improved.
Owner:ZHEJIANG LAB

A gait recognition system and method based on deep learning of self-attention mechanism

The invention discloses a gait recognition system and method based on deep learning of a self-attention mechanism, belonging to the field of gait recognition. The present invention proposes to use the attention mechanism on the original feature map. By learning a matrix between 0 and 1 with the same size as the original feature, the original feature is denoised, and the salient features in the picture are selected to reduce the noise in the feature map. The noise of the classification loss and the comparison verification loss are organically combined to punish the features in combination with the loss function, which not only uses the identity information of the target, but also uses the different relationships between the targets to increase the discrimination between different features; The most important limb features in gait are added to the original depth features as prior knowledge. This combination can not only use the target global body shape features, but also correct the features that are not conducive to classification brought about by the clothes transformation learned by the global features, and enhance the original depth features. The expressive ability of deep features enhances the expressive ability of features from different dimensions.
Owner:HUAZHONG UNIV OF SCI & TECH

Speech and language function impairment level converter based on ICF framework

The invention discloses a speech and language function impairment level converter based on an ICF framework. The speech and language function impairment level converter comprises the following function evaluation indexes: auditory understanding (%), visual understanding (%) and right brain function (%); and S4, the b16710 spoken language expression ability is reclassified and subdivided into wordnaming (%), simple repetition (%), word repetition (%), bisyllable word duration 2cvT(s), bisyllable word fundamental frequency 2cvF0 (Hz), sentence repetition (%), sentence duration (s), sentence fundamental frequency (Hz), series of words (%), spoken language description (%), writing (%) and body language (%), so that accurate language function evaluation can be carried out on the patient, and subsequent treatment of the patient is facilitated.
Owner:上海慧敏医疗器械有限公司

Semantic logic expression and analysis method based on spoken language dialogue characteristics

The invention discloses a semantic logic expression and analysis method based on spoken language dialogue characteristics. A dialogue text of a dialogue role is input according to a spoken language dialogue sequence; according to the method, the comprehensive logic rule expression capability of multiple semantic expressions can be realized by adopting a definition mode of a keyword sequence rule and combining a logic relation expression consisting of and, or, not and brackets, so that information feature extraction and analysis are performed on any spoken language dialogue type text. Accordingto the semantic logic expression and analysis method based on the spoken language dialogue characteristics, the context characteristics in a specific business dialogue scene can be fully explored, alogic and mode more conforming to thinking of people are adopted, a complex logic expression and analysis tool can be provided for specific business scenes and purposes, and the method has wide application prospects.
Owner:慧捷(上海)科技有限公司

Intelligent moral quality online interactive education system and method

The invention discloses an intelligent moral quality online interactive education system and method, and belongs to the technical field of education systems. According to the system, a video course is set, so that students complete course learning and establish a correct moral quality view; through a community law module, students analyze and express hot events from different angles, on one hand, the ability of the students to learn to use knowledge points learned from books is exercised, on the other hand, the oral expression ability and thinking logic of users are exercised, and the comprehensive ability of the users is improved; and through a character community module, students deeply experience different effects generated by executing decisions and better understand the content learned by video courses, intelligent operation is realized, the operating cost of the system is reduced, the students exercise their comprehensive quality and ability while playing and learning, audiences of moral quality education are expanded, and a cultural atmosphere of lifelong moral quality-oriented education is formed.
Owner:盐城国睿信科技有限公司

Natural language word segmentation method and system, natural language classification method and system

The present application discloses a method for constructing natural language word segmentation. The method includes: obtaining Ngram word segmentation of natural language, and discontinuously combining the Ngram word segmentation according to the sequence of positions to obtain the combined word segmentation. It can be seen that on the basis of the original Ngram word segmentation, the obtained Ngram word segmentation is combined according to the order of the position sequence. Through this method, the expressive ability of natural language is increased, and the classification accuracy of natural language is improved. Moreover, in this solution, the window size is further set for the combined word segmentation, so as to avoid generating too many combined word segmentation feature quantities, and improve the classification speed of natural language. Correspondingly, the present invention also discloses a natural language word segmentation construction system, a natural language classification method and system, which also have the above beneficial effects.
Owner:MASHANG CONSUMER FINANCE CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products