1008 results about "Vector generation" patented technology

The invention belongs to the technical field of computer vision and deep learning, and particularly relates to a PCB defect detection method based on combination of a visual converter and conditional diffusion. Comprising the following steps: constructing an unlabeled PCB image data set and carrying out data preprocessing and enhancement to obtain a preprocessed image; executing a self-supervised pre-training task on the preprocessed image to obtain a feature extraction network; based on a conditional diffusion model, generating a synthetic defect PCB image and a label thereof by using the features output by the feature extraction network and the defect type control vector; mixing the synthetic defect image with a small number of real defect images to construct a training set; performing training adjustment on the defect detection model by adopting the training set to obtain a trained defect detection model; performing PCB defect detection by using the trained defect detection model; according to the method, the robustness and the cross-domain generalization ability are remarkably improved, the missed detection risk is reduced, and the rapid and stable quality control requirement of the production line is met.

The invention relates to the technical field of artificial intelligence, can be applied to business scenes such as agent autonomous decision making, financial science and technology and medical health, and discloses a task processing method and device based on visual attention enhancement, equipment and a medium. Visual hierarchical features are extracted, a double fovea attention module processes and fuses high-level visual features, a side suppression network obtains enhanced visual features, and a cross-modal fusion module generates fusion features by taking the enhanced visual features as query vectors and taking language components and action components as key and value vectors; and fusing the feature input decision network to generate target category and position information, generating feedback information based on actual label difference, and updating module parameters to complete a target task. According to the invention, through combination of a bionic vision mechanism and multi-modal attention fusion, the visual feature extraction and background suppression capability is improved, and the target capture efficiency and recognition precision in a complex scene can be improved.

The embodiment of the invention discloses a spinal metastatic tumor treatment scheme generation system. According to one specific embodiment, the system comprises a data processing server, an information fusion server and a scheme generation server which are in communication connection with one another, and the data processing server is used for preprocessing multi-source patient data to obtain standard multi-source patient data; the information fusion server is used for executing the following steps: performing feature code fusion on standard multi-source patient data to obtain a multi-source patient feature vector; performing feature enhancement on the multi-source patient feature vector to obtain a joint patient characterization vector; generating an initial therapeutic schedule result based on the joint patient characterization vector; and the scheme generation server is used for performing feature decision processing on the initial treatment scheme result to obtain a final treatment report. According to the embodiment, waste of computing resources can be reduced, and system response time can be shortened.

The invention relates to the technical field of security auditing, in particular to a security auditing method and system for an industrial control system, and the method comprises the following steps: aiming at a key control task, a communication task and a security task of a real-time operating system, collecting a period, a starting timestamp, a finishing timestamp, a central processing unit occupied time slice and peak memory usage amount data. According to the method, the period, timestamp, central processing unit occupation and memory usage data of a key task of a real-time operating system are collected, a task execution time boundary and a resource consumption envelope are set, and then an expected relation rule set of a task time sequence and resource consumption is constructed by applying historical data statistics and logic rule deduction; and meanwhile, the key length of symmetric and asymmetric encryption, initialization vector generation, hash algorithm selection, key derivation parameters and encryption operation context are stipulated, so that a comprehensive and specific ICS behavior specification baseline is established.

A robot control method based on tactile prediction pre-training comprises the following steps: acquiring and generating a human playing data set consisting of three-channel image tensors in an offline stage, and training a constructed conditional diffusion model comprising a tactile encoder, a tactile decoder and an action and visual encoder; in the online stage, the trained conditional diffusion model is integrated into a standard imitation learning strategy network, and an action instruction of the robot is generated according to the state of the robot, the current visual features and the tactile feature vectors extracted by the imitation learning strategy network. According to the method, a specific agent task is completed by training a deep neural network model, that is, a future tactile signal sequence is predicted according to historical information and future action intentions; the model is enabled to characterize generic haptic features contacting physical dynamic laws for further migration into downstream robot control tasks.

The invention provides a recommendation method and system fusing big language model reasoning and multi-source trajectory information, and the method comprises the steps: firstly obtaining user historical learning behaviors and static attribute information after receiving a user recommendation request, and constructing a static interest vector; in combination with the initial feature vector of the learned knowledge point and the map enhancement vector of the first-order neighbor node of the knowledge map, generating explicit and map extension interest vectors, and fusing to obtain a user interest vector; screening N unlearned knowledge points to form a candidate set through similarity analysis of user interest vectors and unlearned knowledge point vectors and / or reasoning of a large language model on user association information; and screening the target knowledge points through mastery degree verification of the pre-modified knowledge points, and outputting a recommendation result after sorting. According to the method, recommendation accuracy and suitability are improved, and personalized learning requirements are met.

The invention discloses an online video content intelligent pushing method combined with a learning interest model. The method comprises the following steps: constructing a dynamic interest vector based on multi-source user behavior data, generating a user interest portrait vector set, and performing interest dimension clustering and weight distribution; generating a video content feature vector set according to a clustering result of the user interest portrait vector set; establishing a multi-dimensional association relationship between the user interest portrait vector set and the video content feature vector set, and outputting a user-video matching confidence matrix; converting the user-video matching confidence coefficient matrix into a push sequence based on a multi-objective optimization strategy and issuing the push sequence; and feedback behaviors of the user on the pushed video are collected in real time to realize closed-loop optimization. The method has the following advantages and effects: accurate perception and deep semantic matching of the dynamic learning interest of the user can be realized, and the accuracy, timeliness and user satisfaction of content distribution are remarkably improved, so that the learning efficiency and experience are optimized.

The invention relates to the technical field of three-dimensional modeling, in particular to a dynamic modeling method for a geological structure three-dimensional model. The method comprises the steps that multi-source data are acquired and preprocessed, a voxel semantic fusion algorithm based on variational optimization is introduced, after semantic information is extracted from the preprocessed multi-source data, the preprocessed multi-source data are mapped to a target three-dimensional space grid, and an optimal semantic fusion vector is obtained; based on the optimal semantic fusion vector, generating a standard voxel data pool, and constructing a geological structure three-dimensional model; monitoring data change, calculating the position of a newly added data point, combining the standard voxel data pool to obtain a space updating area, and modeling the space updating area to realize model updating; and after the modeling of the space updating region is completed, optimizing the boundary continuity. The problems that multi-source geological data cannot be directly used for structure construction and semantic fusion of a three-dimensional model, dynamic response to newly-added data is lacked, and geometric discontinuity and structural logic discontinuity exist at the boundary are solved.

The invention discloses an adaptive evaluation method for the health degree of an electrolytic cell, and the method comprises the following steps: collecting the multi-dimensional operation parameters of the electrolytic cell in real time, and carrying out the preprocessing of the collected time series data, so as to construct a training sample with a time window; extracting a multi-scale time sequence feature from the training sample to form a feature vector; inputting the feature vector into a weight adjustment network, and outputting a dynamic weight vector; weighting the feature vector by using the dynamic weight vector to generate a weighted feature vector; inputting the weighted feature vector into a performance prediction model, and outputting a short-term performance prediction value of the electrolytic cell at a future moment; after the corresponding real performance value is obtained, calculating a prediction error of the short-term performance prediction value, and constructing a reinforcement learning reward signal based on the prediction error; updating a strategy of the weight adjustment network through a reinforcement learning algorithm by utilizing a reward signal, thereby optimizing dynamic weight vector generation at a subsequent moment; based on the dynamic weight vector and the feature vector at the current moment, a comprehensive health degree index of the electrolytic bath is obtained through calculation; according to the method, main factors influencing the equipment health degree in different stages are intuitively revealed, and a basis is provided for operation and maintenance decision making.

The invention discloses a network security attack path prediction system based on a graph neural network, and relates to the technical field of network security, and the system comprises a data collection module which collects network full-link time sequence security data, and outputs standardized time sequence data through time sequence alignment and abnormal noise reduction processing; the time sequence diagram construction module is used for constructing a dynamic attack graph containing nodes and time sequence edges; the feature learning module introduces a time sequence attention mechanism, calculates a time-space fusion attention coefficient based on a graph attention network framework, and outputs a node embedding vector; the reasoning and pruning module is used for generating attack paths based on node embedding vectors and outputting a high-value attack path set; and the analysis decision module is used for carrying out importance sorting on all nodes on the high-value attack path, determining a path core risk point and generating a key node decision basis of the attack path. According to the method, the problem that the traditional technology cannot accurately capture the attack behavior time sequence dependence is solved, and the high-precision prediction of the attack path is realized.

One embodiment sets forth a method for providing relevant search results for search queries. According to some embodiments, the method can be implemented by a client computing device, and includes the steps of (1) receiving a query, wherein the query is associated with a user account, and the user account is associated with a user account vector, (2) generating a query vector based at least in part on the query, (3) generating an output vector based at least in part on the query vector and the user account vector, (4) obtaining, based at least in part on the query, a plurality of digital asset vectors, wherein each digital asset vector of the plurality of digital asset vectors corresponds to a respective digital asset, (5) comparing the output vector to the plurality of digital asset vectors to generate respective similarity scores for the plurality of digital asset vectors, (6) filtering the plurality of digital asset vectors in accordance with the similarity scores to establish a filtered plurality of digital asset vectors, and (7) displaying, in accordance with the filtered plurality of digital asset vectors, respective affordances for the respective digital assets that correspond to the filtered plurality of digital asset vectors.

The invention discloses a multi-source unmanned aerial vehicle track fusion method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring unmanned aerial vehicle monitoring data; performing analysis and feature extraction on the unmanned aerial vehicle monitoring data, and forming a standardized data feature vector; generating a confidence score of each sensor at each moment based on the data feature vectors; dynamically adjusting parameters of a UKF or PF algorithm according to the confidence score, and performing trajectory fusion based on the adjusted parameters to obtain an initial trajectory fusion result; and performing optimization based on the initial trajectory fusion result, and dynamically adjusting computing resources according to a system load and a confidence score to obtain a final trajectory fusion result. By implementing the method provided by the invention, the fusion precision, the system robustness and the real-time calculation efficiency in a complex environment can be remarkably improved, and intelligent multi-platform data processing without manual intervention is realized.

The invention discloses a bridge inclination and settlement monitoring method and system based on multi-sensor fusion, and belongs to the field of bridge structure monitoring, and the method comprises the steps: obtaining the data of a multi-source heterogeneous sensor; the sensor data is converted into space-time diagram data, and the space-time diagram data comprises the steps that each sensor is mapped into nodes in a diagram, edges between the nodes are defined according to the physical connection relation of the bridge structure, and multi-source heterogeneous sensor data are unified into dynamic feature vectors with the same dimension on the nodes through learnable feature mapping; inputting the time-space diagram data into a preset neural network model, performing spatial feature aggregation on the dynamic feature vector through a diagram attention mechanism, performing time feature extraction through a time convolutional network, and generating a hidden state vector fused with time-space information; and generating a monitoring state value of the bridge based on the hidden state vector. According to the invention, depth feature fusion of spatial perception is realized, and the sensitivity of anomaly recognition is improved.

The invention relates to the technical field of geological disaster monitoring and early warning, and discloses a landslide susceptibility prediction method based on knowledge map and spatial-temporal feature fusion, and the method comprises the steps: extracting an inference feature vector from a geological knowledge map through a map neural network; extracting a spatial feature vector from the multi-source spatial data by using a convolutional neural network; extracting a time sequence feature vector from the rainfall time sequence data by using a Transform model; generating an attention weight based on the reasoning feature vector, and performing adaptive weighted fusion on the space and time sequence feature vectors by using the weight to obtain a fused feature vector; and inputting the fusion feature vector into the prediction model, and outputting the landslide occurrence probability. The geological priori knowledge in the knowledge graph is introduced to guide the fusion process of the spatial-temporal characteristics, so that the model can focus on the key disaster-inducing factor combination, the accuracy and reliability of prediction are remarkably improved, and the interpretability of the model is enhanced at the same time.

The invention discloses a traffic flow prediction method and device based on multi-level space-time and perception fusion, and the method comprises the steps: dynamic multi-level feature embedding, space-time and local mutation perception modeling and super-domain interaction fusion: firstly constructing a dynamic multi-level feature embedding module, and fusing original flow data, periodic labels and adaptive feature vectors; generating high-dimensional feature representation; then, a space-time and local mutation perception attention module is constructed, time dependence features, space dependence features and local mutation perception features are extracted through parallel time, space and local mutation perception attention mechanisms, finally, a super-domain interaction fusion module is constructed, and multi-source features are integrated through a cross attention and gating mechanism; uniform space-time representation is generated, and prediction robustness is improved. According to the method, an end-to-end framework for traffic flow prediction is formed, joint modeling and efficient prediction can be carried out on space-time dependence and non-stationary sudden change in a complex traffic scene, and butt joint with a traffic management system is facilitated.

The invention relates to the technical field of online education, and discloses an online course learning management method based on a knowledge graph. The method comprises the following steps: acquiring multi-modal learning behavior data of a learner, and extracting a deep learning state vector reflecting knowledge understanding depth, learning input degree and cognitive confusion through semantic fusion; and dynamically calculating and updating the logical relationship strength among the knowledge points in the course knowledge graph by using the vector, so that the knowledge structure can adaptively evolve along with the actual cognitive state of the learning group. And generating a real-time personalized learning path based on the updated knowledge graph and the current state vector of the learner. Meanwhile, according to cognitive confusion features in the state vector, intervention measures such as pushing of remedial resources, adjusting of content sequence or starting of self-adaptive testing are triggered in real time. According to the method, the dynamic optimization of the knowledge graph and the accurate and immediate response of learning intervention are realized, and the adaptability and management efficiency of online learning are improved.

The present application discloses a data processing method and apparatus, a device, and a readable storage medium. The method comprises: combining M vision mapping vectors generated from media data and N text mapping vectors generated from text information into a mapping vector sequence, and in the mapping vector sequence, inserting a compression token vector between the M vision mapping vectors and the N text mapping vectors to obtain a vision compression sequence; performing attention processing on the vision compression sequence to obtain an attention result vector; determining a unit attention vector associated with the compression token vector in the attention result vector as a global compression vector, the vector length of the global compression vector being less than the sum of vector lengths of the M vision mapping vectors; and generating a question-answer result on the basis of the global compression vector and unit attention vectors associated with the N text mapping vectors. By using the present application, vision mapping vectors can be compressed, thereby reducing calculation costs while improving model performance.

The invention discloses an intelligent education robot question answering system based on voice recognition and a knowledge graph, and particularly relates to the technical field of artificial intelligence. The method comprises the following steps: performing feature extraction on a user voice signal to obtain a text sequence and a multi-modal context feature; recognizing subject domain judgment and question answering intentions based on supervised classification and keyword rule fusion, and outputting a subject domain prior probability and a question answering target vector; determining polysemy words in the text sequence by using a context window, generating a semantic item sequence and semantic item confidence, constructing a subject domain sub-graph based on a subject domain prior probability, and obtaining a candidate reasoning path and a path scoring vector; determining a target teaching concept and an optimal reasoning path by combining Bayesian inference and consistency verification; generating a personalized question answering result aiming at the question asked by the user through fact retrieval and knowledge derivation in combination with the question answering target vector; accurate and efficient intelligent teaching question answering can be realized, and question answering accuracy and intelligent interaction capability of the education robot are effectively improved.

The invention discloses a data-driven perfusion process automatic adjustment method based on machine learning, and the method comprises the steps: collecting and preprocessing multi-source time sequence data, and generating an alignment feature vector sequence; performing perfusion stage division and stage coding vector generation based on the aligned feature vector sequence; constructing a three-layer liquid state machine model, and determining a stage liquid pool activation sequence; stage gating coding is executed, the liquid pool is driven to generate a dynamic state, and cross-stage migration is completed; integrating the dynamic state sequence, generating a joint state, and inputting a multi-task readout layer to output adjustment parameters; and executing distribution drift detection, topology updating and parameter calibration, and outputting final perfusion adjustment parameters. Through a data driving method based on stage topology modeling, liquid state machine dynamic evolution and a multi-task readout mechanism, accurate prediction, risk identification and self-adaptive adjustment of the perfusion process are achieved, and the perfusion quality and long-term operation stability are improved.

The invention relates to the field of digital power supply control, and provides a parameterized digital control method and system suitable for a wide power range charging module, and the method comprises the steps: collecting an original sensor signal, constructing a slow working condition feature vector through adaptive-temperature drift compensation, and driving a physical constraint reference prediction model to generate a second baseline behavior parameter vector; monitoring a gate driving current by using a gate-drain capacitor, extracting a transient characteristic value, and quantifying the transient characteristic value into a ZVS quality factor; the slow working condition feature vector and the ZVS quality factor are fused, a correction vector is generated through an adaptive prediction model, a control parameter vector is generated in combination with a second baseline behavior parameter vector, and online updating is carried out according to feedback; executing hardware physical limit constraint according to the security constraint set, generating security parameters and abnormal flag bits, and writing the security parameters and the abnormal flag bits into a hardware register in a fixed-point manner after cooperating with ZVS quality factor arbitration. According to the method, physical constraint prediction and transient perception correction are fused, and a digital control system considering dynamic response, drift compensation and safety redundancy is constructed.

The invention relates to the technical field of medical RFQ accelerators, and particularly discloses a medical RFQ accelerator multi-safety redundancy linkage protection system and method, and the method comprises the steps: collecting the working state image data of an accelerator, and generating a three-dimensional fusion feature map containing heat distribution, electromagnetic radiation intensity and mechanical vibration; a multi-dimensional safety index evaluation vector is output by using a multi-mode Transform model; constructing a dynamic safety boundary prediction model, and generating a real-time safety threshold curve of each subsystem according to the evaluation vector; performing parameter optimization by adopting a federated learning framework; constructing a non-tampering security event traceability chain by using a block chain; constructing a multi-stage protection strategy of the medical RFQ accelerator; and verifying the redundancy control channel through a cross-layer security verification mechanism, and finally executing a protection action. According to the invention, multiple safety redundancy protection of the medical RFQ accelerator is realized, the risk of false shutdown is reduced, and the requirement of high reliability of clinical radiotherapy is met.

The invention discloses a mountainous area sudden change wind field active control wind tunnel simulation method based on deep learning, and relates to the technical field of wind tunnel experiment research, and the method comprises the steps: obtaining a target wind field parameter in a mountainous area target region, constructing an MLP input vector according to the target wind field parameter in combination with wind tunnel real-time sensor data, a deep learning control model comprising an MLP forward mapping model and a Transform time sequence feedback model is established, the MLP forward mapping model generates a multi-fan array control signal according to the MLP input vector, a multi-fan array is controlled to generate a simulated wind field, the Transform time sequence feedback model generates a multi-fan array control correction signal according to a wind field difference value between the simulated wind field and a target wind field, and the multi-fan array control correction signal is controlled to generate a multi-fan array control correction signal according to a wind field difference value between the simulated wind field and the target wind field. The simulated wind field parameters are dynamically corrected in real time, and target wind field parameters are approached; according to the method, high-fidelity dynamic reproduction and intelligent adaptive control of the complex wind field in the mountainous area are realized through MLP feedforward generation and Transform feedback correction.

The invention relates to the technical field of power data analysis, in particular to a line loss abnormity associated mutual inductor insulation fault positioning method. Collecting multi-source operation data of a target station area, and obtaining a datum line loss rate sequence, a residual component sequence and an auxiliary feature vector based on the multi-source operation data; according to the reference line loss rate sequence, obtaining a comprehensive sensitivity vector representing the influence degree of the mutual inductor on the line loss; obtaining a preliminary diagnosis result according to the comprehensive sensitivity vector, the residual component sequence and the real-time load rate; performing spatial logic verification according to the preliminary diagnosis result and the electrical topological graph of the transformer area to obtain a fault positioning list; generating a correction amount according to the fault positioning list and the corresponding comprehensive sensitivity vector, and correcting the line loss rate according to the correction amount; and meanwhile, feedback optimization is performed on the machine learning model according to complete data of the diagnosis. According to the invention, accurate and efficient diagnosis and correction of the line loss abnormity of the target station area can be realized.

The invention discloses a data tamper-proofing method, system and device based on blockchain evidence storage, and a storage medium, relates to the field of data processing, and is used for solving the problems that multi-source heterogeneous data lacks a unified processing scheme and evidence storage verification is incomplete in existing property management. The method specifically comprises the following steps: multi-modal feature processing: extracting text and image data features respectively, and generating a unified feature vector by combining cross-modal attention mechanism weighted fusion; generating a hash value and a block chain evidence, generating the hash value by adopting an SHA-256 function, and generating a dynamic two-dimensional code in an association manner; and user verification: after scanning the two-dimensional code, locally reproducing feature processing and Hash calculation, and comparing a Hash value on a chain through an elliptic curve cryptography technology to complete verification. According to the method, full-process credible evidence storage of the property data is realized, the data tamper-proofing capability and verification convenience are improved, and property scene data management requirements are met.

The invention relates to the technical field of data processing, and provides an internet text translation system based on artificial intelligence, a feature extraction module generates genre, style vectors, culture mark sequences and rhythm feature vectors through a deep neural network, the problem that specific attributes of internet texts are difficult to accurately capture in the prior art is solved, and a fine-grained basis is provided for translation; the dynamic translation module dynamically adjusts decoding parameters, interruption opportunity and translation method selection to realize style restoration, culture adaptation and rhythm matching, and overcomes the defects of traditional translation style distortion, culture element processing rigidity and rhythm dislocation; the feedback optimization module enables the system to continuously learn new language phenomena and user requirements through a knowledge vector generation and increment updating mechanism, solves the problem that a static system is poor in adaptability, and improves the accuracy, adaptability and efficiency of internet translation through cooperation of the three mechanisms.

The invention discloses a large-scale building group layout optimization method based on deep learning. The method comprises the following steps: S1, obtaining a three-dimensional rasterized matrix and a meteorological environment vector of a to-be-optimized building group; s2, on the basis of a pre-trained three-dimensional building layout-wind field mapping network, generating a global wind environment distribution map according to the three-dimensional rasterized matrix and the meteorological environment vector; s3, based on the global wind environment distribution diagram, calculating a building layout parameter gradient through a gradient back propagation optimizer, and generating a layout optimization scheme; and S4, based on the layout optimization scheme, outputting optimized building group space layout parameters. Intelligent optimization of large-scale building group layout is realized through a deep learning technology, and urban wind environment quality and planning and design efficiency are remarkably improved.

A system converts a first input idea vector representing an idea into a first contracted vector. The system generates one or more second contracted vectors in a multivariate space to which the first contracted vector belongs, based on a value of a first predetermined component of the first contracted vector in the multivariate space. The system generates, respectively from the one or more second contracted vectors, one or more first output idea vectors representing a candidate idea to be proposed to a user. The multivariate space is configured to maintain a similarity between an input idea vector for generating a contracted vector and an output idea vector generated from the contracted vector and a similarity between the first predetermined component of the contracted vector and a first predetermined index value.

The invention relates to the field of image deblurring, and discloses a multi-scale image deblurring method based on a potential space condition diffusion model, comprising the following steps: constructing a multi-scale image deblurring network which comprises a condition diffusion model and a sliding window attention module, the conditional diffusion model is used for generating a multi-scale prior feature from the fuzzy condition vector in a potential space; the sliding window attention module is a U-shaped network based on an encoder-decoder and is used for executing image deblurring feature regression guided by multi-scale prior features; training the network by adopting a two-stage strategy comprising pre-training and post-training; and inputting a blurred image to be processed into the trained multi-scale image deblurring network, and outputting a final deblurred image. According to the method disclosed by the invention, the common problems of excessive smoothness and artifacts in the deblurring process can be effectively inhibited while the calculation efficiency is ensured, and the recovery precision of texture details and edge structures is improved.

The invention relates to the technical field of robots, and discloses a robot action prediction method and device, computer equipment and a storage medium, and the method comprises the steps: generating an input observation sequence according to a position code and a current feature vector, and predicting an action sequence through Transform and the input observation sequence; and exponential decay weighted average processing is carried out on the predicted action sequence, and a target action is determined. Through the above mode, the multi-modal observation data is converted into the unified feature vector through the feature extraction network, key information in the observation data is reserved, the time sequence dependency relationship and context information in the observation data are fully captured by using the Transform decoder, and the robot action sequence is accurately predicted. And exponential decay weighted average processing is performed on the predicted action sequence, so that the action sequence is further smoothed, the prediction instability is reduced, the finally determined target action is more accurate, and the performance and success rate of the robot during task execution are improved.

The invention discloses an interactive question and answer task processing method based on an AI large model, and relates to the technical field of AI questions and answers, and the method comprises the following steps: performing correlation screening and function label labeling on high-confidence sub-queries in a retrieval result set, performing weight reduction on low-confidence sub-queries, constructing a cross-source consistency constraint vector, and generating an input data set; based on the input data set, a multi-source evidence consistency verification channel is constructed, entity-by-entity alignment and conflict detection are executed, the credibility interval of answers is calculated, meanwhile, voice emotions are recognized, and an answer data set is generated; and converting the answer data set into multi-modal feedback, monitoring user behaviors in real time, calculating behavior response strength indexes, dynamically adjusting a feedback form, and generating an interactive question and answer data set. The semantic consistency constraint of the multi-modal evidence is realized, and the robustness of answer credibility evaluation in the natural language processing task is improved.