1376results about How to "Reduce overhead" patented technology

Systems and methods for managing divergence of best effort transactional support mechanisms in various transactional memory implementations using a portable transaction interface are described. This interface may be implemented by various combinations of best effort hardware features, including none at all. Because the features offered by this interface may be best effort, a default (e.g., software) implementation may always be possible without the need for special hardware support. Software may be written to the interface, and may be executable on a variety of platforms, taking advantage of best effort hardware features included on each one, while not depending on any particular mechanism. Multiple implementations of each operation defined by the interface may be included in one or more portable transaction interface libraries. Systems and / or application software may be written as platform-independent and / or portable, and may call functions of these libraries to implement the operations for a targeted execution environment.

The invention provides a Python monitoring task resource use method and system, and relates to the technical field of intelligent monitoring and resource optimization, and the method comprises the steps: extracting process call stack information through a multilayer convolutional neural network, constructing a process dependency graph, calculating a process criticality score, and recognizing a core monitoring process. And obtaining resource use data of the core monitoring process through a resource monitoring coroutine, determining an optimal sampling period by adopting a self-adaptive feedback control algorithm, and writing the data into a time sequence database. And performing data aggregation analysis by using a sliding time window algorithm to obtain a performance index, and constructing an anomaly detection model. Dimensionality reduction is carried out based on a principal component analysis method to construct feature vectors, and load types are divided by adopting a clustering algorithm; and generating a differentiated process optimization strategy according to the feature vector and the load type. When performance fluctuation is detected, an optimization strategy is executed, the anomaly detection model and the deep reinforcement learning model are updated according to the performance improvement index, and finally an optimization diagnosis report is generated.

The invention discloses an intelligent server management method and system in an edge computing environment, and relates to the field of electrical digital data process.The method comprises the steps that a dynamic load buffer database is established; calculating a load stability coefficient of each server node; when the load stability coefficient of the server node is smaller than a first preset threshold value, marking the server node as a fluctuation node; calculating a node cooperation efficiency value; determining a target server candidate set, and calculating to obtain a task allocation weight; distributing tasks on the fluctuation nodes to server nodes in the target server candidate set; when the continuous preset duration of the real-time load value of the fluctuation node is lower than a third preset threshold value, calculating a processing efficiency improvement proportion; for the task of which the processing efficiency improvement proportion is greater than a fourth preset threshold value, keeping continuous processing at the current server node; and for the tasks of which the processing efficiency improvement proportion is smaller than or equal to a fourth preset threshold value, returning to the original fluctuation node. The method and the device are used for reducing communication overhead and service processing delay between nodes.

The invention discloses a tensor core-based method for accelerating arbitrary precision sparse matrix multiplication and addition operation, which comprises the following steps of: aiming at an architecture of a tensor core on A100 and a programming model thereof, combining program characteristics of deep learning parallel processing sparse matrix, combining an arbitrary precision sparse matrix multiplication and addition operation operator with parallel computing acceleration equipment, and performing parallel computing acceleration on the sparse matrix. The method comprises the following steps: decomposing a general matrix into a sparse matrix in a set format, partitioning the sparse matrix and assigning a corresponding parallel computing hardware unit for the sparse matrix, reducing memory conflicts and access delay by using a shared memory, calling a high-level language API function library to execute loading, computing and storing processes of a computing data volume, and finally performing specification on output data of a function. Therefore, the final result of the random-precision sparse matrix multiplication and addition operation on the parallel computing acceleration equipment is obtained. According to the method, parallel computing resources such as the tensor core on the A100 can be fully utilized, the program calculation amount and the communication overhead are effectively reduced, and the program operation performance is greatly improved.

The invention discloses a self-adaptive data caching and synchronizing method, system and device for an edge gateway of the Internet of Things and a storage medium, and the method comprises the steps: obtaining network quality parameters, such as link bandwidth, time delay and packet loss rate, between nodes, detecting and solving instruction conflicts based on a vector clock mechanism, generating an instruction execution sequence, and carrying out the self-adaptive data caching and synchronizing of the edge gateway of the Internet of Things; local caching and incremental synchronization strategies are adopted for hotspot data, a data access mode is dynamically predicted in combination with a machine learning algorithm, and a cache updating period is optimized. The future data is predicted according to the data characteristics of the Internet of Things application, and the data cache updating period of each node is dynamically adjusted, so that each node can keep a relatively consistent updating rhythm when caching a hotspot data copy; therefore, the hot spot data copies of different nodes can be ensured to have similar accuracy and timeliness at the same time point, so that the data consistency is maintained.

The invention relates to the technical field of data synchronization, and discloses a distributed account book remote data synchronization method and system, and the method comprises the steps: carrying out the Hash mapping calculation of all nodes of a distributed account book system, and obtaining a distributed storage architecture; performing hierarchical network construction on the nodes, and setting an inter-region remote synchronization link and an intra-region local synchronization link to obtain a double-layer communication architecture; performing parallel compression coding, digital signature generation, transmission scheduling arrangement and consistency verification processing on the data blocks to obtain a node synchronous data set; calculating a correlation coefficient of node processing delay and network transmission delay to obtain a synchronous delay analysis model; based on a data block corresponding to the synchronous delay analysis model, adding a time sequence identifier and calculating an integrity check code, and generating a synchronous reliability guarantee mechanism; according to the method, the consistency and the traceability of the distributed account book data are ensured.

The invention relates to the technical field of knowledge questions and answers, in particular to a method and system for detecting and repairing large language model output illusion, which comprises the following steps of: dividing answer content into a plurality of sentences, calculating the confidence coefficient of each sentence, judging whether the confidence coefficient of each sentence exceeds a set threshold value or not, performing fact extraction on the sentences with the confidence coefficient lower than the set threshold value, obtaining triple data of each sentence; generating a subgraph A of each sentence based on the triple data of each sentence; extracting a sub-graph B from the local knowledge graph based on the subject in the triple data of each sentence; calculating the similarity between the sub-graph A and the sub-graph B, and if the similarity is smaller than a set threshold value, repairing the sub-graph A according to the sub-graph B to obtain a repaired sub-graph A; according to the triad data corresponding to the repaired subgraph A, obtaining a repaired sentence; and after all the sentences with the confidence lower than a set threshold value are repaired, final answer content is obtained. And the accuracy and reliability of model output are improved.

The invention relates to the technical field of data packet transmission, and discloses a data packet transmission control method and device for a low-power-consumption dual-mode Bluetooth chip, and the method comprises the steps: obtaining a signal strength parameter, a power consumption parameter, a transmission rate parameter and an environment parameter of the dual-mode Bluetooth chip, inputting the parameters into a deep time convolution network, and carrying out the time sequence feature extraction, obtaining a transmission mode switching judgment result; calculating transmission state evaluation data; the chip transmitting power is adjusted in a grading mode, standby power consumption is controlled through a nanoamp LDO circuit, and a dynamic power control parameter is obtained; performing compression coding on the data packet to obtain a dual-mode transmission protocol configuration parameter; executing time sequence sampling, and constructing a dual-mode state feature sequence; the data packet delivery success rate, the end-to-end transmission time delay and the total power consumption of the system are calculated, transmission control parameter optimization is carried out through double-layer deep reinforcement learning, target transmission control parameters are obtained, and the optimal balance of the transmission performance and the power consumption of the dual-mode Bluetooth chip is achieved.

The invention belongs to the technical field of model training, and particularly relates to a large model increment training method and system based on dynamic sparsification. The method comprises the following steps: S1, evaluating the parameter importance of a large model by adopting a multi-dimensional evaluation method, and reflecting the contribution of parameters to the performance of the large model; s2, performing dynamic sparsification control on the large model structure by adopting a self-adaptive structure adjustment method; s3, performing incremental training management on the large model by constructing a capacity-adjustable empirical playback buffer pool and proposing a multi-objective loss function; and S4, carrying out model compression optimization on the large model by adopting a mixed precision quantification strategy. The method has the characteristics that the network structure can be dynamically adjusted by evaluating the importance of model parameters in real time, so that the efficient training is realized while the model performance is ensured.

The invention provides a pattern multi-row matching method and device for ATE equipment, and the method comprises the steps: determining a cyclic pattern sequence comprising a plurality of rows of target patterns based on the number of matched rows when a first to-be-tested pattern carrying a LOOPMATCH instruction is recognized after a test is started; lOOPMATCH circulation is executed on the basis of the circulation Pattern sequence, and in each LOOPMATCH circulation, the Pattern matching result of the target Pattern in each row is determined; when all Pattern matching results pass in one cycle, jumping out of the LOOPMATCH cycle and determining that the cycle Pattern sequence test passes, otherwise, continuing to circulate to the maximum cycle number and then determining that the cycle Pattern sequence test does not pass; according to the technical scheme of the embodiment of the invention, when the first to-be-tested Pattern carrying the LOOPMATCH instruction is detected, multiple rows of target Patterns are constructed for loop testing, and LOOPMATCH loop is jumped out after any complete matching succeeds, so that the depth of the Pattern sequence can be effectively reduced, multi-row synchronous multi-time level matching is realized, the test resource overhead is reduced, and the test efficiency is improved.

The invention discloses a large language model training method and system based on elastic federal low-rank adaptive fine tuning, which adopts federal aggregation based on singular value decomposition (SVD) to reconstruct a local LoRA module of a client into a local increment matrix, then aggregates the local increment matrix into a global increment matrix, and generates a LoRA module of the next round through SVD, thereby effectively eliminating aggregation errors of traditional FedLoRA. According to the method, the LoRA rank is dynamically distributed for each weight matrix, and the balance between statistics and hardware efficiency is realized. In order to further reduce communication overhead and maintain model precision, a self-adaptive matrix freezing mechanism is introduced, and whether freezing is carried out or not is judged according to training stability of a weight matrix. According to the method, challenges of traditional FedLoRA in the aspects of training efficiency, communication overhead, model precision and the like are solved.

The invention discloses a two-way staggered assembly line parallel method for accelerating large model training, and compared with an existing method, the method introduces V-shaped staggered scheduling into a two-way assembly line to form a two-way staggered assembly line parallel mode, and improves the training efficiency of a large model and the use balance of a video memory. A V-shaped staggered assembly line and an inverted V-shaped staggered assembly line are fused to form a bidirectional staggered assembly line, the number of parallel devices is increased, meanwhile, the calculation time of a single micro-batch is shortened, the cavitation rate of the assembly line can be effectively reduced, and more balanced video memory occupation is achieved; in addition, a communication optimization strategy is introduced for overlapping or reducing part of communication overhead, and global gradient synchronization is performed by optimizing pipeline equipment mapping and utilizing a high-speed network in a server, so that the parallel training efficiency of the large model is further improved. The invention further provides a two-way interleaving assembly line parallel system and equipment which are used for achieving the two-way interleaving assembly line parallel method.

The invention discloses an in-tunnel construction vehicle meeting scheduling system based on positioning equipment. The system comprises a basic service module which is used for storing basic data for business service analysis in an acquisition and collection mode; the business service module is used for carrying out detailed analysis of the corresponding service through the data acquired by the basic service, and providing an analysis result to the display end for display; and the display end module is used for displaying the analysis result of the business service module, wherein the display modes comprise a mobile phone APP, a webpage large screen, an LED panel, a tunnel loudspeaker and a tunnel traffic light. The system carries out weight grading on vehicle and personnel positioning labels by standardizing a tunnel construction process, and integrates data of avoidance points of a transverse tunnel. And in combination with an intelligent technology, efficient vehicle meeting scheduling, intelligent construction path planning and avoidance point path planning of the construction area are realized.

The invention relates to an automatic switching method of a terminal and a satellite based on multi-attribute decision, and belongs to the technical field of satellite communication. The objective of the invention is to solve the problems of frequent switching and network performance reduction caused by high-speed movement of satellites in a 5G NTN low-orbit satellite communication scene. The method comprises the following steps: firstly, a terminal collects information of satellites in a visible range, including longitude and latitude, signal strength, serial numbers, channel resources and the like, and reports the information to a service satellite; and the service satellite calculates the remaining coverage time of the current satellite on the terminal according to the reported information, and judges whether to trigger switching. And if switching is triggered, the service satellite screens a candidate satellite set, quantitative evaluation is carried out according to attributes such as a satellite elevation angle, residual service time, the number of channels and switching time delay, and the satellite with the maximum comprehensive quantitative value is selected as a target satellite. And finally, the service satellite initiates a switching request to the target satellite and starts a switching process to complete switching between the terminal and the satellite.

The invention provides an in-orbit calculation autonomous system for a low-orbit satellite. A satellite system is managed in a mode of dividing satellites in a constellation into a plurality of autonomous groups, a unified management mechanism and an in-autonomous-group management mechanism are constructed to construct a management system of the satellite autonomous groups, and a streaming computing task migration scheme sensitive to a geographic position is constructed to construct an over-the-ground communication optimization scheme of the satellite system. Through a grouping mode, the low-orbit satellites can be divided into a plurality of autonomous groups, and the pressure of management nodes is reduced; a dynamic grouping mechanism is introduced during grouping, so that the balance of calculation load borne by each group in the whole operation period is ensured; a decentralized satellite cluster self-organization and cooperation mode is provided, so that the high availability of the cluster and the robustness of task processing are improved; according to the designed streaming computing task migration scheme sensitive to the geographic position, the expenditure caused by frequent migration of the on-satellite computing task is reduced from two aspects of migration path selection and migration mechanism implementation.

The invention discloses an ESG report text analysis method and system based on prompt, and the method comprises the steps: S1, extracting the content of each chapter in an ESG report, and obtaining a plurality of keywords of the content of each chapter based on the content of each chapter; s2, extracting substantive topics in the ESG report, calculating the similarity between each substantive topic and the keyword of each chapter, and screening the substantive topics based on the similarity; and S3, constructing a prompt template, inputting the screened substantive issue into the prompt template to obtain a cue word text, splicing the cue word text and the content of the corresponding chapter, and inputting the spliced cue word text and the content of the corresponding chapter into the large language model to obtain a text analysis result of the corresponding chapter of the ESG report.

The invention discloses a vehicle-mounted message middleware processing method and system and an intelligent cabin thereof. The method comprises the steps that an abstract interface layer used for achieving data type abstraction and decoupling is defined, a communication interface is created based on a hardware abstract layer, and cross-device communication and cross-process communication are conducted through the abstract interface layer; based on the communication interface, a Fast DDS is adopted in a communication protocol transmission layer to serve as message middleware, an abstract interface layer is packaged in an application layer, communication nodes are created in a communication interface layer according to a communication protocol and a QoS strategy, and the communication nodes are used for managing objects of a communication mode in a cross-process and cross-device mode and in-process communication. According to the vehicle-mounted message middleware provided by the invention, a user interface is simplified, and two main communication modes are supported, so that a system can flexibly select a communication mode according to an application scene.

The invention provides a data encryption method and device, a computer program product and a data encryption system. The method comprises the following steps: acquiring initial data and a network transmission rate; according to the network transmission rate, the initial data is segmented to obtain data to be transmitted, and the size of the network transmission rate and the size of a data segment are in positive correlation; performing multi-layer encryption on the to-be-transmitted data to obtain encrypted data; and sending the encrypted data to the target terminal. According to the scheme, the problems that a single-layer encryption method in the prior art is easy to crack and the data security is relatively low are solved.

The invention provides an HIVE grouping operation performance optimization and scheduling method based on key column partitioning, which relates to the technical field of data processing, and comprises the following steps: preprocessing data by constructing a key column distribution histogram and determining a key column partitioning strategy; optimizing data partitions based on resource evaluation and a dynamic programming algorithm; and calculating an optimal data transmission path by adopting a minimum spanning tree algorithm, and dynamically adjusting an allocation scheme. According to the method, HIVE grouping operation performance can be improved, dynamic balance of calculation loads is achieved, the problem of data skew is effectively solved, and large-scale data processing efficiency is improved.

The invention provides an image target removal method and apparatus, and an electronic device. The method comprises the steps of obtaining a to-be-processed image and a corresponding foreground target mask; inputting the to-be-processed image and the foreground target mask into a pre-training diffusion model; wherein the pre-training diffusion model comprises an auto-encoder module and a noise prediction module; encoding the to-be-processed image through an encoder in the self-encoder module to obtain an initial hidden variable corresponding to the to-be-processed image; performing forward diffusion noise adding processing on the initial hidden variable to obtain a noise-containing hidden variable; based on the foreground target mask, executing a reverse diffusion sampling process on the noise-containing hidden variable through a noise prediction module, and obtaining and correcting a self-attention map of each time step; and guiding a reverse diffusion sampling process based on the corrected self-attention map of each time step and a self-attention redirection guiding mode so as to realize target removal of the to-be-processed image. According to the invention, the target removal quality and stability can be improved.

The invention relates to the technical field of federated learning, and discloses a discriminator knowledge distillation personalized federated learning method based on a conditional generative adversarial network. The problems of appearance of massive participants and influence of non-independent identical distribution of data on federal learning data cost and personalized learning efficiency are solved; a discriminator knowledge distillation model fused with the conditional generative adversarial network is put forward to be embedded into a personalized federated learning framework; the differentiated feature representation of the client model is enhanced through the discriminator knowledge distillation model, so that the personalized model effectively recognizes and captures specific modes and features of individuals, the representation capability and the personalized effect of the model are enhanced, the higher model compression rate is achieved, and meanwhile, the user experience is improved. The conditional generative adversarial network fused with the discriminator generates synthetic data specific to each client, and the synthetic data improves the generalization ability of personalized federal learning and the adaptability of the model on the premise of fully ensuring the data privacy.

The invention provides an NPU instruction level context switching method and device, and relates to the technical field of data processing.The method comprises the steps that when a context switching instruction is received, a non-refresh instruction is forbidden, it is waited for completion of an instruction being executed, and context data is stored in a main storage module; after completion, recovering the context switching related instruction of the NPU module, and executing the target task; and after the target task is executed, recovering the context data to the NPU module, and recovering the execution of the non-refresh instruction. According to the method, optimization is carried out aiming at the problem that an existing NPU architecture generally can only support limited preemption at an algorithm boundary or a specific context switching point and even cannot preemption, so that a high-priority task must wait for a low-priority task to be completed, and remarkable priority inversion and / or key inversion are / is caused. A full-stack solution from a system on chip to an instruction set architecture and then to an operating system kernel is provided, and preemption of an NPU instruction level is achieved.

The invention relates to a multi-back-end separated memory system and an optimization control method thereof, the multi-back-end separated memory system comprises an intelligent far memory multi-back-end management control module and a multi-back-end far memory switch, the intelligent far memory multi-back-end management control module analyzes and processes calling of page errors and page switching triggering, and the multi-back-end far memory switch is used for switching the multi-back-end far memory switch. Switching instruction information and parameter regulation and control instruction information are obtained; and the multi-rear-end remote memory exchanger receives and executes the switching instruction information and the parameter regulation and control instruction information, runs applications on the separated memory architecture and the multi-rear-end remote memory software system, and releases resources after running is finished. By supporting a memory exchange strategy and a system architecture of a plurality of heterogeneous far-end separated memory rear ends, parallel access of a plurality of paths of far memories is realized to improve data throughput, application characteristics are analyzed, and intelligent rear end switching and data exchange parameter adjustment are realized; parallel access of a plurality of memory exchange back-end devices, fine-grained parameter configuration of a remote memory data access path, real-time switching of a plurality of memory exchange back-ends and an intelligent control and management method of the remote memory access path can be realized.

The invention provides a real-time energy-saving control method and device for a central air conditioner and a readable storage medium. The method comprises the steps that real-time position information of personnel in an area where the central air conditioner acts is obtained; according to the real-time position information of the personnel, the number of the personnel in each sub-area in the area where the central air conditioner acts is determined, and each sub-area correspondingly has at least one central air conditioner air outlet; central air conditioner control parameters of the corresponding sub-areas are determined according to the number of people in each sub-area; and the movement information of the personnel is obtained in real time, and the central air conditioner control parameters of all the sub-areas are adjusted in real time based on the movement information of the personnel. An intelligent air conditioner control strategy is adopted, the air speed of the air conditioner is intelligently adjusted according to the position of the person and motion prediction, the dynamic control method can optimize energy consumption, and meanwhile the comfort degree of the environment is kept.

The invention relates to a real-time optimization control method for a central air-conditioning system based on an agent model. The machine learning algorithm is adopted to construct the agent model used for predicting the optimal control strategy result, the real-time operation condition is input into the agent model in the actual application process, the optimal control strategy result under the operation condition is predicted, and compared with a traditional control method based on model optimization, the method has the advantages that the operation efficiency is improved, and the cost is reduced. According to the method, the calculation complexity is reduced, the calculation efficiency of the optimal control strategy is remarkably improved, and the real-time optimization control requirement of the central air conditioning system is met. The mathematical optimization model of the central air-conditioning system is established based on the equipment composition and the topological structure of the target central air-conditioning system as well as the optimization target and the constraint condition during model training, so that a large number of optimal control strategies corresponding to the operation condition sample are generated through the randomly generated operation condition; and forming a training data set for agent model training, so that the trained agent model can accurately predict an optimal control strategy.

The present invention discloses a dynamic low-overhead traffic obfuscation method based on data packet length distribution, and belongs to the technical field of defense against encrypted traffic fingerprint detection. The problem of insufficient dynamics and high resource overhead of the traditional traffic obfuscation method in the prior art is solved; the present invention combines the distribution-based data packet length mapping method and the segmentation and stacking-based data packet modification method, and finally proposes a dynamic low-overhead traffic obfuscation method based on data packet length distribution. For each data packet in a data packet sequence, the target data packet length is first obtained by the distribution-based data packet length mapping method, and then the data packet is modified to the target length by the segmentation and stacking-based data packet modification method, and finally the obfuscated data packet sequence is obtained. The present invention effectively avoids the operation of introducing padding data in the data packet modification process, reduces the additional bandwidth overhead, and can be applied to obfuscate traffic in an actual network environment.

The invention discloses an image restoration method and system based on gated convolution and a fused gated attention mechanism, and belongs to the field of image super-resolution restoration frameworks, and the method comprises the steps: taking a preprocessed image as the input of an image restoration model, carrying out the feature extraction and dimension raising, and obtaining a feature map; the image features are extracted by fusing the attention and self-attention mechanism of the convolutional neural network type, and iterative optimization is carried out by using the loss function, so that the local and global features of the image can be processed more effectively, the image reconstruction quality is improved, and the authenticity and detail richness of the high-resolution image are ensured at the same time.

The invention relates to an optical network routing intelligent planning method and system based on multi-objective optimization. The method comprises the following steps: performing recursive partitioning on a target optical network to obtain a plurality of sub-network units and a boundary node set; inputting the plurality of sub-network units into a three-layer deep learning model for processing to obtain a routing decision matrix and a wavelength allocation vector; inputting the plurality of sub-network units and the boundary node set into a distributed controller for processing to obtain an error compensation matrix; based on the routing decision matrix and the wavelength allocation vector, performing dual strategy processing of time delay priority and hop count priority to obtain a time delay weighted evaluation function and a hop count weighted evaluation function; performing multi-target joint optimization to obtain an optimal routing scheme; and performing reconstruction cost calculation on the optical path set according to the error compensation matrix and the optimal routing scheme, and outputting a resource reconfiguration scheme. According to the implementation of the invention, the resource overhead in the network reconstruction process is reduced, and the overall operation efficiency of the network is improved.

The invention relates to the technical field of intelligent power grids, in particular to an inspection vehicle tracking and communication optimization method and system, and the method comprises the steps: constructing an RIS-ISAC-NOMA system framework oriented to an intelligent power grid application scene, converting a non-line-of-sight NLOS path into an equivalent virtual line-of-sight LOS path through a reconfigurable intelligent metasurface RIS in a non-line-of-sight NLOS scene, and carrying out the tracking and communication optimization of an inspection vehicle through the virtual line-of-sight LOS path. A propagation path of an ISAC signal is adjusted in real time, and a vehicle tracking and predicting method based on an EKF algorithm is designed, so that efficient vehicle real-time sensing and communication are realized; a multi-inspection vehicle power distribution optimization scheme is provided in combination with the NOMA technology, the communication rate and the resource utilization efficiency of the system are further improved, meanwhile, the simulation result also verifies the advantages of the proposed system in the aspects of communication performance and resource utilization, and the overall efficiency of the system can be remarkably improved while the pilot signal overhead is reduced.

The embodiment of the invention provides a knowledge question-answering system and a data processing method. The knowledge question-answering system comprises a multi-modal knowledge graph, a question analysis module, a knowledge query module and an answer generation module, the question analysis module can perform semantic analysis on a target question currently asked by a user based on a large language model, extract target entity information and target user intention in the target question, and generate a knowledge query result; the knowledge query module can construct target query logic according to the target entity information and the target user intention and query the multi-modal knowledge graph according to the target query logic to obtain a target prompt information list, and the answer generation module determines a target answer according to all attribute information in the target prompt information list to generate prompt words and sends the prompt words to the user. And based on the large language model, generating a target answer according to the target answer generation cue word and returning the target answer to the user. According to the knowledge question-answering system, the accuracy and timeliness of the system can be improved, and the generalization ability of the system can also be improved.