51 results about "Reference analysis" patented technology

The invention provides a method and system based on NLP file analysis, and relates to the technical field of natural language processing. According to the method, time and identifier unification and format and character set standardization are carried out on the multi-source file, layout segmentation, table structure extraction, reference analysis, term standardization and anaphora resolution are combined, semantic representation is constructed, a hierarchical index and a unique traceability identifier are generated, intention recognition, retrieval sorting, incremental updating and consistency verification are supported, and the method is suitable for large-scale popularization and application. Unification, semantization and traceability of the file analysis process are achieved, and the processing efficiency and accuracy are improved.

The invention discloses a phishing document de-obfuscation and feature extraction method and an application thereof in attack detection. The de-obfuscation comprises the steps of obtaining an obfuscation macro code of a phishing document, constructing a prompt engineering template by utilizing a pre-training language model, analyzing an obfuscation logic structure and generating a de-obfuscation rule and a reduction strategy; the method comprises the following steps: structuring a confused macro code into an abstract syntax tree through an analysis tool, matching a typical confusion mode based on a regular expression, and performing simulation and cell reference analysis in combination with a function to realize structure preliminary reduction, control flow semantic reduction and operation path construction; according to the unmixing rule and the abstract syntax tree, the confusion structure is converted into a readable macro statement, a macro code instruction sequence with confusion semantics removed is generated, and a semantic sequence obtained after unmixing is output. The feature extraction comprises word feature extraction, Token feature extraction, abstract syntax tree feature extraction and relation feature extraction. According to the method, the bottleneck that traditional phishing attack detection and confusion documents are difficult to recognize is broken through, and the detection accuracy of phishing document attacks is improved.

The invention provides a multi-source data fusion analysis system for electronic commerce digitization insight, and relates to the technical field of data processing, and the system comprises a reference analysis unit used for selecting a reference analysis unit in a spatial relation structure, calculating fractal dimension features of feature vectors, carrying out multi-level recursive division on the analysis unit based on fractal dimension values, and carrying out multi-level data fusion analysis on the fractal dimension values; each level of division keeps a self-similarity characteristic, and feature partitions are obtained; the partition space distribution module is used for mapping the feature vector distribution to corresponding feature partitions and calculating partition space distribution weights based on vector density distribution and space position relations of the feature partitions; the value evaluation module is used for performing parameter adjustment on the deep learning network by using the partition space distribution weight to obtain a value evaluation result; and the value evaluation module is used for processing a value evaluation result through a multi-modal fusion algorithm, generating a personalized recommendation strategy based on a reinforcement learning framework, and calculating resources and an intelligent routing algorithm so as to complete a digital insight analysis process. The e-commerce data utilization efficiency and the decision-making effect can be improved.

The invention belongs to the technical field of information, and particularly relates to an event co-reference resolution method based on multiple agents. The method specifically comprises the following steps: splicing prompt words of an event attribute recognition agent with a chapter text to be subjected to event co-reference analysis, and then inputting the event attribute recognition agent; the event attribute recognition agent follows the input content output event type and role or the output is null; when the output of the event attribute recognition agent is null, performing question query on the exogenous information integration agent; after a return result of the exogenous information integration agent is added, the event attribute identification agent generates a corresponding event type and role again; splicing the event type, the role and the chapter text, and inputting an event co-reference prediction agent; and the event co-reference prediction agent realizes event co-reference resolution for each event mention according to the sequence.

The invention relates to the technical field of data processing, can be applied to business scenes of battery management, financial science and technology, medical health and the like, and discloses a battery unit anomaly detection method, device, equipment and medium, and the method comprises the steps: obtaining static attribute data and operation state parameters of a plurality of battery units, and generating a homogenized battery unit group through dynamic clustering; performing dynamic reference analysis by using the internal resistance data, identifying abnormal battery units and generating a preliminary abnormal battery unit set; classifying the abnormal battery units by using a time sequence analysis model, and generating and confirming an abnormal battery unit list; and generating a battery unit maintenance schedule based on the confirmation list and a preset maintenance knowledge base. According to the invention, through combination of the dynamic reference analysis model and the time sequence analysis model, the abnormal state of the battery unit is accurately identified, a potential problem battery can be found in time, the efficiency and accuracy of a battery monitoring system are improved, and the reliability of the battery at a critical moment is guaranteed.

The invention relates to the technical field of education, and discloses an online learning environment-oriented education interaction behavior recognition system, which is characterized in that a standardized cognitive model is constructed offline based on interaction data of a high-performance learning group; in the online stage, interaction events of individual learners are obtained in real time, high-dimensional cognitive trajectories representing cognitive states of the learners are generated, a trajectory dynamics analysis module executes double reference analysis, and internal dynamics indexes representing stability of the trajectories are obtained by calculating the maximum Lyapunov index, recognizing phase change points and the like; the external deviation measure of the difference between the trajectory and the normalized cognitive model is obtained by calculating the state deviation degree, the path efficiency and the like, and finally, the cognitive mode diagnosis module integrates the indexes and generates a structured diagnosis description according to a rule base. According to the method, through combination of internal dynamics and an external reference system, deep quantification and fine identification of the cognitive process of the learner can be realized, and an objective basis is provided for adaptive intervention.

The invention provides a cross-file semantic association analysis system based on a hierarchical rule engine. The cross-file semantic association analysis system comprises a plurality of modules, wherein a code compilation analysis module is used for converting an input code file into an AST object and executing cross-file dependency and reference analysis; the semantic fact construction module generates an initial fact set according to the data output by the code compiling and analyzing module; the rule interpreter receives the rule set of the hierarchical rule base and constructs a corresponding hierarchical rule network; the rule matcher injects the initial fact set into the hierarchical rule network in a streaming mode to execute rule matching; the condition verification module verifies a rule matching condition according to the input fact data and returns a result; when rule matching succeeds, the rule executor executes corresponding processing logic and updates the fact base; and the result analyzing and processing module analyzes and integrates the final reasoning fact data and writes a result into a conclusion library. The method can be used for constructing a universal cross-file code context analysis system, and the code static analysis performance is improved.

The embodiment of the invention discloses a key anaphora analysis method and device in voice-to-text conversion, a medium and a product. The method comprises the following steps: constructing an interpersonal relationship knowledge graph of a user and splitting according to an interpersonal relationship type; when the text sequence of the user voice conversion contains the third-person pronouns and meets a preset condition, acquiring context information and acquiring anaphora objects of the third-person pronouns, a relationship type between the anaphora objects and the user and the confidence coefficient of the relationship type; sorting and traversing the relation sub-atlases based on the relation type and the confidence coefficient to obtain candidate relation sub-atlases; and when the reference object is successfully matched with any one of the sub-nodes in the candidate relation sub-graph and the first annotation fields of the sub-nodes, acquiring a character sequence containing a correct third-person pronoun based on the second annotation fields of the sub-nodes. According to the method, the anaphora objects of the third-person pronouns are obtained through semantic analysis, and the anaphora objects are matched in a multi-dimensional mode in combination with the interpersonal relationship knowledge graph so as to obtain the correct third-person pronouns.

The embodiment of the invention provides a story text processing method and device and a related product, and the method comprises the steps: obtaining a story text, and obtaining a plurality of text analysis dimensions corresponding to the story text; according to the content features of the story text, determining a first analysis dimension matched with the story text in the plurality of text analysis dimensions, and taking the text analysis dimensions except the first analysis dimension in the plurality of text analysis dimensions as second analysis dimensions; determining a first analysis mode of the story text in the first analysis dimension, and analyzing the story text according to the first analysis mode to obtain a first analysis result of the story text in the first analysis dimension; and analyzing the story text according to a reference analysis mode configured for the second analysis dimension to obtain a second analysis result of the story text in the second analysis dimension.

The invention discloses an ecological and economic multi-target collaborative decision-making system and method for comprehensive treatment engineering in the technical field of ecological environment treatment, and the system comprises a data collection module which is used for obtaining the historical data of carbon reserves and undetermined influence factors; the data screening module is used for acquiring model influence factors and dividing the model influence factors into governance regulation factors and non-regulation factors; the reference analysis module is used for obtaining a governance regulation factor reference value based on the carbon reserves and the species diversity index; the adjustment amplitude analysis module is used for setting the adjustment amplitude and determining the numerical change of a non-adjustment factor; the model construction module is used for obtaining a carbon reserve prediction model; and the carbon reserve assessment decision module is used for carbon reserve prediction and adjustment amplitude determination. According to the method, the ecological target and the economic target are jointly used as the basis of reference definition, the regulation and control scheme is generated based on the basis, and decision support with both ecological and economic benefits is provided for treatment engineering through modular cooperative operation.

The invention relates to the technical field of mineral exploration, and discloses an intelligent evaluation method and system for mineral resource exploration and a storage medium. The method comprises the steps of obtaining multi-source geological data of a to-be-evaluated area, dividing remote sensing physical data into first remote sensing physical data corresponding to a sampling point and second remote sensing physical data not corresponding to the sampling point, and constructing a preset model based on geological exploration data and the first remote sensing physical data, the model is used for estimating geological exploration data of a virtual sampling point, the estimated data and original data are combined to generate new multi-source geological data, a first geological space structure and a plurality of second geological space structures are respectively generated through two preset geological modeling methods, and the space structure closest to the first geological space structure is selected as a reference; according to the method, the reference geological space structure is analyzed, the mineral average abundance and the ore volume are calculated, then the mineral resource reserves are predicted and evaluated, the mineral resource evaluation accuracy and efficiency are improved, and the exploration cost is reduced.

The invention relates to the technical field of intelligent decision making, can be applied to business scenes such as financial science and technology and medical health, and discloses an analysis method, device and equipment based on distribution alignment and a medium, and the method comprises the steps: obtaining historical business data and reference analysis label data, constructing reference probability distribution, and training an analysis model to obtain an optimization model; and inputting the to-be-analyzed business data into the optimization model to generate target analysis probability distribution, calculating a statistical feature value and a confidence interval, and generating an analysis report containing abnormal prompt information based on the statistical feature value and the confidence interval. According to the method, distribution alignment is carried out on the model by referring to probability distribution, so that the model outputs probability distribution with group judgment features; providing result credibility expression through statistical feature values and confidence intervals; the interpretation and reliability of an analysis report are enhanced through abnormal prompt, and the stability and practicability of an intelligent analysis system in an uncertain scene are improved.

The present invention relates to the field of cable monitoring technology, and specifically to a cable health status assessment system and method based on edge computing, comprising: a cable data acquisition module, a monitoring reference analysis module, a trigger condition setting module and a health assessment management module, wherein the cable data acquisition module collects historical operation data and historical monitoring data of the cable; the monitoring reference analysis module uses an edge computing processor to preprocess the cable data, and the preprocessed data is composed of training samples and a monitoring reference model of the cable is established; the trigger condition setting module analyzes and sets trigger conditions for cable fracture monitoring based on the monitoring reference model; the health assessment management module monitors the cable for fracture and assesses the cable health status when the trigger conditions are met, and sends an alarm signal to the monitoring terminal when the cable health status is assessed to be abnormal, thereby reducing the monitoring cost and improving the cable monitoring efficiency while ensuring the safe operation of the cable.

The invention discloses a document-level relation extraction method and system based on knowledge enhancement and evidence guidance, and belongs to the technical field of natural language processing and information extraction. According to the invention, three core modules are mainly used for cooperative work: a document graph enhancement module is used for constructing a hierarchical heterogeneous graph and fusing co-reference analysis to enhance semantic representation; the knowledge enhancement module introduces an external knowledge base and adopts a confidence coefficient filtering mechanism to reduce knowledge noise; the evidence guidance reasoning module realizes multi-hop reasoning through axial attention and evidence supervision, and solves the technical problems of decentralized modeling of reasoning capability, large knowledge integration noise, insufficient evidence guidance and limited long-range dependence capture capability in the existing method. Experiments show that the method can effectively capture inter-sentence dependence, suppress knowledge noise and improve multi-hop reasoning stability, and can be widely applied to scenes such as knowledge graph construction, intelligent question and answer and information retrieval.

The invention provides a DFMEA analysis method and a computer program product, and the method comprises the steps: firstly determining an analysis main body and an analysis intention based on a failure question after obtaining the failure question related to a to-be-tested product; based on the analysis main body and the failure questions, a keyword query mode and a vector query mode are adopted to retrieve various reference analysis knowledge from a pre-constructed multi-source knowledge base; and finally, based on the analysis subject, the analysis intention and all reference analysis knowledge, calling a large language model, and generating a DFMEA analysis result corresponding to the analysis subject. The multi-source knowledge base is created on the basis of various heterogeneous data sources and comprises a large amount of analysis knowledge, so that reference analysis knowledge can be accurately searched from the multi-source knowledge base by adopting a keyword query mode and a vector query mode, and finally, a large language model is called to generate a DFMEA analysis result, so that DFMEA analysis is automatically performed, and the efficiency is improved. Manual analysis is not needed, and the analysis efficiency is improved.

A method and system for interaction analysis are disclosed. An example method for evaluation of an interaction between an analyte in a fluid sample and a ligand immobilized on a sensor surface of a biosensor includes providing a reference binding curve, representing a reference interaction for a predetermined acquisition cycle, by acquiring, using the biosensor, one or more binding curves for a reference-analyte ligand interaction at the predetermined acquisition conditions, acquiring, using the biosensor, a sample binding curve for the analyte ligand interaction for the predetermined acquisition cycle including at least one association phase wherein the sensor surface is put into contact with a fluid sample including analyte at a predetermined concentration, and generating a graphical user interface, including an upper threshold curve and a lower threshold curve defined with respect to the reference binding curve.

The present invention is provided with: a learning model holding unit (8) that holds a learning model created using learning model creation data, the learning model creation data includes a plurality of parameters of analysis data acquired by normal analysis, and a plurality of parameters of each of a plurality of analysis data acquired by analysis in a state in which a plurality of types of abnormalities are individually present or mixed. The learning model trends to learn the influence of each of the plurality of anomalies on the plurality of parameters; an analysis data holding unit (10) that holds reference analysis data obtained by normally analyzing a reference sample containing a specific component under specific analysis conditions, and analysis data to be diagnosed; the analysis data of the diagnosis object is obtained by analyzing a sample containing at least the specific component under the specific analysis conditions; and an abnormality estimation unit (12) configured to apply the plurality of parameters of each of the reference analysis data and the analysis data to be diagnosed to the learning model, thereby estimating an abnormality type existing in an analysis performed when the analysis data to be diagnosed is acquired.

The application discloses a kind of project intelligent review method and system based on data analysis, it is related to engineering project analysis management technical field, comprising: according to the data retrieval of database system to the project name to be reviewed.This application filters out multiple groups of relevant project historical data according to the key features of different stages of the project to be reviewed, then reviews multiple groups of relevant project historical data according to the review standards of different stages of the review project, obtains multiple groups of reference review scores, then analyzes the importance of different stages of the project to be reviewed in the whole stage, obtains the evaluation weight corresponding to different stages of the project to be reviewed, finally, calculates the reference review score and the evaluation weight corresponding to different stages of the project to be reviewed, determines the evaluation score of the project to be reviewed, determines the evaluation result of the project to be reviewed according to the evaluation score of the project to be reviewed, therefore, by reference analysis to the completed project with relevance, it avoids that review result appears larger error.

The invention provides an underwater robot navigation method and system combined with multi-modal data, and the method comprises the steps: firstly calling a pre-trained multi-modal navigation deep learning model to carry out navigation path reference analysis on multi-modal navigation data such as underwater acoustic detection data and underwater optical imaging data of an underwater robot operation area, and generating a navigation reference scheme; then, the underwater robot is driven to execute initial navigation operation according to the navigation reference scheme, and real-time multi-mode feedback data are collected; and inputting the feedback data into the model to carry out navigation deviation correction analysis, generating a correction instruction, adjusting navigation parameters, and then executing corrected navigation operation. And continuously collecting multi-modal monitoring data in an operation execution process after correction, and inputting the multi-modal monitoring data into the model for navigation state monitoring until the underwater robot reaches a target navigation position. The multi-modal data is fused, the navigation precision and adaptability are improved, the underwater environment change can be timely handled, and the navigation stability and safety are guaranteed.

The present invention relates to a communication system (140), a monitoring system (110) for in-situ monitoring of at least one substance (112) used in a gas scrubbing process, and related methods (310, 312), wherein the monitoring system (110) includes the communication system (140). The monitoring system (110) can be used to monitor the at least one substance (112) and provide processing data for processing the at least one substance (112) via the communication system (140). In this article, the communication system (140) includes a cloud server (144), a first server (146), at least one second server (148, 148') and at least one third server (150, 150'); wherein the first server (146) further has a first communication interface (156) configured to provide reference spectral information and reference analytical data related to at least one reference sample to the cloud server (144); wherein each second server (148, 148') has a second communication interface (158, 158') configured to provide spectral information related to at least one substance (112) to the cloud server (144); wherein the cloud server (144) is configured to: generate a calibration model by using the reference spectral information and reference analytical data related to at least one reference sample provided by the first server (146), wherein the calibration model includes at least one parameter; applying a calibration model to spectral information associated with at least one substance (112) provided by a second server (148, 148'), thereby extracting at least one value for the at least one parameter; providing at least one value for the at least one parameter to the first server (146) via a first communication interface (156); wherein the first server (146) is further configured to determine processing data by using the at least one value for the at least one parameter provided by the cloud server (144), wherein the processing data includes at least one piece of data associated with a recommended treatment of the at least one substance (112); wherein the first server (146) further has at least one third communication interface (160, 160'), wherein each third communication interface (160, 160') is configured to provide processing data to at least one third server (150, 150').

The invention relates to the technical field of safety detection of nuclear reactor equipment, in particular to an irradiation damage nondestructive detection method for a reactor pressure vessel and related equipment. The method comprises the following steps: carrying out material magnetic non-uniformity reference analysis before a reactor pressure vessel is put into operation to obtain corresponding original material magnetic distribution characteristics and a magnetic reference database, carrying out detection region division, and determining a plurality of vessel detection regions; performing local magnetization reference retest on each container detection area to calibrate the magnetic reference database; in the process that the reactor pressure vessel is put into operation, each vessel detection area is magnetized again, and corresponding target area magnetic field signals are collected; and according to the magnetic reference database and the target area magnetic field signal, performing damage evaluation on each container detection area to obtain a damage evaluation result. Therefore, the nondestructive testing and evaluation of the material irradiation damage of the reactor pressure vessel can be carried out, and the method has better accuracy performance.

The invention discloses a data reference representation method, which comprises the following steps of: acquiring a configuration file for describing a page component, and distributing an identifier for a reference object in the configuration file; in the configuration file, representing a data reference relationship in the configuration file by using a preset reference prefix and an identifier to obtain a reference expression; the preset reference prefix is used for distinguishing reference types; defining a reference mapping item in the configuration file; each reference mapping item records corresponding information between the identifier and the external file path. Correspondingly, the invention further discloses a data reference analysis method which comprises the following steps: acquiring the configuration file containing the reference expression; identifying the type of a reference prefix in the reference expression; determining a corresponding analysis strategy according to the type of the reference prefix; analyzing the reference expression into actual data based on an analysis strategy; and executing a data replacement operation, and replacing the reference expression with an analysis result.

The invention discloses an intelligent gas turbine regulation and control system based on multi-source operation reference analysis, and relates to the technical field of intelligent regulation and control. And the blade fracture risk prediction module analyzes the double-source abnormal value and calculates a risk coefficient, and if the risk is pre-judged, the airflow regulation and control module is triggered to execute a strategy of suddenly pressurizing to the maximum so as to blow away blockages, so that closed-loop intelligent regulation and control are formed. According to the method, the collaborative anomaly index is constructed through air pressure and vibration multi-source data fusion, and early-stage accurate early warning is realized in combination with time sequence trend analysis; after the risk is predicted, sudden pressurization unblocking is actively triggered, and passive shutdown is avoided; non-intrusive deployment and standardized risk coefficient calculation are adopted to adapt to multiple models; the risk is judged based on the continuous deterioration trend, instantaneous interference is filtered, and the interference immunity and decision reliability are improved.

The application relates to the technical field of mineral exploration, and discloses an intelligent evaluation method and system for mineral resource exploration and a storage medium. The method comprises the following steps: obtaining multi-source geological data of a region to be evaluated, dividing remote sensing physical data into first remote sensing physical data corresponding to sampling points and second remote sensing physical data not corresponding to the sampling points, constructing a preset model based on geological exploration data and the first remote sensing physical data, estimating geological exploration data of virtual sampling points by using the model, combining the estimated data with original data to generate new multi-source geological data, generating a first geological spatial structure and a plurality of second geological spatial structures by two preset geological modeling methods, selecting the closest one to the first geological spatial structure as a reference, analyzing the reference geological spatial structure, calculating the average abundance of minerals and the volume of ores, and then predicting the reserves of mineral resources and performing evaluation. The application improves the accuracy and efficiency of mineral resource evaluation and reduces the exploration cost.

This method enables the fully automated creation of a wiki from a text corpus. The lists of articles to be created are generated using either a vector-based or a purely frequency-based analysis. The actual text creation step always employs a RAG (Reference Analysis Group) method, in which the LLM (Licensed Life Sciences) uses a vector database to create well-founded, context-appropriate articles. This results in a scalable, well-structured, and linked knowledge system that can be used in various disciplines—from corporate and legal or medical documentation to large scientific archives.

The invention discloses a training data generation method and electronic equipment, and relates to the technical field of model data training, and through precise symbol tracking and cross-file reference analysis of a trained analysis tree model, generated structured data more accurately reflects the use condition and semantic relationship of symbols in a code library, so that the user experience is improved. And a solid foundation is provided for subsequently generated training data. Structured data and technical specification documents are associated, and the generated knowledge base contains official definitions and functional descriptions of code symbols, so that the understanding ability of a large model to underlying system codes can be greatly enhanced. Conditional compilation and multi-architecture support are considered in the construction of the global symbol table, so that the generated training data can cover different compilation configurations and architectures, and the generalization ability and practicability of the model are enhanced. Therefore, the problem that effective training data is difficult to generate in a complex scene in an existing scheme can be solved.

The invention relates to the technical field of printing process optimization, and discloses an intelligent printing process parameter optimization method and equipment, and the method comprises the steps: obtaining full-process monitoring data of printing equipment through a multi-dimensional sensing module, carrying out the process quality correlation analysis to obtain printing quality analysis characteristics, constructing quality influence factors and corresponding factor value information, and carrying out the analysis of the printing quality. A potential optimization strategy is obtained through genetic algorithm optimization, a feasible optimization strategy is screened out through physical simulation of a digital twinborn body of printing equipment, the feasible strategy is allocated to multiple pieces of printing equipment to be executed for collaborative reference analysis, an iterative optimization strategy is obtained, and intelligent optimization of printing process parameters is achieved. The printing quality and the production efficiency are improved, the manual intervention complexity and cost are reduced, the strategy collaboration and continuous optimization capacity between equipment is enhanced, and the problem that in the prior art, the effect of manually adjusting printing process parameters is poor is solved.

The invention relates to the technical field of image area detection, in particular to a tea wall breaking rate detection method, equipment and system. The method comprises the following steps: acquiring gray images of leaves at different angles after each batch of rolling, and initially quantifying the wall breaking condition through a pixel value difference area to obtain an initial wall breaking index; determining a suspected crack edge according to the edge direction consistency degree and the width smoothness condition, obtaining a wall breaking sufficient evaluation index according to the length position distribution uniformity condition of the suspected crack, and analyzing the boundary complexity to adjust the initial index to obtain an adjusted wall breaking index; and obtaining a confidence coefficient by adjusting index consistency and a historical batch similarity degree through different angles, and determining a final wall breaking rate based on the confidence coefficient. According to the method, wall breaking evaluation is adjusted by considering the influence of leaf veins and tearing characteristics, abnormal conditions caused by process drift are found in time through real-time detection and correlation historical process reference analysis after rolling, and stable quality of tea production is guaranteed while the accuracy of wall breaking evaluation is improved in factory production.

A transmission shaft life optimization method based on a digitalized generation structure comprises the steps that according to original geometry and material attributes, after the rigidity and fatigue life of an original shaft are obtained through finite element reference analysis, a preliminary CAD model is generated through neural network topology; and performing iterative finite element performance evaluation to obtain an optimization model of which the rigidity and the service life meet the requirements for manufacturing. According to the invention, through topology generation driven by the neural network and a multi-objective optimization strategy, rigidity and strength maintenance of the transmission shaft is realized, and structural parameters are automatically and intelligently regulated and controlled through a required life value input by a user, so that light weight is realized, and a test period is shortened.

The invention provides a video generation method and device, electronic equipment, a storage medium and a program product. The method comprises the following steps: in response to an acquisition operation for a reference material in a display interface, confirming a reference video and an initial prompt text related to the reference video; in response to an input operation for an input area in the display interface, displaying input content in the input area; wherein the input content comprises an input prompt text obtained based on the initial prompt text and a material identifier about a target object; performing multi-modal analysis on the reference video to obtain a reference analysis text describing a preset dimension of the reference video about a plurality of modals; generating a target video for the target object based on the reference analysis text, the input prompt text and an input material indicated by the material identifier; and the target video and the reference video have similarity in the preset dimension, so that the creation quality of the video can be improved.