66 results about "Evidence mapping" patented technology

The invention discloses a judicial evidence atlas construction and contradiction detection method based on a graph neural network, and the method comprises the following steps: importing a multi-mode evidence through a judicial personnel terminal, adding a label, and transmitting the multi-mode evidence to a background server; the method comprises the following steps: carrying out storage and standardized preprocessing on multi-modal evidence uploaded by a judicial officer terminal through a background server to obtain a standardized evidence text set; performing evidence entity extraction on each piece of data in the standardized evidence text set to obtain an entity list; performing relation prediction on entities in the entity list to generate an entity-relation pair list; according to the entity-relation pair list, constructing an initial judicial evidence atlas; based on a graph neural network and a judicial evidence ontology library, optimizing the initial judicial evidence atlas to obtain an optimized judicial evidence atlas; and according to the optimized judicial evidence atlas, detecting dominant contradictions and invisible contradictions. According to the invention, the integration of multi-modal evidence is realized, and dominant and implicit contradictions can be effectively identified.

The invention discloses an operation interaction method and system applied to camera image editing, and relates to the technical field of image processing, and the method comprises the steps: obtaining a voice semantic heat map, a pointing intensity map, a touch confidence map and a gazing confidence map based on a multi-modal interaction data packet, and calculating an image feature matrix at the same time; fusing into a multi-modal evidence graph through a normalized scale; performing semantic segmentation according to the image feature matrix to obtain a semantic segmentation first draft and a pixel-by-pixel category confidence coefficient, and performing position correlation weighting on the pixel-by-pixel category confidence coefficient by taking the multi-modal evidence graph as a confidence coefficient modulation factor to generate a candidate object mask sequence; and performing highlight display on the candidate object mask sequence, and performing conflict resolution and priority rearrangement in combination with the multi-mode evidence graph to generate a target object mask. According to the method, deep fusion of the interaction intention and image segmentation is realized, the precision and consistency of candidate region detection are improved, and the stability of real-time rendering and the reliability of an editing result are improved.

The invention discloses a mineralization potential evaluation and target region optimization method and system based on machine learning, and relates to the technical field of mineral resource exploration. The method comprises the following steps: acquiring multi-dimensional data of a target area; preprocessing the multi-dimensional data to obtain preprocessed key data; a metallogenic system conceptual model is constructed, and a hydrothermal migration path and a metallogenic fluid pressure field of the key data are calculated through finite element simulation for visual display; generating an evidence layer based on the key data to obtain multi-dimensional evidence layer data; and constructing a mineralization potential evaluation model, and inputting the multi-dimensional evidence layer data into the mineralization potential evaluation model for prediction to obtain a mineralization probability prediction result. Through multi-source data fusion and machine learning algorithm innovation, intelligentization and precision of mineralization potential evaluation and target region optimization are realized, and an efficient technical means is provided for mineral resource exploration.

The invention discloses a multi-mode large model police file and image association reasoning method, and belongs to the technical field of artificial intelligence and police information processing. The method comprises the following steps: preprocessing police affair texts and images; respectively extracting word-level and sentence-level features and local and global visual features by using a text encoder and an image encoder; realizing alignment and joint representation through a cross-modal fusion layer and an attention mechanism; generating a preliminary reasoning result by adopting a task output head; a heterogeneous evidence graph is constructed, and priori knowledge is injected in combination with the knowledge graph; and finally, joint reasoning is carried out through the graph neural network, and a cross-modal reasoning result and a complete evidence chain are output. According to the method, deep correlation analysis of police affair texts and images is realized, and the accuracy and interpretability of case research and judgment are improved.

The invention relates to the field of legal science and technology and generative artificial intelligence, and discloses a legal service project three-party agent pre-auditing arrangement and double-threshold gating method and system based on a large language model, and a storage medium. A platform, party and law institution three-party agent collaborative architecture is constructed, and role recognition, evidence structuring, law mapping, intelligent evaluation and document generation are completed through a chat mechanism. The method comprises the following steps: judging a party role based on LLM and generating an authorization certificate; demand fingerprints and evidence maps are extracted under platform arrangement, and legal mapping and reference chains are generated; hierarchical visible control is realized through a double-threshold scoring function; generating a neutral summary, a strategy tree and a document first draft through combination; and finally, outputting a desensitization packaging result and completing auditing anchoring. According to the method, the compliance, comparability and auditing performance of the project establishment stage are effectively improved, and the method is suitable for multiple types of legal service pre-docking scenes.

The invention discloses a multi-modal evidence chain management and control method and device, computer equipment and a medium, and the method comprises the steps: carrying out the standardization processing of multi-modal data, and obtaining standardized data, the multi-modal data comprising image data, video data and text data; performing feature extraction and low-rank projection according to the standardized data to obtain unified feature representation; constructing a causal evidence graph based on unified feature representation to obtain a causal relationship network; performing causal enhanced retrieval on the causal relationship network to obtain a retrieval result; performing limited decoding and attention consistency control by adopting a retrieval result to obtain a generated content; and generating a structured evidence chain record based on the generated content and the retrieval result, and taking the structured evidence chain record as tracing output information. According to the invention, the quality and credibility of the evidence chain are improved, and the generation controllability and the resource utilization rate are enhanced.

The invention discloses a data-mechanism-knowledge combined driving dam safety state self-diagnosis method and system, and the method comprises the steps: obtaining dam operation state data, extracting a state quantity based on a pre-constructed computable evidence graph, and generating a monitoring evidence; performing inversion correction on mechanism model parameters under the priori constraint of the atlas, and generating mechanism evidence; the monitoring evidence and the mechanism evidence are accessed to a probability graph model for joint inference, and arbitration correction is carried out when a conflict triggering condition is met; and carrying out credibility calibration on the diagnosis result by adopting a conformal prediction method based on the historical calibration sample set, and outputting a security state diagnosis conclusion containing a confidence interval. According to the method, the problems of multi-source evidence conflict, mechanism parameter distortion and low diagnosis result credibility are effectively solved, and credible and automatic diagnosis of dam safety is realized.

The application discloses a dynamic prototype multi-model medical image classification method based on feature credibility evaluation, and relates to the technical field of medical image processing.The method effectively solves the problems in traditional medical image subtype classification, such as lack of high-quality labeled data, high uncertainty of pseudo-labels, and difficulty of static prototypes in adapting to dynamic changes of lesions, and the like.Through construction of a multi-structure feature extraction network and completion of hierarchical feature fusion, the method combines feature cross learning, local attention modeling and expanded convolution to supplement context information, and strengthens semantic consistency and structural continuity of lesion region features.Meanwhile, the method constructs a spatial similarity graph through cosine similarity, and models and fuses a spatial uncertainty graph with any and cognitive uncertainty based on a Dirichlet distribution, obtains a reliable evidence graph through exponential fusion, and generates a pseudo-label with sample-level confidence, so that effective supervision information is accurately screened from a feature level, cumulative deviation of false pseudo-labels is greatly reduced, and stability of a semi-supervised learning process is improved.

The invention discloses a data decision generation system, method and device and electronic equipment. The system comprises a processing module, an evidence graph construction and calculation module, a multi-target decision module and a verification module. The processing module processes the obtained multi-source data to obtain an evidence unit set; the evidence graph construction and calculation module constructs an evidence graph according to the evidence unit set, and calculates the consistency score of each evidence unit in the evidence graph; the multi-target decision module generates at least two candidate decisions based on the evidence graph and the consistency score of each evidence unit, and obtains a target decision according to a rearrangement result of the at least two candidate decisions; the verification module updates the evidence unit set according to the verification result under the condition that the verification result of the target decision does not meet the preset verification threshold value, and reprocesses the multi-source data until the verification result of the target decision meets the preset verification threshold value; the integration of multi-source data is realized, and the credibility and the practical applicability of the target decision are improved.

This application discloses a temporal role recognition method based on knowledge graphs, relating to the field of data processing technology. It includes: acquiring first text data, second temporal record data, and third temporal event data of a target object; constructing event units for the target object and dividing them into first, second, and third temporal stages; extracting behavioral features from the first text data and mapping them to event nodes, extracting capability features from the second temporal record data and connecting them to basis nodes, and connecting the event nodes to result nodes to construct a role evidence graph; extracting continuous evidence chains and converting them into stage observation sequences; inputting the sequence inference model to generate role state trajectories, determining the role category based on the trajectory's continuity across the three stages, and outputting classification labels. This method transforms role recognition into a staged inference process based on temporal evidence chains, effectively distinguishing between continuous dominance and partial participation, thus improving the objectivity and accuracy of recognition.

This invention discloses an AI vision-based method for fire door status recognition, comprising the following steps: collecting and preprocessing data to obtain standardized visual data; extracting structural visual features of the fire door; performing evidence mapping processing to convert it into structural evidence and constructing state support parameters; constructing and correcting structural constraint relationships to obtain constraint-corrected state support parameters; inputting the structural visual features into a deep random vector functional connection network and updating the state support parameters; constructing an evidence set and calculating the degree of constraint violation to obtain evidence conflict measurement results; performing reliability assessment to obtain weight parameters corresponding to each structural piece of evidence; constructing support channels respectively, and incorporating the evidence conflict measurement results for constraint suppression processing to obtain a fire door status judgment value; and outputting the fire door status result. This invention employs an AI vision-based state recognition method based on structural constraint relationships to achieve fire door status determination with high accuracy and robustness.

The invention belongs to the technical field of literature processing, particularly relates to a literature element extraction method and system, electronic equipment and a storage medium, and aims to solve the problems that related technical knowledge updating is lagged, model illusion is difficult to restrain, and cross-modal semantic understanding is shallow. The method comprises the steps of obtaining a target literature; carrying out analysis processing on the target literature by utilizing a pre-trained multi-modal model so as to identify literature elements in the target literature; constructing an evidence graph based on the identified literature elements, and determining a target prompt word in combination with the evidence graph and a preset literature element ontology; according to the target cue word, executing a literature element extraction task by utilizing a pre-trained element extraction model, and extracting a target element from the target literature; and verifying the target element by using the evidence graph, and outputting the target element meeting the verification. According to the method, illusion output of a large model can be effectively identified and inhibited, and the accuracy of literature element extraction is remarkably improved.

The invention discloses an automatic medical record generation method and system based on multi-modal data and active perception, and belongs to the technical field of data processing.The method specifically comprises the steps that multi-modal original data of a patient and a doctor are collected, a field evidence graph is constructed, a constraint set is generated according to a clinical ontology and a time sequence rule, and configuring a priority and a dynamic threshold for each medical record field, generating an additional collection action based on information gain, and performing iterative update on the field evidence graph, the iterative update being to control the additional collection action of a target field until the target field reaches the dynamic threshold or meets a constraint set. Performing constraint decoding on medical record fields in the iteratively updated field evidence graph to obtain each target field value, and outputting a medical record with an evidence fingerprint chain; according to the method, the multi-modal information required by the medical record can be comprehensively covered, the information gap can be dynamically found and actively supplemented in the acquisition process, and the logic consistency and the time sequence reasonability between the fields are ensured.

The invention discloses a construction and evaluation method for an applicant drug clinical test quality management system, and aims to solve the problems that traceable evidence chains are lacked among applicant clinical test quality system elements, evidence materials and evaluation conclusions, evaluation is difficult to quantify, and closed-loop improvement based on evaluation results is difficult. According to the method, quality system element data, evidence material data and evidence source data are obtained and standardized, and a time heterogeneous evidence graph comprising element nodes, evidence nodes, version nodes, responsibility nodes and evidence source nodes is constructed; executing graph neural network representation learning on the evidence graph to generate a relationship likelihood parameter and an evidence credibility likelihood parameter; constructing a Bayesian factor graph containing conflict factors, and executing belief propagation to obtain element compliance posterior distribution and evidence credibility posterior distribution; and further calculating a compliance score, an uncertainty score and a coverage rate score to form a three-dimensional score result, and extracting an evidence chain to generate traceable output, thereby achieving the technical effects of automatic establishment of the evidence chain, compliance quantitative evaluation, conflict evidence suppression and evaluation result driven closed-loop improvement.

The invention discloses a full life cycle intelligent supervision method and system oriented to somatic cell clinical research, and relates to the technical field of somatic cell clinical research supervision. Based on the associated evidence atlas, intelligent auditing is achieved, and an interpretable report is generated; an in-process real-time supervision stage: real-time acquisition and verification of physical parameters, operation specifications and follow-up visit data are realized through an Internet of Things sensor, computer vision and an intelligent terminal; and a post-event intelligent auditing and tracing stage: integrating full-process data based on the unique ID of the project to form a digital file, mining compliance defects by using a deep learning model, and realizing credible tracing through a block chain. According to the invention, the problems of high manual dependence, blind monitoring areas, difficulty in tracing and full-flow splitting in traditional supervision are solved, automatic, intelligent and traceable supervision of the whole life cycle of somatic cell clinical research is realized, and the supervision efficiency and accuracy are improved.

The application discloses an online legal automatic question-answering method and system based on semantic matching, relates to the field of online legal automatic question-answering, and comprises the following steps: receiving a user legal consultation question and preprocessing; generating legal evidence basic data; generating user legal question element data; inputting a legal element initialization layer, generating a question-driven activation gate and a multi-layer heterogeneous legal evidence graph; inputting an element difference perception propagation layer, generating a difference evidence message; inputting a recent case path injection layer, generating a recent case path representation; and inputting a question-answering credibility output layer, generating a legal automatic question-answering result. The application utilizes evidence path enhancement matching to realize legal question-answering verification generation, and has the advantages of accuracy, professionalism and credibility.

This invention relates to a method, device, and medium for retrieving problem-related subgraphs based on evidence graph patterns. It retrieves a small-scale evidence subgraph from a graph-structured knowledge base that provides factual support for solving natural language problems. First, the natural language problem is transformed into a vector representation using an association resource pair retrieval model, and highly relevant candidates are retrieved from an offline-built association resource pair vector library. Then, all possible evidence graph patterns are enumerated based on the association resource pairs. Finally, all evidence graph patterns are uniformly ranked using a ranking model, and the top-ranked patterns are used to derive the problem-related subgraph. This invention transforms the retrieval of problem-related subgraphs into the retrieval of a small number of evidence graph patterns, avoiding the pruning overhead caused by iterative subgraph expansion with pruning on large-scale graph data, and effectively improving subgraph retrieval performance for multi-hop problems.

The invention relates to the technical field of teaching methods, in particular to a collaborative teaching method and system based on deep fusion of a knowledge base and an AI large model, and the method comprises the steps: obtaining a current teaching demand, and generating an executable constraint set according to the current teaching demand based on a course target constraint language; according to the executable constraint set, cutting is carried out on an evidence graph to generate a minimum closed subgraph, and the evidence graph is generated based on a knowledge base; according to the minimum closed subgraph, calling an AI large model through limited decoding to generate an assisted teaching draft; and judging whether the assistant teaching draft accords with a provincial condition, if so, generating credible teaching content, and carrying out teaching according to the credible teaching content. The problems of inaccurate retrieval results and wrong versions in the prior art can be solved.

The application discloses a kind of multi-agent cooperation methods based on retrieval enhancement's knowledge integration, specifically related to compliance audit question and answer technical field, document analysis agent splits compliance document into clause fragment and generates clause positioning mark, constructs knowledge unit and calculates evidence qualified score, and stores into knowledge unit library;Inquiry analysis agent forms structured inquiry representation, and retrieval agent retrieves accordingly and generates evidence package under the filter rearrangement of evidence qualified score in audit mode;Reasoning agent constructs integrated evidence graph based on evidence package and generates conclusion unit, and outputs candidate reply with reference drop point;Consistency check agent calculates support score and contradiction score and determines auditable conclusion strength, and output control or trigger closed-loop rectification process is carried out to reply.

The application discloses a non-homogeneous token investment research Agent high-dimensional sparse evidence constraint conclusion generation method, which is executed by a processor and includes the following steps: collecting investment research data and converting the investment research data into layered evidence nodes; establishing a target strong identity anchor point based on chain identity information and contract identity information, and blocking entity merging when there is identity conflict and no mapping proof; constructing a layered sparse evidence graph, generating a support path set and a counter-evidence path set; performing typed closed path verification on weak event layer evidence nodes; determining the allowed output sentence strength according to the support path strength, the counter-evidence path strength and the strong counter-evidence condition; generating a constrained investment research text, and deleting, downgrading or rewriting an unauthorized factual statement through a declared five-tuple countercheck. The method can reduce the risks of entity mismatch, weak event determination and evidence expression strength unauthorized risk in the non-homogeneous token investment research scene.

This application relates to the field of image processing technology and provides a barcode recognition method and apparatus that integrates depth stripe perception and orientation alignment. The method includes: acquiring a test image containing a barcode; generating multiple physical feature evidence maps corresponding to the test image based on multi-scale features extracted from the stripe physical features of the barcode in the test image; fusing the multi-scale features and the multiple physical feature evidence maps to generate a fused evidence map; and performing instance-level quadrilateral regression geometric localization processing on the fused evidence map based on the dense distribution features of the evidence in the fused evidence map to obtain the barcode localization result of the test image. This method can directly perceive and utilize these stripe physical features of the barcode, enabling accurate, robust, and efficient geometric localization of the barcode, thereby eliminating reliance on general detectors and fundamentally improving the accuracy and efficiency of barcode recognition.

The application relates to the technical field of CAD drawing auditing, in particular to an industrial CAD drawing auditing method and system based on multi-agent cooperation. The method comprises the following steps: performing structured analysis and standardization processing based on an acquired industrial CAD drawing file; performing hybrid representation construction of graph elements-semantics based on the standard industrial CAD drawing; generating executable auditing tasks based on the hybrid representation; performing multi-agent task auditing and auditing problem summarization based on the auditing tasks; performing multi-agent collaborative arbitration and feedback correction based on the auditing problems; and the application introduces a task context construction and candidate problem evidence expression mechanism in the multi-agent professional auditing process, so that each auditing conclusion comprises an auditing target, an executing agent, a rule basis, an evidence graph element, a graph element relationship, a rule violation confidence and a severity level.

The invention discloses an aviation plug health assessment method based on evidence reasoning, and relates to the technical field of equipment health assessment, and the method comprises the steps: determining assessment indexes reflecting the health state of an aviation plug, and obtaining the data of the aviation plug under each assessment index; obtaining a reference value corresponding to each health level in each evaluation index, converting data of the aviation plug under each evaluation index into confidence distribution of the reference value based on the reference value, and forming evidence of each evaluation index; for each health level, fusing the evidence of each evaluation index by using an evidence reasoning algorithm to obtain the evidence of each input evaluation index, and mapping the evidence into the confidence of output relative to each health level; and calculating a final utility value for representing the health state of the aviation plug according to the utility value of each health level and the confidence coefficient. According to the method, information generated in the electrical performance testing process and the mechanical performance testing process is comprehensively utilized, the uncertainty in the evaluation process is reduced, and the credibility of the evaluation result is improved.

The application discloses a false news detection method based on graph evidence interpretability, which comprises the following steps: collecting various news; expressing the false news detection task as a binary classification problem; constructing an evidence graph for the news in an offline manner; collecting the representation of historical news, and establishing positive and negative example evidence graphs; giving test news, and forwarding it to a multi-view feature extraction and learning module to obtain multi-view features; obtaining a matching score by referencing the evidence graph; then combining the multi-view features and the matching score and using them for prediction; searching for the predicted explanation by referencing the corresponding evidence graph; given a piece of news to be detected, comparing the news with the evidence graph to generate a matching score and related graph evidence, and predicting the news as false news or true news. The application utilizes the evidence graph constructed by historical labeled data; a new method for constructing an evidence graph from existing labeled data is proposed, which can be used to provide guidance and provide explicit explanations when new messages appear.

This invention discloses a method and apparatus for cervical lesion decision-making based on heterogeneous skill invocation and reinforcement learning. Its core lies in constructing a meta-controller driven by reinforcement learning and a decoupled heterogeneous intelligent agent skill library. The skill library encapsulates feature extraction and clinical logic into independent skills such as multi-instance learning visual perception, retrieval enhancement guide alignment, and safety backoff. The meta-controller models clinical reasoning as a Markov decision process, dynamically outputting skill scheduling instructions based on the slot state of the evidence graph and the model's cognitive uncertainty. This invention employs an action masking mechanism and a joint reward function including pathological consistency, multi-instance frame selection concentration, and invocation cost penalties to optimize the meta-controller's offline strategy, and introduces behavioral cloning constraints to inhibit out-of-distribution exploration. Finally, it synchronously outputs disease risk prediction and a traceable clinical evidence chain composed of structured skill invocation trajectories, achieving highly transparent and secure medical assistance decision-making.

The application discloses a nursing education traceable, interpretable and verifiable large model intelligent teaching assistant system and method, which comprises an input capture and semantic analysis module, a multi-source knowledge retrieval and versioned traceability module, an executable reference and piece-by-piece binding module, a causal evidence graph and structured explanation generation module, an evidence consistency and counterfactual stability verification module, and a replayable decoding and evidence storage module; the input analysis module generates a standardized problem template and risk level information; the knowledge retrieval and traceability module outputs a candidate evidence set with source segment fingerprints; the information binding module binds answer units and evidence fingerprints to generate reference mapping; the causal evidence graph module constructs a structured explanation graph; the multi-dimensional verification module performs logical consistency and counterfactual stability tests and outputs verification results; and the decoding and evidence storage module stores replayable decoding information. The application realizes answer segment level traceability, structured machine verifiable explanation and deterministic result replay through the construction of a modular system.

An evaluation system, method and program product for multi-view evaluation. The system acquires multi-view original data of the same evaluation object, aligns the data according to object identification, time identification and task identification, and forms a time sequence multi-view data set; generates a quality vector, an uncertainty vector and a view representation based on the data set, constructs a cross-time multi-view evidence graph, and calculates a credible weight; performs consistency propagation and conflict suppression on the view representation based on the evidence graph and the uncertainty vector to obtain a modified representation; generates an intervention variable and an intervention response by taking the modified representation as a variable node of a structural causal model; calculates a fusion evaluation score based on the credible weight, the modified representation and the intervention response, and calculates a conflict contribution value according to the fusion evaluation score and the intervention response; when the conflict contribution value exceeds a threshold, the modified representation, the credible weight and the edge weight of the evidence graph are updated jointly, and the updated fusion evaluation score and the explainable information are output. The present application can improve the accuracy, robustness and explainability of the evaluation.

The invention discloses a complex threat intelligence analysis and logical reasoning method based on an intelligent agent, and belongs to the technical field of intelligent agent intelligence processing. The objective of the invention is to solve the problem of accurate and explainable analysis of complex threat intelligence. The method comprises the following steps: performing question decomposition on a question input by a user based on a multi-strategy decomposition mechanism to obtain decomposed sub-questions; adopting self-adaptive hybrid retrieval and information refining to obtain preliminarily screened candidate documents; constructing a hybrid reordering model based on a spectrogram theory and a determinant point process, including construction of a hybrid semantic weighted graph, adaptive spectrum embedding and noise reduction and greedy k-DPP selection based on a spectrum space, and reordering the preliminarily screened candidate documents to obtain topological data of the reordered hybrid semantic weighted graph; and reasoning and generating a final answer based on the constructed global evidence graph. According to the method, logic coherence of outputting complex threat intelligence is realized, and a final answer which is accurate in reference and has a global view angle is quoted.

This invention discloses a method and system for intelligent monitoring of the entire clinical trial cycle based on multidimensional data fusion. To address the problems of high false alarm rates, lack of interpretability, and difficulty in forming a monitoring closed loop under multidimensional evidence fusion conditions, this invention acquires and standardizes multidimensional clinical trial data and its source data, constructs an evidence layer graph and a source layer graph, and establishes cross-layer associations to form a two-layer evidence graph. Based on source features, the source credibility is calculated and assigned to evidence layer nodes and edges. Source credibility gating is introduced into the heterogeneous graph Transformer attention mechanism for risk inference and risk score output. The risk score is calibrated for confidence and grouped by trial stage and research center. Group-based shape-preserving prediction is used, and the group warning threshold is determined based on the target false alarm rate to output warning labels and risk levels. The minimum sufficient evidence chain is extracted from the triggering warning object for interpretation. Under the condition of frozen model weights, feedback is used to update the credibility parameters, calibration parameters, and thresholds to form a closed loop, achieving the technical effects of low false alarms, interpretable risk warning grading, and a closed-loop intelligent monitoring throughout the entire cycle.

The present application relates to a fraud risk prevention and control method and system, wherein the method inputs multi-source heterogeneous data into a pre-trained channel risk assessment model for channel risk assessment, splices the generated channel risk assessment result with the individual feature vector of the user, inputs the obtained risk portrait feature into a Gaussian mixture model for unsupervised clustering to obtain K potential fraud scenarios and a probability vector of each user belonging to each potential fraud scenario. Based on the real-time context feature vector of all users, the risk portrait feature and the probability vector of each user belonging to each potential fraud scenario, fraud risk assessment is carried out, a heterogeneous evidence graph is constructed for fraud reasoning to generate a fraud risk list, and the risk prevention and control feedback data is used for reverse optimization to realize dynamic risk prevention and control. Thus, the present application can quickly respond to the changing new fraud scenarios while realizing the source perception of fraud, and realize accurate fraud risk prevention and control.