Risk detection method and device and electronic equipment

Abstract

The embodiment of the invention discloses a risk detection method and device and electronic equipment. The method comprises the following steps: receiving a risk detection request for sharing and exchanging government affair data in a mesh network mode; in response to the risk detection request, determining to-be-detected mesh network shared exchange data corresponding to the risk detection request; determining risk information corresponding to the mesh network shared exchange data by using a pre-trained multi-modal large model according to the mesh network shared exchange data; wherein the multi-modal large model is obtained by training according to historical shared exchange data, historical risk information and neural perception data, and the historical risk information comprises risk information determined by a preset auditing party according to the historical shared exchange data; the neural perception data is used for representing brain neural activities of the preset auditing party when the historical risk information is determined; and according to the risk information, determining whether there is a risk in sharing exchange of the government affair data in the mesh network mode.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer technology, and in particular to a risk detection method and device and electronic equipment. BACKGROUND

[0002] With the gradual increase of government data development and utilization scenarios, and government data involving financial management, asset management, environmental facility management and other aspects of data, therefore, how to ensure the security of government data sharing and exchange in mesh mode has become the focus of public attention.

[0003] The sharing and exchange of government data in mesh mode can be detected by artificial risk detection according to the sharing and exchange characteristics of government data, but due to the characteristics of the sharing and exchange of government data in mesh mode, such as many-to-many data relationship and dynamic data flow path, the detection efficiency and detection effect of artificial risk detection are low, therefore, a technical solution is needed to improve the risk detection efficiency and accuracy of government data sharing and exchange in mesh mode to ensure the security of government data sharing and exchange in mesh mode. SUMMARY

[0004] The purpose of the embodiment of the present application is to provide a technical solution that can improve the risk detection efficiency and accuracy of government data sharing and exchange in mesh mode to ensure the security of government data sharing and exchange in mesh mode.

[0005] To solve the above technical problems, the embodiment of the present application is implemented as follows: In a first aspect, the present application provides a risk detection method, which comprises: receiving a risk detection request for sharing and exchanging government data in mesh mode; In response to the risk detection request, determining the mesh sharing and exchange data to be detected corresponding to the risk detection request; Using a pre-trained multi-modal large model, according to the mesh sharing and exchange data, determining the risk information corresponding to the mesh sharing and exchange data; wherein the multi-modal large model is trained according to historical sharing and exchange data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset audit party according to the historical sharing and exchange data, and the neural perception data is used to represent the brain neural activity of the preset audit party when determining the historical risk information; According to the risk information, determining whether there is a risk in the sharing and exchange of government data in mesh mode.

[0006] In a second aspect, the present application provides a risk detection device, which comprises: The request receiving module is configured to receive a risk detection request for sharing and exchanging government data in a mesh mode. The data determining module is configured to determine, in response to the risk detection request, mesh sharing and exchanging data to be detected corresponding to the risk detection request. The information determining module is configured to determine, by using a pre-trained multi-modal large model, risk information corresponding to the mesh sharing and exchanging data according to the mesh sharing and exchanging data; wherein the multi-modal large model is obtained by training according to historical sharing and exchanging data, historical risk information, and neural perception data, the historical risk information includes risk information determined by a preset auditing party according to the historical sharing and exchanging data, and the neural perception data is used to represent brain neural activity of the preset auditing party when determining the historical risk information. The risk detection module is configured to determine, according to the risk information, whether there is a risk in sharing and exchanging the government data in the mesh mode.

[0007] In a third aspect, an electronic device is provided, which includes a processor, a memory, and a computer program stored in the memory and executable on the processor, and when the computer program is executed by the processor, the steps of the risk detection method provided in the above embodiments are implemented.

[0008] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the risk detection method provided in the above embodiments are implemented.

[0009] In a fifth aspect, a computer program product is provided, which includes a computer program, and when the computer program is executed by a processor, the steps of the risk detection method provided in the above embodiments are implemented. BRIEF DESCRIPTION OF DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0011] Figure 1 It is a flowchart of a risk detection method of the present application; Figure 2 It is a flowchart of a process of obtaining historical risk information and neural perception data of the present application; Figure 3A flowchart of a training process of a multi-modal large model of the present application; Figure 4 A schematic diagram of a risk detection process of the present application; Figure 5 A structural schematic diagram of a risk detection device of the present application; Figure 6 A structural schematic diagram of an electronic device of the present application. DETAILED DESCRIPTION

[0012] The embodiments of the present application provide a risk detection method, device and electronic equipment.

[0013] In order to enable persons skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor should belong to the protection scope of the present application.

[0014] Embodiments of the present specification provide a risk detection method, device and equipment. At present, the security of government data sharing and exchange in a mesh mode has become the focus of public attention. Generally, risk detection of government data sharing and exchange in a mesh mode can be performed manually according to the sharing and exchange characteristics of government data. However, due to the characteristics of the sharing and exchange of government data in a mesh mode, such as a many-to-many data relationship and a dynamically changing data flow path, the manual risk detection method has low detection efficiency and detection effect. In order to improve the detection efficiency and detection effect, in the present scheme, a risk detection request for the sharing and exchange of government data in a mesh mode is received, and in response to the risk detection request, the mesh sharing and exchange data to be detected corresponding to the risk detection request is determined. A pre-trained multi-modal large model is used to determine the risk information corresponding to the mesh sharing and exchange data according to the mesh sharing and exchange data. The multi-modal large model can be trained according to historical sharing and exchange data, historical risk information and neural perception data. The historical risk information includes risk information determined by a preset reviewer according to historical sharing and exchange data. The neural perception data is used to represent the brain neural activity of the preset reviewer when determining the historical risk information. Whether there is a risk in the sharing and exchange of government data in a mesh mode is determined according to the risk information. In this way, on the one hand, when the data structure of the mesh sharing and exchange data to be detected is relatively complex, the multi-modal large model can be used to process the mesh sharing and exchange data, thereby improving the risk detection efficiency and accuracy. On the other hand, since the multi-modal large model is trained according to historical sharing and exchange data, historical risk information and neural perception data, the multi-modal large model can combine the subconscious neural features and the data exchange characteristics in a mesh mode to perform comprehensive risk detection. That is, the multi-modal large model can quickly and accurately determine the risk information, thereby improving the risk detection efficiency and accuracy of the sharing and exchange of government data in a mesh mode and ensuring the security of the sharing and exchange of government data in a mesh mode. Specific processing can be referred to the specific content in the following embodiments.

[0015] As shown in Figure 1 Embodiments of the present application provide a risk detection method. The execution subject of the method can be a terminal device or a server. The terminal device can be a mobile terminal device such as a mobile phone, a tablet computer, a smart watch, etc., or a terminal device such as a computer. The server can be a standalone server or a server cluster composed of multiple servers. The method can specifically include the following steps: In step S102, a risk detection request for the sharing and exchange of government data in a mesh mode is received.

[0016] The government affair data can be data related to one or more application scenarios such as financial management, asset management, environmental facility management, meeting management, reception management, service management, personnel management and system management. The government affair data can include one or more types of data such as text, image, audio and video. The mesh network is a wireless network architecture, and each node in the mesh network can send and receive signals.

[0017] In implementation, the server can trigger a risk detection request for the government affair data shared and exchanged in the mesh network based on a preset detection period, such as once every half hour or 2 hours.

[0018] Alternatively, the server can trigger a risk detection request for the government affair data shared and exchanged in the mesh network mode upon receiving a sharing and exchange request for the government affair data.

[0019] In step S104, in response to the risk detection request, the mesh network sharing and exchange data to be detected corresponding to the risk detection request is determined.

[0020] The mesh network sharing and exchange data can include data sharing and exchange mode, sharing and exchange rule, sharing and exchange data, sharing and exchange scenario and process, etc. For example, the sharing and exchange mode can be a data open sharing mode, the sharing and exchange rule can be a message exchange sharing rule, the sharing and exchange data can be a development and utilization data set, and the sharing and exchange scenario and process can include sharing and exchange scenario description and sharing and exchange flowchart, etc.

[0021] In implementation, the server can obtain the sharing and exchange data to be detected corresponding to the risk detection request, and perform feature extraction processing on the sharing and exchange data by using a large language mode to obtain related features such as sharing and exchange mode, sharing and exchange rule, sharing and exchange scenario and process, etc., so as to construct the mesh network sharing and exchange data to be detected according to the sharing and exchange data and the above-mentioned features extracted.

[0022] In addition, the above-mentioned method for obtaining the mesh network sharing and exchange data to be detected is an optional and implementable method. In actual application scenarios, there can be various different methods for obtaining, and different methods can be selected according to different actual application scenarios, which are not limited in the embodiments of the present application.

[0023] In step S106, the pre-trained multi-modal large model is used to determine the risk information corresponding to the mesh network sharing and exchange data according to the mesh network sharing and exchange data.

[0024] The multi-modal large model can be obtained by training according to historical sharing exchange data, historical risk information, and neural perception data. The historical risk information can include risk information determined by a preset reviewer according to historical sharing exchange data. The neural perception data can be used to represent brain neural activity of the preset reviewer when determining the historical risk information. The multi-modal large model can be a model obtained by training multi-modal information such as text, image, video, and audio. For example, the multi-modal large model can be a multi-modal large language model (MLLM). The multi-modal large model can be a model that combines the natural language processing capabilities of a large language model (LLM) and the understanding and generation capabilities of multiple modalities (such as vision and audio) data. The multi-modal large model integrates various types of input and output such as text, image, video, and audio, and fuses these information to complete complex tasks, and provides users with a more rich and natural interactive experience.

[0025] In implementation, the server can send the historical sharing exchange data to the preset reviewer, and receive the historical risk information and the neural perception data fed back by the preset reviewer. There can be multiple preset reviewers, for example, the preset reviewer can be no less than 12 professional security management personnel with rich security experience in mesh data sharing exchange scenarios.

[0026] The server can collect brain neural activity data of each preset reviewer when determining the historical risk information for the historical sharing exchange data, and determine the neural perception data according to the collected brain neural activity data.

[0027] In step S108, whether the sharing and exchange of the government affair data in the mesh mode exists risk is determined according to the risk information.

[0028] In implementation, taking the risk information including a risk score as an example, the server can determine whether the sharing and exchange of the government affair data in the mesh mode exists risk according to a preset risk threshold and the risk score. That is, if the risk score is not less than the preset risk threshold, it can be determined that the sharing and exchange of the government affair data in the mesh mode exists risk, otherwise, it does not exist risk.

[0029] The above determination method of whether the sharing and exchange of the government affair data in the mesh mode exists risk is an optional and implementable determination method. In actual application scenarios, there can be various different determination methods, and different determination methods can be selected according to different actual application scenarios, which are not limited in the embodiments of the present application.

[0030] In a case where it is determined that there is a risk in sharing and exchanging the government affair data in the mesh mode, the server can output preset alarm information to prompt a preset management party to perform risk control processing on the sharing and exchange in the mesh mode.

[0031] The embodiment of the present application provides a risk detection method, receiving a risk detection request for sharing and exchanging government affair data in a mesh mode, in response to the risk detection request, determining the mesh sharing and exchange data to be detected corresponding to the risk detection request, using a pre-trained multi-modal large model, according to the mesh sharing and exchange data, determining the risk information corresponding to the mesh sharing and exchange data, wherein the multi-modal large model can be obtained by training according to historical sharing and exchange data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset audit party according to historical sharing and exchange data, and the neural perception data is used to represent the brain neural activity of the preset audit party when determining the historical risk information, and according to the risk information, determining whether there is a risk in sharing and exchanging the government affair data in the mesh mode. In this way, on the one hand, in a case where the data structure of the mesh sharing and exchange data to be detected is relatively complex, the multi-modal large model can be used to process the mesh sharing and exchange data, thereby improving the risk detection efficiency and accuracy. On the other hand, since the multi-modal large model is obtained by training according to the historical sharing and exchange data, the historical risk information and the neural perception data, the multi-modal large model can combine the subconscious neural features and the data exchange features in the mesh mode to perform risk detection more comprehensively, that is, the multi-modal large model can quickly and accurately determine the risk information, thereby improving the risk detection efficiency and accuracy of the sharing and exchange of government affair data in the mesh mode, and ensuring the security of the sharing and exchange of government affair data in the mesh mode.

[0032] In actual application, the neural perception data can include amplitudes of brain wave signals in different preset frequency ranges of the preset audit party when determining the historical risk information according to the historical sharing and exchange data within a preset processing time, and blood flow dynamics parameters of the preset audit party collected by using functional near-infrared spectroscopy technology.

[0033] For example, the neural perception data can include amplitude energies of brain wave (EEG) signal frequency ranges of 4-7Hz, 8-10Hz, 11-13Hz, 14-15Hz and 16-20Hz calculated in units of 3 seconds, and mean values and variances of HbO, HbR and HbT in fNIR signals collected by using functional near-infrared spectroscopy (fNIRS) technology.

[0034] The server can synchronously obtain neural activity observation data (i.e., neural perception data) with high temporal resolution and spatial resolution through an EEG-fNIRS device (i.e., a portable neural activity observation device combining electroencephalogram and functional near-infrared spectroscopy technology).

[0035] In actual application, the historical risk information and the neural perception data fed back by the preset auditing party can also be received. The determination method of the historical risk information and the neural perception data can be various. An optional processing manner is provided below, for example, Figure 2 As shown, the processing can specifically include the following steps S202-S204.

[0036] In step S202, historical shared exchange data is obtained, and test data is generated according to the historical shared exchange data and a preset processing time.

[0037] The preset processing time can include display time of the historical shared exchange data, determination time of the historical risk information, etc.

[0038] In implementation, the server can generate the test data according to the historical shared exchange data and the preset processing time by using a pre-trained generation model. The generation model can be a model constructed according to a preset machine learning algorithm.

[0039] For example, assuming that the display time of the historical shared exchange data is 60 seconds and the determination time of the historical risk information is 120 seconds, the test data generated according to the historical shared exchange data and the preset processing time can be: test task and purpose description (10 seconds) + display of the historical shared exchange data (60 seconds) + determination of the historical risk information (120 seconds).

[0040] In addition, the above-mentioned generation method of the test data is an optional and implementable generation method. In actual application scenarios, there can be various different generation methods, and different generation methods can be selected according to different actual application scenarios. The present embodiment does not make specific limitation thereto.

[0041] In step S204, the test data is sent to the preset auditing party, and the historical risk information and the neural perception data fed back by the preset auditing party are received.

[0042] In implementation, the server can send the test data to the preset auditors and collect the neural perception data of the preset auditors when processing the test data. Taking the test data of the test task and the purpose description (10 seconds) + the display of the historical shared exchange data (60 seconds) + the determination of the historical risk information (120 seconds) as an example, the neural perception data collected by the server can include the amplitude energy of the preset auditors' EEG signal frequency range calculated in 3-second units in each stage (i.e., the description, display, and determination stages), which is 4-7 Hz, 8-10 Hz, 11-13 Hz, 14-15 Hz, and 16-20 Hz, and the mean and variance of HbO, HbR, and HbT in the fNIRS signal collected by the functional near-infrared spectroscopy (fNIRS).

[0043] In actual application, the multi-modal large model can also be trained. The training method of the multi-modal large model can be various. The historical risk information can include historical risk types and historical risk scores. Accordingly, an optional processing method is provided as follows. Figure 3 As shown, the method can specifically include the following steps S302-S306.

[0044] In step S302, the first risk type corresponding to the historical shared exchange data is obtained.

[0045] In step S304, the preset auditors are filtered according to the first risk type and the historical risk type fed back by the preset auditors, to obtain filtered preset auditors.

[0046] In implementation, the server can determine the risk perception ability of the preset auditors according to the first risk type and the historical risk type fed back by the preset auditors, then the server can sort the preset auditors according to the risk perception ability of the preset auditors, and filter the preset auditors according to the sorted preset auditors to obtain the filtered preset auditors. For example, the top 50% of the auditors in the sorted preset auditors can be determined as the filtered preset auditors.

[0047] The determination method of the risk perception ability of the preset auditors can be various. For example, the server can determine the risk perception ability of the preset auditors according to the matching result of the first risk type and the historical risk type fed back by the preset auditors.

[0048] In step S306, the multi-modal large model is trained according to the historical shared exchange data, the historical risk information corresponding to the filtered preset auditors, and the neural perception data, to obtain the trained multi-modal large model.

[0049] In actual application, the processing mode of training the multi-modal large model according to the historical shared exchange data, the screened historical risk information corresponding to the preset audit party and the neural perception data in step S306 can be various, and the following provides an optional processing mode, which can specifically include the following steps A1-A2.

[0050] In step A1, a thought chain is constructed by using the risk formation description in the historical shared exchange data, and the multi-modal large model is used to determine the predicted risk information according to the thought chain, the data sharing exchange mode, the sharing exchange rule, the sharing exchange data, the sharing exchange scene and process in the historical shared exchange data, and the neural perception data.

[0051] In step A2, whether the multi-modal large model converges is determined according to the predicted risk information and the screened historical risk information corresponding to the preset audit party, and in the case that the multi-modal large model does not converge, the multi-modal large model is continuously trained according to the historical shared exchange data, the screened historical risk information corresponding to the preset audit party and the neural perception data, until the multi-modal large model converges, and the trained multi-modal large model is obtained.

[0052] In implementation, the server can use the data sharing exchange mode, the sharing exchange rule, the sharing exchange data, the sharing exchange scene and process, and the neural perception data in the historical shared exchange data as input data of the multi-modal large model, use the risk formation description in the historical shared exchange data to construct a thought chain to prompt and train the multi-modal large model, and obtain the predicted risk information including the predicted risk type and the predicted risk score.

[0053] In actual application, the historical shared exchange data can include the data sharing exchange mode, the sharing exchange rule, the sharing exchange data, the sharing exchange scene and process, the first risk type and the risk formation description obtained by using the preset large language model to extract knowledge according to the government data development and utilization cases in the mesh network mode.

[0054] In implementation, the server can use the preset large language model to extract knowledge according to the government data development and utilization cases in the mesh network mode to obtain the historical shared exchange data including the data sharing exchange mode, the sharing exchange rule, the sharing exchange data, the sharing exchange scene and process, the risk type and the risk formation description.

[0055] As shown in Figure 4 The server can detect the risk of government data sharing exchange in the mesh network mode through the following five steps of secure knowledge base construction, neural test signal generation, test data acquisition and processing, multi-modal large model training and security risk perception evaluation, specifically: Step 1: Knowledge base construction.

[0056] The server can construct a knowledge base K = {p, r, d, s, t, m} according to the data types, classification and grading methods, sharing and exchange modes, and security technology framework requirements specified in the National Integrated Government Affairs Data Classification and Grading Standard, the National Integrated Government Affairs Data Sharing and Exchange Standard, and the National Integrated Government Affairs Data Sharing and Open Standard. The knowledge base is constructed by security professionals based on existing experience and analysis of collected mesh network security cases. In this knowledge base, p represents the mesh network-oriented data sharing and exchange mode, r represents the sharing and exchange rules, d represents the involved data sets for sharing and exchange, s represents the data sharing and exchange scenarios and processes, t represents the risk types, and m represents the description of the formation of security risks.

[0057] For example, K = {"data open sharing", "message exchange sharing", "development and utilization data set", s (describing the sharing and exchange scenario and giving the data sharing and exchange process chart), "data desensitization", m (describing how the above data generates security risks in data desensitization)}.

[0058] Based on the above-constructed knowledge base K, the p, r, d, s of each piece of knowledge are taken as input, and their concept definitions and examples are given. According to m, a thought chain is prepared, and a large language model (such as a multi-modal large model) is prompted and trained, and t is output. After the large language model is trained, the p, r, d, s of each piece of knowledge are taken as input, and the large language model is asked "In the above data sharing and exchange scenario of the mesh network, what other security risks exist?" According to the answer of the large model, the risk types of hidden dangers are sorted out and supplemented to the knowledge base.

[0059] Step 2: Generation of neural test signals.

[0060] Based on the above-constructed knowledge base, the following method is used to generate neural test signals (i.e. test data) for computer screen display: S = test task and purpose explanation (10 seconds) + test scene sequence number and p, r, d, s display (60 seconds) + risk type, risk score, and risk formation description questionnaire filling (120 seconds). The risk score can be given in the continuous interval [0, 5].

[0061] Step 3: Test data acquisition and processing.

[0062] Take at least 12 professional security management personnel with rich security experience in mesh data sharing exchange as test objects (i.e. preset audit parties) to conduct synchronous testing, collect their EEG-fNIRS data (i.e. neural perception data) and questionnaire filling data in the "test scene number and p, r, d, s display (60 seconds)" time period, and process the data as follows: Compare the risk types given in the questionnaire data filled by the tested personnel with the corresponding risk types t (i.e. the first risk type) not displayed in their test signals to determine the safety risk perception ability of the tested personnel. The top 50% personnel with higher safety risk perception ability ranking can use their EEG-fNIRS data to calculate the amplitude energy of the EEG signal frequency range of 4-7Hz, 8-10Hz, 11-13Hz, 14-15Hz, 16-20Hz and the mean and variance of HbO, HbR and HbT in fNIRS signal in units of 3 seconds. The above 11 parameters are used as characteristic indexes for large model training, i.e. the neural perception data corresponding to the preset audit parties can be determined according to the 11 parameters.

[0063] Step 4: Multimodal large model training.

[0064] The 11 parameters in the EEG-fNIRS data of the top 50% personnel in terms of safety risk perception ability ranking extracted from step 3 and the p, r, d, s data corresponding to step 3 are used as inputs for multimodal large model training, and the risk types and risk scores in the questionnaire data of the tested personnel are used as outputs. The multimodal large model is prompted and trained based on the description of risk formation until the desired accuracy requirement is met, and the trained multimodal large model is obtained.

[0065] Step 5: Safety risk perception evaluation.

[0066] Based on the trained multimodal large model, the mesh data sharing exchange data (i.e. including sharing exchange mode, sharing exchange rules, sharing exchange data set involved and sharing exchange data of data sharing exchange scene) that needs to be evaluated can be evaluated for safety risk, such as inputting the p, r, d, s data of the mesh sharing exchange data to be detected, and the multimodal large model generates risk information including risk type, risk score of each risk classification and description of risk formation, so as to determine whether there is risk in the sharing exchange of government data in the mesh mode according to the risk information. In addition, the server can also store the above processing results.

[0067] By means of multi-modal large model technology, EEG-fNIRS neural activity data acquisition technology, and in combination with the experience of relevant professionals and human neural perception characteristics, a government affair data sharing and exchange neural perception security evaluation method for a mesh network is constructed in accordance with relevant standards and requirements of national integrated government affair data management, which can guarantee the security of government affair data in a mesh network mode.

[0068] By constructing a knowledge base, multi-modal large models can be used to generate scenario signals with generalization ability and related security risks for different data items and sharing and exchange rules, which can more fully consider security risks in various scenarios in a mesh network mode.

[0069] By setting a screening method for the perception ability of security personnel and a feature parameter extraction method for EEG-fNIRS data, the effectiveness and efficiency of training data can be improved. By training the association between neural perception data of security professionals and security risk assessment levels through large model technology, a more comprehensive security risk assessment can be performed in combination with subconscious neural features and scenario features. Compared with existing technologies, the proposed method can more comprehensively evaluate the risks of complex data transmission, sharing and exchange scenarios, and the implicit security information obtained by human understanding and processing of data and their associations.

[0070] The embodiment of the present specification provides a risk detection method, receiving a risk detection request for sharing and exchanging government affair data in a mesh network mode, in response to the risk detection request, determining the mesh network sharing and exchanging data to be detected corresponding to the risk detection request, using a pre-trained multi-modal large model, according to the mesh network sharing and exchanging data, determining the risk information corresponding to the mesh network sharing and exchanging data, wherein the multi-modal large model can be obtained by training according to historical sharing and exchanging data, historical risk information and neural perception data, the historical risk information includes the risk information determined by a preset audit party according to the historical sharing and exchanging data, and the neural perception data is used to represent the brain neural activity of the preset audit party when determining the historical risk information, and according to the risk information, determining whether there is a risk in sharing and exchanging government affair data in a mesh network mode. In this way, on the one hand, in the case that the data structure of the mesh network sharing and exchanging data to be detected is relatively complex, the multi-modal large model can be used to process the mesh network sharing and exchanging data, thereby improving the risk detection efficiency and accuracy. On the other hand, since the multi-modal large model is trained according to historical sharing and exchanging data, historical risk information and neural perception data, the multi-modal large model can combine subconscious neural features and data exchange features in a mesh network mode to perform more comprehensive risk detection, i.e., the multi-modal large model can quickly and accurately determine the risk information, thereby improving the risk detection efficiency and accuracy of government affair data sharing and exchange in a mesh network mode, and ensuring the security of government affair data sharing and exchange in a mesh network mode.

[0071] Based on the same idea, the embodiments of the present specification also provide a risk detection device, as shown in the following. Figure 5

[0072] The risk detection device comprises a request receiving module 501, a data determining module 502, an information determining module 503, and a risk detection module 504, wherein: The request receiving module 501 is configured to receive a risk detection request for sharing and exchanging government data in a mesh mode. The data determining module 502 is configured to determine, in response to the risk detection request, mesh sharing and exchanging data to be detected corresponding to the risk detection request. The information determining module 503 is configured to determine, by using a pre-trained multi-modal large model, risk information corresponding to the mesh sharing and exchanging data according to the mesh sharing and exchanging data; wherein the multi-modal large model is obtained by training according to historical sharing and exchanging data, historical risk information, and neural perception data, the historical risk information includes risk information determined by a preset audit party according to the historical sharing and exchanging data, and the neural perception data is used to represent brain neural activity of the preset audit party when determining the historical risk information. The risk detection module 504 is configured to determine whether there is a risk in sharing and exchanging the government data in the mesh mode according to the risk information.

[0073] In the embodiments of the present specification, the neural perception data includes amplitudes of brain wave signals in different preset frequency ranges of the preset audit party when determining the historical risk information according to the historical sharing and exchanging data within a preset processing time, and hemodynamic parameters of the preset audit party collected by using functional near-infrared spectroscopy technology.

[0074] In the embodiments of the present specification, the device further comprises: The data acquisition module is configured to acquire the historical sharing and exchanging data, and generate test data according to the historical sharing and exchanging data and the preset processing time; The information receiving module is configured to send the test data to the preset audit party, and receive the historical risk information and the neural perception data fed back by the preset audit party.

[0075] In the embodiments of the present specification, the historical risk information includes historical risk types and historical risk scores, and the device further comprises: The first acquisition module is configured to acquire a first risk type corresponding to the historical sharing and exchanging data. ​The screening module is configured to perform screening processing on the preset auditors according to the first risk type and the historical risk types of the preset auditors, to obtain screened preset auditors. The model training module is configured to train the multi-modal large model according to the historical shared exchange data, the historical risk information corresponding to the screened preset auditors, and the neural perception data, to obtain a trained multi-modal large model.

[0076] In the embodiments of the present specification, the historical shared exchange data includes data sharing exchange mode, sharing exchange rule, sharing exchange data, sharing exchange scene and process, the first risk type, and risk formation description obtained by using a preset large language model to extract knowledge according to a utilization case of government affair data in a mesh network mode.

[0077] In the embodiments of the present specification, the model training module is configured to: construct a thinking chain using the risk formation description in the historical shared exchange data, and determine predicted risk information according to the thinking chain, the data sharing exchange mode, the sharing exchange rule, the sharing exchange data, the sharing exchange scene and process in the historical shared exchange data, and the neural perception data using the multi-modal large model; determine whether the multi-modal large model converges according to the predicted risk information and the historical risk information corresponding to the screened preset auditors, and continue to train the multi-modal large model according to the historical shared exchange data, the historical risk information corresponding to the screened preset auditors, and the neural perception data until the multi-modal large model converges, to obtain a trained multi-modal large model, in a case where it is determined that the multi-modal large model does not converge.

[0078] The embodiment of the present specification provides a risk detection device, receives a risk detection request for sharing and exchanging government data in a mesh mode, determines mesh sharing and exchanging data to be detected corresponding to the risk detection request in response to the risk detection request, determines risk information corresponding to the mesh sharing and exchanging data by using a pre-trained multi-modal large model according to the mesh sharing and exchanging data, wherein the multi-modal large model can be obtained by training according to historical sharing and exchanging data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset audit party according to the historical sharing and exchanging data, and the neural perception data is used to represent the brain neural activity of the preset audit party when determining the historical risk information, and determines whether there is a risk in sharing and exchanging government data in the mesh mode according to the risk information. In this way, on the one hand, in the case that the data structure of the mesh sharing and exchanging data to be detected is relatively complex, the multi-modal large model can be used to process the mesh sharing and exchanging data, and the risk detection efficiency and accuracy can be improved. On the other hand, since the multi-modal large model is obtained by training according to the historical sharing and exchanging data, the historical risk information and the neural perception data, the multi-modal large model can combine the subconscious neural features and the data exchange features in the mesh mode to perform risk detection more comprehensively, that is, the multi-modal large model can quickly and accurately determine the risk information, which can ensure the security of the sharing and exchanging of government data in the mesh mode on the basis of improving the risk detection efficiency and accuracy of the sharing and exchanging of government data in the mesh mode.

[0079] The above is the risk detection device provided by the embodiment of the present specification. Based on the same idea, the embodiment of the present specification also provides an electronic device, as shown in Figure 6 .

[0080] The electronic device can be a terminal device or a server provided by the above embodiment.

[0081] The electronic device can have a large difference due to different configurations or performances, and can include one or more processors 601 and memories 602, and the memories 602 can store one or more storage applications or data. The memory 602 can be temporary storage or persistent storage. The application stored in the memory 602 can include one or more modules (not shown in the figure), and each module can include a series of computer executable instructions in the electronic device. Further, the processor 601 can be configured to communicate with the memory 602 and execute a series of computer executable instructions in the memory 602 on the electronic device. The electronic device can also include one or more power supplies 603, one or more wired or wireless network interfaces 604, one or more input and output interfaces 605, and one or more keyboards 606.

[0082] In particular embodiments, an electronic device includes memory, and one or more programs, wherein the one or more programs are stored in the memory and executable by one or more processors of the electronic device and comprise one or more modules that each comprise a series of computer-executable instructions, and the one or more programs configured to be executed by the one or more processors comprise computer-executable instructions for: receiving a risk detection request for sharing and exchanging of government data in a mesh mode; in response to the risk detection request, determining mesh sharing and exchanging data to be detected corresponding to the risk detection request; using a pre-trained multi-modal large model, determining risk information corresponding to the mesh sharing and exchanging data according to the mesh sharing and exchanging data; wherein the multi-modal large model is obtained by training according to historical sharing and exchanging data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset auditing party according to the historical sharing and exchanging data, and the neural perception data is used to represent brain neural activity of the preset auditing party when determining the historical risk information; determining whether there is a risk in sharing and exchanging of the government data in the mesh mode according to the risk information.

[0083] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for the electronic device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant part can be referred to the part of the method embodiment.

[0084] The embodiment of the specification provides an electronic device, receives a risk detection request for sharing and exchanging government data in a mesh mode, determines mesh sharing and exchanging data to be detected corresponding to the risk detection request in response to the risk detection request, determines risk information corresponding to the mesh sharing and exchanging data by using a pre-trained multi-modal large model according to the mesh sharing and exchanging data, wherein the multi-modal large model can be obtained by training according to historical sharing and exchanging data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset audit party according to the historical sharing and exchanging data, and the neural perception data is used to represent brain neural activity of the preset audit party when determining the historical risk information, and determines whether there is a risk in sharing and exchanging the government data in the mesh mode according to the risk information. In this way, on the one hand, in the case that the data structure of the mesh sharing and exchanging data to be detected is relatively complex, the multi-modal large model can be used to process the mesh sharing and exchanging data, thereby improving the risk detection efficiency and accuracy. On the other hand, since the multi-modal large model is obtained by training according to the historical sharing and exchanging data, the historical risk information and the neural perception data, the multi-modal large model can combine the subconscious neural features and the data exchange features in the mesh mode to comprehensively perform risk detection, that is, the multi-modal large model can quickly and accurately determine the risk information, thereby improving the risk detection efficiency and accuracy of the government data sharing and exchanging in the mesh mode and ensuring the security of the government data sharing and exchanging in the mesh mode.

[0085] Further, based on the above Figures 1 to 4 The one or more embodiments of the specification also provide a storage medium for storing computer executable instruction information, and in a specific embodiment, the storage medium can be a U disk, an optical disk, a hard disk, etc. The computer executable instruction information stored in the storage medium can implement the following processes when executed by a processor: receiving a risk detection request for sharing and exchanging government data in a mesh mode; in response to the risk detection request, determining mesh sharing and exchanging data to be detected corresponding to the risk detection request; determining risk information corresponding to the mesh sharing and exchanging data by using a pre-trained multi-modal large model according to the mesh sharing and exchanging data; wherein the multi-modal large model is obtained by training according to historical sharing and exchanging data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset audit party according to the historical sharing and exchanging data, and the neural perception data is used to represent brain neural activity of the preset audit party when determining the historical risk information; determining whether there is a risk in sharing and exchanging the government data in the mesh mode according to the risk information.

[0086] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each of the embodiments focuses on the difference from other embodiments. In particular, for the above-mentioned storage medium embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.

[0087] The embodiment of the specification provides a storage medium, receives a risk detection request for sharing and exchanging government data in a mesh mode, determines mesh sharing and exchanging data to be detected corresponding to the risk detection request in response to the risk detection request, determines risk information corresponding to the mesh sharing and exchanging data by using a pre-trained multi-modal large model according to the mesh sharing and exchanging data, wherein the multi-modal large model can be obtained by training according to historical sharing and exchanging data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset audit party according to the historical sharing and exchanging data, and the neural perception data is used to represent brain neural activity of the preset audit party when determining the historical risk information, and determines whether there is a risk in sharing and exchanging the government data in the mesh mode according to the risk information. In this way, on the one hand, in the case that the data structure of the mesh sharing and exchanging data to be detected is relatively complex, the multi-modal large model can be used to process the mesh sharing and exchanging data, thereby improving the risk detection efficiency and accuracy. On the other hand, since the multi-modal large model is obtained by training according to the historical sharing and exchanging data, the historical risk information and the neural perception data, the multi-modal large model can combine the subconscious neural features and the data exchange features in the mesh mode to perform risk detection more comprehensively, that is, the multi-modal large model can quickly and accurately determine the risk information, thereby improving the risk detection efficiency and accuracy of the government data sharing and exchanging in the mesh mode, and ensuring the security of the government data sharing and exchanging in the mesh mode.

[0088] Further, based on the above-mentioned method, one or more embodiments of the specification further provide a computer program product comprising a computer program, wherein the computer program in the computer program product can implement the following flow when executed by a processor: Figures 1 to 4 receiving a risk detection request for sharing and exchanging government data in a mesh mode; determining mesh sharing and exchanging data to be detected corresponding to the risk detection request in response to the risk detection request; ​According to the pre-trained multi-modal large model, the risk information corresponding to the mesh network sharing exchange data is determined according to the mesh network sharing exchange data; wherein the multi-modal large model is obtained by training according to historical sharing exchange data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset auditing party according to the historical sharing exchange data, and the neural perception data is used to represent the brain neural activity of the preset auditing party when determining the historical risk information. According to the risk information, it is determined whether there is a risk in the sharing and exchange of the government data in the mesh network mode.

[0089] Each of the embodiments in the specification is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for the above-mentioned computer program product embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the related parts can be referred to the part of the method embodiment.

[0090] The embodiment of the specification provides a computer program product, receives a risk detection request for sharing and exchange of government data in a mesh network mode, determines the mesh network sharing exchange data to be detected corresponding to the risk detection request in response to the risk detection request, and determines the risk information corresponding to the mesh network sharing exchange data according to the mesh network sharing exchange data by using a pre-trained multi-modal large model. The multi-modal large model can be obtained by training according to historical sharing exchange data, historical risk information and neural perception data. The historical risk information includes risk information determined by a preset auditing party according to the historical sharing exchange data. The neural perception data is used to represent the brain neural activity of the preset auditing party when determining the historical risk information. According to the risk information, it is determined whether there is a risk in the sharing and exchange of the government data in the mesh network mode. In this way, on the one hand, in the case that the data structure of the mesh network sharing exchange data to be detected is relatively complex, the multi-modal large model can be used to process the mesh network sharing exchange data, thereby improving the risk detection efficiency and accuracy. On the other hand, since the multi-modal large model is obtained by training according to historical sharing exchange data, historical risk information and neural perception data, the multi-modal large model can combine the subconscious neural features and the data exchange features in the mesh network mode to perform risk detection more comprehensively, that is, the multi-modal large model can quickly and accurately determine the risk information, thereby improving the risk detection efficiency and accuracy of the sharing and exchange of government data in the mesh network mode, and ensuring the security of the sharing and exchange of government data in the mesh network mode.

[0091] The above described embodiments of the present description have been described. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order and still achieve desirable results. Additionally, the processes depicted in the figures do not necessarily require the particular order shown, or sequential order, to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous.

[0092] In the 1990s, it was possible to distinguish whether an improvement in a technology was a hardware improvement (e.g., an improvement in the circuit structure of a diode, transistor, switch, etc.) or a software improvement (an improvement in a method flow). However, as technology has advanced, many improvements in method flows today can be considered as direct improvements in hardware circuit structures. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into a hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (e.g., a field programmable gate array (FPGA)) is an integrated circuit whose logic function is determined by the user programming the device. A digital system is "integrated" on a PLD by the designer programming it himself, without having to ask a chip manufacturer to design and manufacture a special integrated circuit chip. Moreover, instead of manually manufacturing an integrated circuit chip, this programming is now mostly implemented using "logic compiler" software, which is similar to the software compiler used when developing a program, and the original code before compilation must also be written in a specific programming language, which is called a hardware description language (HDL), and there are many types of HDL, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., and the most commonly used are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also be aware that it is only necessary to logically program the method flow using the above-mentioned hardware description languages and program it into an integrated circuit to easily obtain a hardware circuit that implements the logical method flow.

[0093] The controller can be implemented in any suitable way, for example, the controller can take the form of a microprocessor or processor and a computer readable medium storing computer readable program code, such as software or firmware, executable by the microprocessor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320, the memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also know that, in addition to implementing the controller in pure computer readable program code, it is possible to implement the same functionality in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. The controller can thus be considered a hardware component, and the means included therein for implementing the various functions can be considered structures within the hardware component. Alternatively, or even additionally, the means for implementing the various functions can be considered both software modules implementing the method and structures within the hardware component.

[0094] The systems, apparatuses, modules or units illustrated by the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0095] For the sake of description, the above apparatuses are described in various units by functions respectively. Of course, the functions of each unit can be implemented in one or more software and / or hardware in implementing one or more embodiments of the present specification.

[0096] Those skilled in the art will understand that the embodiments of the present specification can be provided as a method, a system, or a computer program product. Therefore, one or more embodiments of the present specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, one or more embodiments of the present specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0097] The embodiments of the present specification are described with reference to flowcharts and / or block diagrams of the method, device (system) and computer program product according to the embodiments of the present specification. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable electronic devices to produce a machine, so that the instructions executed by the computer or other programmable electronic devices generate a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions of one or more flows and / or blocks in the flowcharts and / or block diagrams can be implemented by an apparatus. Figure 1 The functions of one or more flows and / or blocks in the flowcharts and / or block diagrams can be implemented by an apparatus.

[0098] These computer program instructions can also be stored in a computer readable memory that can direct the computer or other programmable electronic devices to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including instruction apparatus, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions of one or more flows and / or blocks in the flowcharts and / or block diagrams can be implemented by an apparatus. Figure 1 The functions of one or more flows and / or blocks in the flowcharts and / or block diagrams can be implemented by an apparatus.

[0099] These computer program instructions can also be loaded into a computer or other programmable electronic devices, so that a series of operation steps are performed on the computer or other programmable electronic devices to produce a computer implemented process, so that the instructions executed on the computer or other programmable electronic devices provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions of one or more flows and / or blocks in the flowcharts and / or block diagrams can be implemented by an apparatus. Figure 1 The functions of one or more flows and / or blocks in the flowcharts and / or block diagrams can be implemented by an apparatus.

[0100] In a typical configuration, the computing device includes one or more processors (CPU), input / output interface, network interface and memory.

[0101] The memory can include non-persistent memory in the computer readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer readable media.

[0102] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0103] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or apparatus that comprises a list of elements does not only include those elements, but can also include other elements not expressly listed or inherent to such process, method, article or apparatus. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0104] Those skilled in the art will appreciate that embodiments of the present specification can be provided as methods, systems or computer program products. Therefore, one or more embodiments of the present specification can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, one or more embodiments of the present specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0105] One or more embodiments of the present specification can be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. One or more embodiments of the present specification can also be practiced in a distributed computing environment, where tasks are performed by remote processing devices that are connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0106] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each of the embodiments focuses on the difference from other embodiments. In particular, for the system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiments.

[0107] The above only describes the embodiments of the specification and is not used to limit the file. The specification can have various changes and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the specification shall be included in the claim range of the specification.

Claims

1. A risk detection method, characterized by, The method comprises: receiving a risk detection request for sharing and exchanging government data in a mesh mode; in response to the risk detection request, determining mesh sharing and exchanging data to be detected corresponding to the risk detection request; using a pre-trained multi-modal large model, determining risk information corresponding to the mesh sharing and exchanging data according to the mesh sharing and exchanging data; wherein the multi-modal large model is trained according to historical sharing and exchanging data, historical risk information, and neural perception data, the historical risk information includes risk information determined by a preset reviewer according to the historical sharing and exchanging data, and the neural perception data is used to represent brain neural activity of the reviewer when determining the historical risk information; determining whether there is a risk in sharing and exchanging government data in a mesh mode according to the risk information.

2. The method of claim 1, wherein, The neural perception data includes the amplitude of the brain wave signal in different preset frequency ranges of the reviewer within a preset processing time when determining the historical risk information according to the historical sharing and exchanging data, and the blood flow dynamics parameters of the reviewer collected by functional near-infrared spectroscopy technology.

3. The method of claim 2, wherein, Before the step of using a pre-trained multi-modal large model to determine risk information corresponding to the mesh sharing and exchanging data according to the mesh sharing and exchanging data, the method further comprises: obtaining the historical sharing and exchanging data, and generating test data according to the historical sharing and exchanging data and the preset processing time; sending the test data to the reviewer, and receiving the historical risk information and the neural perception data fed back by the reviewer.

4. The method of claim 3, wherein, The historical risk information includes historical risk types and historical risk scores, and before the step of using a pre-trained multi-modal large model to determine risk information corresponding to the mesh sharing and exchanging data according to the mesh sharing and exchanging data, the method further comprises: obtaining a first risk type corresponding to the historical sharing and exchanging data; screening the reviewer according to the first risk type and the historical risk type fed back by the reviewer to obtain a screened reviewer; training the multi-modal large model according to the historical sharing and exchanging data, the historical risk information corresponding to the screened reviewer, and the neural perception data, to obtain a trained multi-modal large model.

5. The method of claim 4, wherein, The historical sharing and exchanging data includes data sharing and exchanging mode, sharing and exchanging rules, sharing and exchanging data, sharing and exchanging scenarios and processes, the first risk type, and risk formation description obtained by using a preset large language model to extract knowledge from government data development cases in a mesh mode.

6. The method of claim 5, wherein, The step of training the multi-modal large model according to the historical sharing and exchanging data, the historical risk information corresponding to the screened reviewer, and the neural perception data to obtain a trained multi-modal large model comprises: construct a thought chain by using the risk formation description in the historical sharing exchange data, and determine predicted risk information according to the thought chain, a data sharing exchange mode, a sharing exchange rule, sharing exchange data, a sharing exchange scene and a process in the historical sharing exchange data, and the neural perception data by using the multi-modal large model; determine whether the multi-modal large model converges according to the predicted risk information and the historical risk information corresponding to the preset audit party after the screening, and continue to train the multi-modal large model according to the historical sharing exchange data, the historical risk information corresponding to the preset audit party after the screening and the neural perception data until the multi-modal large model converges, so as to obtain the trained multi-modal large model, in the case where it is determined that the multi-modal large model does not converge.

7. A risk detection apparatus characterized by comprising: The apparatus comprises: a request receiving module configured to receive a risk detection request for sharing and exchanging government data in a mesh mode; a data determining module configured to determine, in response to the risk detection request, mesh sharing exchange data to be detected corresponding to the risk detection request; an information determining module configured to determine, by using a pre-trained multi-modal large model, risk information corresponding to the mesh sharing exchange data according to the mesh sharing exchange data; wherein the multi-modal large model is obtained by training according to historical sharing exchange data, historical risk information and neural perception data, the historical risk information includes risk information determined by a preset audit party according to the historical sharing exchange data, and the neural perception data is used to represent brain neural activity of the preset audit party when determining the historical risk information; a risk detection module configured to determine whether there is a risk in sharing and exchanging the government data in the mesh mode according to the risk information.

8. An electronic device, comprising: The computer program is stored on the computer readable storage medium and is executed by the processor to implement the steps of the risk detection method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer program is stored on the computer readable storage medium and is executed by the processor to implement the steps of the risk detection method according to any one of claims 1 to 6.

10. A computer program product, characterised in that, The computer program is stored on the computer readable storage medium and is executed by the processor to implement the steps of the risk detection method according to any one of claims 1 to 6.

Images