Risk control processing method and device based on credit investigation

Abstract

The embodiment of the invention provides a risk control processing method and device based on credit investigation, and the method comprises the steps: calling a credit investigation platform to carry out the credit investigation data collection through a risk control engine according to a risk control request which is transmitted by a service system and carries a credit investigation authorization voucher of a user in a risk control processing process for the user, and obtaining desensitization credit investigation data uploaded by the credit investigation platform from the data isolation space of the trusted data space, sending the desensitization credit investigation data to the privacy calculation space, further obtaining a service risk control rule corresponding to the risk control request based on the service authorization of the service system, and transmitting the service risk control rule and the risk control model to the privacy calculation space, and inputting the desensitization credit investigation data and the service risk control rule into a risk control model to carry out service risk detection, and finally synchronizing a risk detection result output by the privacy calculation space to a service system so as to realize risk control processing aiming at the user from the credit investigation data of the user.

Description

TECHNICAL FIELD

[0001] The present document relates to the technical field of data processing, and in particular to a risk control processing method and device based on credit investigation. BACKGROUND

[0002] With the continuous development of Internet technology and big data application, various online service scenarios are continuously enriched, and the role of user credit investigation data in risk identification and management is increasingly prominent. For example, in the recruitment service scenario, the relevant service party often needs to reasonably evaluate the credit status of the user to assist in decision-making and control potential risks in the process of providing recruitment services. With the gradual application of multi-source data analysis technology in related services, in addition to relying on manual screening, evaluation and investigation, the demand for comprehensive analysis of user-related data in the service process is increasing. In this case, how to reasonably utilize user credit investigation data and risk assessment becomes the focus of attention of all parties. SUMMARY

[0003] One or more embodiments of the present specification provide a risk control processing method based on credit investigation, applied to a risk control engine, the method comprising: calling a credit investigation platform for credit investigation data collection according to a risk control request sent by a service system, the risk control request carrying a credit investigation authorization credential of a user. Obtaining desensitized credit investigation data uploaded by the credit investigation platform from a data isolation space of a trusted data space, and sending the desensitized credit investigation data to a privacy computing space. Obtain the service risk control rules corresponding to the risk control request based on the service authorization of the service system, and input the service risk control rules and a risk control model into the privacy computing space, so as to input the desensitized credit investigation data and the service risk control rules into the risk control model for service risk detection. Synchronize the risk detection result output by the privacy computing space to the service system.

[0004] One or more embodiments of the present specification provide a risk control processing device based on credit investigation, running in a risk control engine, the device comprising: a data collection module configured to call a credit investigation platform for credit investigation data collection according to a risk control request sent by a service system, the risk control request carrying a credit investigation authorization credential of a user. A data sending module configured to obtain desensitized credit investigation data uploaded by the credit investigation platform from a data isolation space of a trusted data space, and send the desensitized credit investigation data to a privacy computing space. A risk detection module configured to obtain the service risk control rules corresponding to the risk control request based on the service authorization of the service system, and input the service risk control rules and a risk control model into the privacy computing space, so as to input the desensitized credit investigation data and the service risk control rules into the risk control model for service risk detection. A result synchronization module configured to synchronize the risk detection result output by the privacy computing space to the service system.

[0005] The one or more embodiments of the specification provide a credit-based risk control processing device, comprising: a processor; and a memory configured to store computer executable instructions which, when executed, cause the processor to: according to a risk control request carrying a credit authorization credential of a user sent by a service system, call a credit platform for credit data collection. Obtain the desensitization credit data uploaded by the credit platform from the data isolation space of the trusted data space, and send the desensitization credit data to the privacy computing space. Obtain the service risk control rules corresponding to the risk control request based on the service authorization of the service system, and input the service risk control rules and the risk control model into the privacy computing space to input the desensitization credit data and the service risk control rules into the risk control model for service risk detection. Synchronize the risk detection result output by the privacy computing space to the service system.

[0006] The one or more embodiments of the specification provide a computer readable storage medium for storing computer executable instructions, which, when executed, implement the following processes: according to a risk control request carrying a credit authorization credential of a user sent by a service system, call a credit platform for credit data collection. Obtain the desensitization credit data uploaded by the credit platform from the data isolation space of the trusted data space, and send the desensitization credit data to the privacy computing space. Obtain the service risk control rules corresponding to the risk control request based on the service authorization of the service system, and input the service risk control rules and the risk control model into the privacy computing space to input the desensitization credit data and the service risk control rules into the risk control model for service risk detection. Synchronize the risk detection result output by the privacy computing space to the service system. BRIEF DESCRIPTION OF DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the one or more embodiments of the specification 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 specification, and those skilled in the art can also obtain other drawings according to these drawings without creative labor; Figure 1 A schematic diagram of an embodiment environment for providing a credit-based risk control processing method is provided in the specification; Figure 2 A processing flowchart of a credit-based risk control processing method is provided in the specification; Figure 3 A processing flowchart of a credit-based risk control processing method applied to a risk control processing scene is provided in the specification; Figure 4A credit-based risk control processing method flowchart provided by one or more embodiments of the present specification for a recruitment service scenario; Figure 5 A schematic diagram of a credit-based risk control processing device embodiment provided by one or more embodiments of the present specification; Figure 6 A structural schematic diagram of a credit-based risk control processing device provided by one or more embodiments of the present specification. DETAILED DESCRIPTION

[0008] In order to enable personnel in the technical field to better understand the technical solutions in one or more embodiments of the present specification, the technical solutions in one or more embodiments of the present specification will be described clearly and completely below in conjunction with the drawings in one or more embodiments of the present specification. Obviously, the described embodiments are only a part of the embodiments of the present specification, rather than all the embodiments. Based on one or more embodiments of the present specification, all other embodiments obtained by those of ordinary skill in the art without creative labor should fall within the protection scope of the present document.

[0009] The credit-based risk control processing method provided by one or more embodiments of the present specification can be applied to the implementation environment of a risk control engine, which is described with reference to Figure 1 The implementation environment at least includes: a service system 101, a risk control engine 102, a risk control model 103 deployed in the risk control engine, and a credit platform 104; The service system 101 is configured to send a risk control request to the risk control engine 102, cooperate with the risk control engine 102 to perform service risk detection, and obtain a risk detection result synchronized by the risk control engine 102. The service system 101 can be deployed in a server. The server can be one or more servers, a server cluster composed of several servers, or a cloud server of a cloud computing platform. The risk control engine 102 is configured to obtain the risk control request sent by the service system 101, call the credit platform 104 to collect user credit data in response to the risk control request, obtain desensitized credit data uploaded by the credit platform 104 from a data isolation space of a trusted data space, call the risk control model 103 to perform service risk detection, and synchronize the risk detection result obtained by the service risk detection to the service system 101. The risk control engine 102 can be deployed in a server or independently. The server in which the service system 101 is deployed and the server in which the risk control engine 102 is deployed can be the same server or different servers. The risk control model 103 performs service risk detection based on the desensitized credit data and service risk control rules in response to the call of the risk control engine 102, and outputs a risk detection result. The risk control model 103 can be deployed in the risk control engine 102. The credit investigation platform 104 is configured to store the desensitized credit investigation data of the user. The credit investigation platform 104 can be a server, a server cluster composed of a plurality of servers, or one or more cloud servers in a cloud computing platform.

[0010] The implementation environment can further include a user terminal 105. The user terminal 105 is configured to cooperate with the service system 101 to perform credit investigation authorization processing of the user. The user terminal 105 can be a mobile phone, a personal computer, a tablet computer, an electronic book reader, a device for information interaction based on VR (Virtual Reality) and AR (Augmented Reality), a vehicle-mounted terminal, an IoT device, a wearable smart device, a laptop computer, and a desktop computer, etc.

[0011] In the implementation environment, the risk control engine 102 invokes the credit investigation platform 104 to collect credit investigation data according to the risk control request sent by the service system 101 and carrying the credit investigation authorization credential of the user. Then, the risk control engine 102 obtains the desensitized credit investigation data uploaded by the credit investigation platform 104 from the data isolation space of the trusted data space, and sends the desensitized credit investigation data to the privacy computing space. Further, the risk control engine 102 obtains the service risk control rules corresponding to the risk control request based on the service authorization of the service system 101, and transmits the service risk control rules and the risk control model into the privacy computing space. The risk control model 103 is invoked in the privacy computing space to input the desensitized credit investigation data and the service risk control rules into the risk control model 103. The risk control model 103 performs service risk detection based on the desensitized credit investigation data and the service risk control rules and outputs the risk detection result. Thereafter, the risk control engine 102 obtains the risk detection result output by the privacy computing space and synchronizes the risk detection result to the service system 101. In this way, the risk control processing of the user is realized based on the credit investigation data of the user.

[0012] It should be noted that the desensitized credit investigation data, the user service data, the user resume data, and the contract storage and other related data involved in the present specification may, to some extent, belong to the privacy of the user. Therefore, in order to collect the desensitized credit investigation data, the user service data, the user resume data, and the contract storage and other related data, the authorization of the user can be obtained before collecting the data, so that the operation of collecting the data complies with the relevant data management regulations. For example, the user can authorize the data when accessing the service system. The specific way of data authorization can be to send a data authorization reminder to the user. The user can obtain data authorization by confirming the reminder through an instruction. Alternatively, the data authorization can be obtained by signing a data authorization agreement. The present embodiment is not limited in this regard.

[0013] One or more embodiments of the credit investigation-based risk control processing method provided in the present specification are as follows: Reference Figure 2 The credit-based risk control processing method provided in this embodiment is applied to a risk control engine. The method specifically includes steps S202 to S208.

[0014] Step S202: Based on the risk control request sent by the service system carrying the user's credit authorization certificate, call the credit reporting platform to collect credit data.

[0015] In this embodiment, the risk control engine refers to a processing engine integrated into the service system for risk control processing. Specifically, the risk control engine can be a processing engine that performs risk detection based on user credit data and other relevant data in a specific service scenario. In the process of risk detection, the risk control engine can acquire user credit data, configure service risk control rules, and / or call risk control models to perform risk detection processing. In addition, the risk control engine can also be used to perform other tasks related to risk detection processing. The risk control engine can be in the form of a plug-in or a component.

[0016] For example, in a recruitment service scenario, a risk control engine can be a recruitment risk control engine that performs risk detection based on relevant data of job applicants during the recruitment process. For instance, a recruitment risk control engine can perform risk detection based on the credit data and / or resume data of job applicants.

[0017] The service system refers to a system provided by an enterprise or a third-party platform for performing specific service processing. For example, a service system can be a recruitment service system, a leasing service system, or a resource service system. For instance, in the scenario where the service system is a recruitment service system, the recruitment service system can be a data platform used by an enterprise to carry out recruitment services. Specifically, the enterprise or third-party platform can use the recruitment service system to post job openings, manage resumes, and / or process application processes.

[0018] In practice, during the risk control process, once the user has completed credit authorization, the service system can generate a risk control request carrying the user's credit authorization credentials and send it to the risk control engine. Correspondingly, this system retrieves the risk control request sent by the service system and, based on the request carrying the user's credit authorization credentials, calls the credit reporting platform to collect credit data. Optionally, the risk control request can be sent after the service system performs service risk control checks.

[0019] The credit authorization certificate refers to an electronic certificate generated by the service system or issued by the authorization management module after the user agrees to the authorization in the service system, which serves to prove that the user has authorized the credit information. Credit authorization refers to the act of the user authorizing the service provider or a third party entrusted by the service provider to query the user's personal credit information. For example, credit authorization can be the user authorizing the service provider or a third party entrusted by the service provider to query the user's credit information by means of electronic signature or ticking a consent form.

[0020] In practical applications, when service providers are processing services for users, they can assess users' credit data and whether users have potential risks. In this case, in order to avoid indiscriminate credit inquiries for all users and reduce the waste of risk control resources, credit authorization processing for users can be carried out when the service process node is at a preset process node. In one optional implementation of this embodiment, service risk control detection includes: If a service process node is detected to be at a preset process node, check whether the service processing performed on the user triggers service risk control detection. If so, perform credit authorization processing on the user to obtain credit authorization credentials.

[0021] Among them, preset process nodes refer to specific process nodes that are pre-configured to trigger service risk control detection. For example, in the recruitment service scenario, preset process nodes can be the initial screening passing node in the recruitment service process, such as the node where the applicant passes the initial resume screening and enters the interview stage; it can also be the initial interview passing node, such as the node where the applicant passes the initial interview; it can also be the background check node, such as the node where the recruiter will conduct a background check on the applicant after the applicant has completed all interview stages, or the node where the background check is conducted on the applicant when the applicant enters the onboarding preparation stage; in addition, in the recruitment service scenario, preset process nodes can also be special risk control nodes for specific positions, such as the security review initiation node for positions involving confidentiality, or the overseas background investigation node for applicants with overseas work experience.

[0022] Specifically, during the risk control process for users, service process nodes can be monitored in real time. If a service process node is detected to be at a preset process node, it is further checked whether the service processing for the user triggers service risk control detection. If so, it indicates that risk control processing needs to be initiated for the user performing the current service processing, and credit authorization processing can be performed for the user to obtain the corresponding credit authorization certificate; if not, it indicates that risk control processing does not need to be initiated for the user performing the current service processing, and no processing is required.

[0023] For example, in a recruitment service scenario, during the risk control process for job applicants, the service risk control detection can be recruitment risk control detection. Specifically, during the recruitment risk control detection process, if the recruitment process node of the job is detected to be at a preset process node, it is checked whether the job has triggered recruitment risk control detection. If so, the credit authorization process for the job applicant is carried out to obtain a credit authorization certificate.

[0024] In the specific execution process, during the detection of whether service processing for a user triggers service risk control detection, service risk assessment can be performed based on the user's service data, service processing data, and / or service evaluation task text. In one optional implementation of this embodiment, detecting whether service processing for a user triggers service risk control detection includes: Input the user's service data, service processing data, and service evaluation task text into the risk control trigger detection model to conduct service risk assessment. If the service assessment result indicates that risk detection is required, then the service risk control detection is triggered.

[0025] User service data refers to data related to a user's own attributes and basic information about the user in the service system. For example, in a recruitment service scenario, user service data can be user resume data, which may include the user's education, work experience, professional qualifications and / or historical application records, as well as other data related to the user's resume. As another example, in a rental service scenario, user service data can be user rental data, which may include the user's occupation, proof of income and / or historical performance records, as well as other data related to the user's rental records.

[0026] The service processing data refers to data related to the current service scenario, service task, and / or service type; for example, in a recruitment service scenario, the service processing data can be job data, which specifically includes data used to characterize the basic information, job attributes, and / or job characteristics of the job, such as job title, job category, salary range, work location, and / or core responsibility description; the service evaluation task text refers to prompt text used to guide the model to perform comprehensive analysis, such as prompt text used to guide the risk control trigger detection model to perform service risk assessment.

[0027] The risk control trigger detection model refers to a model used to determine whether risk detection for service processing needs to be initiated. The risk control trigger detection model can be an algorithm model, such as a model based on a decision tree algorithm, a model based on a neural network algorithm, or a model based on other algorithms. The input of the risk control trigger detection model includes user service data, service processing data, and / or service evaluation task text, and the output includes service evaluation results.

[0028] Specifically, in the process of detecting whether a service processing performed on a user triggers a service risk control detection, the user service data, service processing data, and service evaluation task text can be obtained and then input into the risk control trigger detection model. This allows the risk control trigger detection model to perform a service risk assessment and output a service assessment result. Subsequently, the service assessment result output by the risk control trigger model is obtained. If the service assessment result indicates that a risk detection is required, then it is determined that a service risk control detection has been triggered. Conversely, if the service assessment result indicates that a risk detection is not required, then it is determined that a service risk control detection has not been triggered.

[0029] In this process of service risk assessment, the risk control trigger detection model can include a text encoder, a feature fusion unit, a risk assessment network, and / or a threshold comparator. Specifically, the text encoder standardizes and encodes the unstructured user service data, service processing data, and service assessment task text, outputting standardized structured feature vectors and text semantic vectors. These vectors are then input into the feature fusion unit, which uses an attention-based neural network to calculate the correlation weights of the three types of features, weighting and concatenating the vectors to generate a multi-dimensional fused feature vector. Further, the fused feature vector is input into the risk assessment network, which uses a deep neural network (DNN) to extract high-order risk features, performs classification reasoning on the feature vectors based on preset risk threshold parameters, and outputs a risk assessment score. Subsequently, the threshold comparator determines whether risk control is triggered based on the risk assessment score and a preset threshold, outputting the service assessment result.

[0030] Using the previous example, in a recruitment service scenario, specifically in the process of detecting whether a job posting triggers recruitment risk control detection, the user's resume data, the job posting data, and the job evaluation task text can be input into the risk control trigger detection model to conduct a job risk assessment, thereby determining whether recruitment risk control detection is triggered. That is, in a recruitment service scenario, detecting whether a job posting triggers recruitment risk control detection includes: inputting the user's resume data, the job posting data, and the job evaluation task text into the risk control trigger detection model to conduct a job risk assessment; if the job assessment result indicates that risk detection is required, then recruitment risk control detection is determined to be triggered.

[0031] Similarly, in the recruitment risk control detection process, the risk control trigger detection model first standardizes and encodes unstructured user resume data, job data, and job evaluation task text using a text encoder, outputting standardized structured feature vectors and text semantic vectors. Then, these vectors are input into a feature fusion unit, which uses an attention-based neural network to calculate the correlation weights of the three types of features, weighting and concatenating the vectors to generate a multi-dimensional fused feature vector. Further, the fused feature vector is input into a risk assessment network, which uses a deep neural network to extract high-order risk features, combines them with preset risk threshold parameters to classify and infer the feature vectors, and outputs a risk assessment score. Finally, a threshold comparator determines whether risk control is triggered based on the risk assessment score and a preset threshold, outputting the service assessment result.

[0032] In this process, during the service risk assessment by the risk control trigger detection model, the model can perform risk prediction, anomaly detection, and / or authenticity detection based on user service data and / or service processing data. In one optional implementation of this embodiment, the service risk assessment includes: Based on service type tags, risk prediction is performed on service processing data, anomaly detection is performed on the adaptation between service processing data and user service data, and / or authenticity detection is performed on key data in user service data, and service evaluation results are determined based on risk prediction results, anomaly detection results and / or authenticity detection results.

[0033] Here, in the case of a recruitment service scenario, the service risk control assessment can specifically be a job risk assessment conducted within the recruitment service scenario. When the service risk control assessment is a job risk assessment, the above-mentioned service risk assessment process can be replaced by: predicting the risk of the recruitment job based on the job data, detecting anomalies in the matching between the job data and the user resume data, and / or verifying the authenticity of key data in the user resume data, and determining the job assessment result based on the risk prediction result, the anomaly detection result, and / or the authenticity detection result.

[0034] For example, in the process of job risk assessment in recruitment service scenarios, the risk level of the job can be assessed based on the job category, job level and permissions, job salary and / or data permissions in the job data; in the process of adaptation anomaly detection, if the job data and user resume data match too closely, such as exceeding the adaptation threshold, it may indicate that the applicant has fabricated a resume, and the adaptation anomaly detection result is determined to be an adaptation anomaly; alternatively, job data can be matched with user resume data to identify suspicious situations between the two, such as the applicant's qualifications being too short but achievements being too high, or the skills and experience in the user resume data not matching; in the process of authenticity detection, third-party interfaces can be connected to verify the key data in the user resume data.

[0035] In addition, in the process of detecting whether service processing triggers service risk control detection, in addition to conducting service risk assessment through the risk control trigger detection model, a large language model can also be called to detect whether service processing triggers service risk control detection. In this case, detecting whether service processing performed on users triggers service risk control detection includes: inputting user service data, service processing data and service assessment task text into the large language model to conduct service risk assessment. If the service assessment result is to conduct risk detection, then it is determined that service risk control detection has been triggered. Here, Large Language Model (LLM) refers to a pre-trained natural language model. Large language models can use foundation models or pre-trained models. The architecture of a large language model can be a neural network architecture with a large number of parameters, a Transform architecture, or other architectures. Specifically, a large language model can directly use a foundation model or a pre-trained model. Alternatively, it can be fine-tuned based on the foundation model or pre-trained model for the specific task of service risk assessment to obtain a large language model that can perform the specific task of service risk assessment. For example, in the process of service risk assessment using a large language model, the data standardization component first standardizes user service data, service processing data, and service assessment task text to obtain structured data and denoised plain text data. Then, the embedding layer converts the discrete fields of the structured data into low-dimensional dense vectors, and the Transformer encoder captures the semantic context of the text through a self-attention mechanism, converting the service assessment task text into a fixed-dimensional semantic vector. This maps the three types of data to the same vector space, outputting user service feature vectors, service processing feature vectors, and task text semantic vectors. Finally, a cross-attention mechanism... The Mechanism module calculates the attention weights between different modal vectors to quantify the strength of data associations. It then performs weighted concatenation and dimensional unification of the three types of vectors to generate a globally fused feature vector containing multi-source information. Based on this, the decoder of the large language model performs semantic decoding on the fused feature vector and combines it with a pre-trained risk control knowledge graph to perform risk correlation reasoning, outputting the probability distribution of two categories of labels: triggering risk control or not triggering risk control. Finally, the logic determiner compares the output probability with the preset risk control trigger threshold and outputs the service evaluation result.

[0036] Similarly, in the recruitment service scenario, during the process of detecting whether a job posting triggers recruitment risk control detection, the risk control trigger detection model can also be replaced by a large language model. Correspondingly, the large language model in the process of detecting whether a job posting triggers recruitment risk control detection can directly adopt a base model or a pre-trained model. Furthermore, based on the base model or pre-trained model, the base model or pre-trained model can be fine-tuned for the specific task of job risk assessment to obtain a large language model capable of performing the specific task of job risk assessment. During the job risk assessment process, the specific job risk assessment process of the large language model is similar to the service risk assessment process of the large language model described above, and will not be elaborated further in this embodiment.

[0037] In practical applications, during the process of calling the credit reporting platform to collect credit data, in order to further ensure that the service system's acquisition of credit data is within the scope of user authorization, enhance the compliance of data collection, and prevent unauthorized calls, the service authorization identifier of the service system can be verified and matched with the user and institution's contract notarization stored on the blockchain. In an optional implementation of this embodiment, after calling the credit reporting platform to collect credit data based on the risk control request sent by the service system carrying the user's credit authorization certificate, the method further includes: Based on the service authorization identifier carried in the risk control request, a service authorization verification request is sent to the trusted management and control node of the trusted data space to perform service authorization verification; If the verification is successful, proceed to step S204 below to obtain the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space.

[0038] Optionally, service authorization verification includes: querying the contract records between the user and the institution from the blockchain and parsing the contract to obtain the scope of contract authorization; checking whether the service authorization identifier matches the scope of contract authorization; if so, confirming that the verification is successful.

[0039] Trusted data space refers to a secure data space used for storing and processing specific data. Trusted data space only allows authorized trusted entities to operate on the relevant data within the authorized scope, ensuring that the relevant data is not accessed by unauthorized parties, thereby preventing data leakage or abuse. Contract notarization refers to an agreement between users and institutions regarding the authorization of data use. For example, contract notarization can be an employment contract between users and institutions authorizing the use of user credit data. After being signed by both users and institutions, it is uploaded to the blockchain to form an immutable and traceable notarization record. Here, the service authorization identifier can also be replaced by a service authorization certificate.

[0040] Specifically, after a user signs a contract with an institution in the service system, the contract can be written to the blockchain, generating a contract notarization between the user and the institution. Subsequently, during the risk control process for the user, after the risk control engine calls the credit reporting platform to collect credit data based on the risk control request sent by the service system carrying the user's credit authorization certificate, since the credit reporting platform has already uploaded the user's credit data to the data isolation space of the trusted data space, the risk control engine further sends a service authorization verification request to the trusted management node of the trusted data space based on the service authorization identifier carried in the risk control request. Subsequently, the trusted management node responds to the service authorization verification request and performs service authorization verification. If the verification is successful, it indicates that the service authorization identifier carried in the current risk control request is consistent with the contract authorization scope in the contract notarization between the user and the institution stored on the blockchain, and conforms to the data authorization scope agreed upon by the user. Then, the de-identified credit data uploaded by the credit reporting platform is obtained from the data isolation space of the trusted data space. Conversely, if the verification fails, it indicates that the current risk control request exceeds the data authorization scope of the user in the contract. In this case, the trusted management node rejects the risk control request, does not provide the user's de-identified credit data, and returns an authorization failure notification to the risk control engine. In this process, when the trusted management node responds to the service authorization verification request and performs service authorization verification, the trusted management node first queries the contract certificate between the user and the institution from the blockchain and parses the contract certificate to obtain the scope of contract authorization. Then, it checks whether the service authorization identifier matches the scope of contract authorization. If they match, it means that the current risk control request is within the scope of data authorization granted by the user in the contract, and the verification is confirmed to be successful. If not, it means that the current risk control request exceeds the scope of data authorization granted by the user in the contract, and the verification is confirmed to be unsuccessful.

[0041] Furthermore, in practical applications, to provide a data foundation for subsequent risk control processing and thus improve the accuracy of subsequent service risk detection, after calling the credit reporting platform to collect credit data, service processing data and user service data can also be passed to the privacy computing space. In one optional implementation of this embodiment, after calling the credit reporting platform to collect credit data based on the risk control request sent by the service system carrying the user's credit authorization certificate, the method further includes: Based on the access authorization identifier of the data sandbox synchronized by the service system, the service processing data corresponding to the service type tag carried in the risk control request and the user's user service data are read from the data sandbox. Service processing data and user service data are transmitted into the privacy computing space in encrypted form.

[0042] Among them, service type tags refer to tags used to identify specific service scenarios of service systems; for example, in a recruitment scenario, service type tags can be used to identify different job positions; data sandboxes refer to controlled isolation spaces used for data security isolation and controllable use. In a data sandbox, process restrictions on data access or processing can be implemented to prevent data leakage, abuse and / or unauthorized tampering; access authorization identifiers can specifically be access authorization tokens.

[0043] Specifically, during the risk control process, the risk control engine can also obtain the access authorization identifier of the data sandbox synchronized by the service system, construct a data read request based on the access authorization identifier, and send it to the data sandbox. After parsing and verifying the access authorization identifier, the data sandbox returns the service processing data corresponding to the service type tag carried in the risk control request and the user's user service data to the risk control engine if the verification is successful. Accordingly, the risk control engine reads the service processing data and user service data from the data sandbox. On this basis, the risk control engine can obtain the encryption key of the privacy computing space and transmit the service processing data and user service data to the privacy computing space in encrypted form. Subsequently, the privacy computing space can decrypt and verify the data, and if the verification is successful, store the service processing data and user service data in the privacy computing space.

[0044] For example, in a recruitment service scenario, job data corresponding to the job posting and resume data of the applicant can be read from the data sandbox and passed into the privacy computing space. Specifically, in a recruitment service scenario, after the credit data is collected by calling the credit reporting platform based on the risk control request sent by the recruitment service system carrying the credit authorization certificate of the applicant, the following operations can be performed: based on the access authorization identifier of the data sandbox synchronized by the recruitment service system, the job data corresponding to the job posting and resume data of the applicant carried in the risk control request are read from the data sandbox, and the job data and resume data are passed into the privacy computing space in encrypted form.

[0045] Step S204: Obtain the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space.

[0046] As mentioned above, based on the risk control request sent by the service system carrying the user's credit authorization certificate to call the credit reporting platform to collect credit data, here, the de-identified credit data uploaded by the credit reporting platform is obtained and sent to the privacy computing space.

[0047] In practice, during the risk control process, in order to ensure data security, strengthen user privacy protection, and enhance data processing compliance, de-identified credit data uploaded by the credit reporting platform is obtained from the data isolation space of the trusted data space, and the de-identified credit data is sent to the privacy computing space for subsequent risk control processing.

[0048] In the specific execution process, during the acquisition of de-identified credit data, the risk control engine can obtain a temporary access token from the trusted management node after the aforementioned service authorization verification is passed. The risk control engine can also determine the data isolation space by parsing the risk control request. Subsequently, the risk control engine can construct a data acquisition request based on the temporary access token and send the data acquisition request to the data isolation space. After verifying the permissions of the temporary access token, the data isolation space can read the de-identified credit data and return it to the risk control engine if the verification is successful. Correspondingly, the risk control engine obtains the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space; after that, it sends the de-identified credit data to the privacy computing space.

[0049] Trusted data space refers to a secure data space used for storing and processing specific data. Trusted data space only allows authorized trusted entities to operate on the relevant data within the authorized scope, ensuring that the relevant data is not accessed by unauthorized parties, thereby preventing data leakage or abuse; trusted data space can be used to achieve secure data sharing between different entities. Data isolation space refers to a logical partition within a trusted data space used to store specific data. Each data isolation space can correspond to a specific service type or data type. Each data isolation space can prevent cross-leakage of different data through physical isolation mechanisms or logical isolation mechanisms. De-identified credit data refers to user credit data that has undergone de-identification processing. De-identification processing refers to privacy protection processing of user credit data, specifically, it refers to hiding or obscuring data that can directly or indirectly identify a user's personal identity while retaining key data used for risk detection. For example, certain sensitive fields may be removed, such as deleting the user's exact address; or some characters may be masked, such as replacing some characters with specific symbols; or the data precision may be reduced, such as only retaining the year and month of the date, without retaining the specific date.

[0050] Step S206: Obtain the service risk control rules corresponding to the risk control request based on the service authorization of the service system, and input the service risk control rules and risk control model into the privacy computing space, so as to input the de-identified credit data and the service risk control rules into the risk control model for service risk detection.

[0051] As mentioned above, after obtaining and sending the de-identified credit data to the privacy computing space, the service risk control rules and risk control model are further input into the privacy computing space to perform service risk detection based on the risk control model combined with the service risk control rules and the de-identified credit data.

[0052] In practice, to improve the accuracy of subsequent service risk detection and match different service types with corresponding service risk control rules, the service risk control rules corresponding to the risk control request are first obtained based on the service authorization of the service system. On this basis, in order to strengthen the privacy protection of the aforementioned de-identified credit data and service risk control rules and improve the fairness of service risk detection, the service risk control rules and risk control model are passed into the privacy computing space. The risk control model is then invoked in the privacy computing space, and the de-identified credit data and service risk control rules are input into the risk control model to enable the risk control model to perform service risk detection. The risk detection results output by the risk control model are then obtained, which is to say, the risk detection results output by the privacy computing space are obtained.

[0053] The service risk control rules refer to the risk judgment logic or risk threshold standards preset for specific service scenarios and / or service types. For example, in the recruitment service scenario, the service risk control rules may record at least one of the following: job analysis field for job analysis, detection task field for recruitment risk detection (service risk detection), interpretable output field, and subject role field. The job analysis fields can be fields used to characterize the basic attributes, responsibilities, organizational authority, and / or industry characteristics of the job posting; such as job title, department, job level, work location, industry category, core responsibilities, and / or whether it involves financial approval authority. The detection task fields can be fields used to characterize the specific risk detection tasks performed on the applicant, such as credit information inquiry, educational background verification, and work experience consistency verification. The interpretable output fields can be fields used to characterize the templates and rules for generating risk detection results.

[0054] The risk control model refers to a model used to perform service risk detection tasks. The risk control model can be a large language model; or, the risk control model can also be an algorithm model, such as a model based on decision tree algorithm, or a model based on neural network algorithm, or a model based on other algorithms. The input of the risk control model includes de-identified credit data and service risk control rules, and the output includes risk detection results.

[0055] In the specific implementation process, to ensure the credibility and immutability of the service risk control rules used for service risk detection, these rules can be pre-stored on the blockchain. Simultaneously, to improve the accuracy and flexibility of risk control processing, different service risk control rules can be queried based on service type tags when different service types match different service risk control rules. In one optional implementation of this embodiment, obtaining the service risk control rules corresponding to the risk control request based on the service authorization of the service system includes: Based on the service type tag carried in the risk control request, submit a rule query request carrying the service type tag to the blockchain and obtain the service risk control rules returned by the query.

[0056] Optionally, service risk control rules are configured by the service system for service type tags and stored on the blockchain.

[0057] As mentioned above, service type tags are tags used to identify specific service scenarios of a service system; for example, in a recruitment scenario, service type tags can be used to identify different job positions.

[0058] Specifically, the service system can configure risk control rules and associate the different service risk control rules obtained from the configuration with the corresponding service type tags, and store the service risk control rules with the service type tags on the blockchain. On this basis, in the process of obtaining the service risk control rules corresponding to the risk control request based on the service authorization of the service system, after the risk control engine obtains the risk control request sent by the service system, it can submit a rule query request carrying the service type tag to the blockchain according to the service type tag carried in the risk control request. The blockchain responds to the rule query request and performs a service risk control rule query based on the service type tag. If the service risk control rule is obtained, it returns the service risk control rule to the risk control engine. Accordingly, here, the service risk control rule returned by the blockchain query is obtained.

[0059] For example, in a recruitment service scenario, during the process of obtaining service risk control rules through a query, a rule query request carrying the job posting can be submitted to the blockchain. For instance, obtaining the service risk control rules corresponding to the risk control request based on the service authorization of the service system includes: submitting a rule query request carrying the job posting to the blockchain based on the job posting carried in the risk control request, and obtaining the service risk control rules returned by the query; wherein, the service risk control rules are configured by the recruitment service system for the job posting and stored on the blockchain.

[0060] Furthermore, after submitting a rule query request to the blockchain, if the blockchain's response to the rule query request results in an empty service risk control rule query (i.e., if the blockchain query returns empty), to ensure that the subsequent service risk detection process is not interrupted due to the lack of service risk control rules and to improve the stability of service risk detection, a risk identification model can be introduced. This allows for the configuration of service risk control rules through the risk identification model and rule templates. In an optional implementation of this embodiment, obtaining the service risk control rules corresponding to the risk control request based on the service authorization of the service system further includes: If the query returns empty, the service system’s risk identification model is authorized for permission verification. After the verification is successful, the risk identification model and the rule template read from the data sandbox are passed into the privacy computing space so that the service processing data and the rule template are input into the risk identification model to identify risk items and obtain risk items. Risk control rules are constructed based on risk items and rule templates, and service risk control rules are obtained by configuring parameters of the risk control rules.

[0061] The risk identification model refers to a model used for identifying risk items; the risk identification model can be a large language model; or, the risk identification model can also be an algorithm model, such as a model based on a decision tree algorithm, or a model based on a neural network algorithm, or a model based on other algorithms; the input of the risk identification model includes service processing data and rule templates, and the output includes risk items; As mentioned above, service processing data refers to data related to the current service processing scenario or service type; for example, in a recruitment service scenario, service processing data can be job data, which can specifically include data used to characterize the basic information, job attributes and / or job characteristics of the job being recruited, such as job title, job category, salary range, work location and / or core responsibility description.

[0062] Specifically, in the process of obtaining the service risk control rules corresponding to the risk control request based on the service system's service authorization, if the query result returned by the aforementioned blockchain is empty, the service system's risk identification model is first verified for permissions to check whether the service system has the permission to call the risk identification model. After the verification is passed, the risk identification model provided by the service system and the applicable rule template are obtained from the data sandbox, and the risk identification model and rule template are passed into the privacy computing space. In the privacy computing space, the service processing data and rule template are input into the risk identification model to identify risk items and obtain risk items. Subsequently, risk control rules are constructed based on the identified risk items and rule templates, and the risk control rules are further configured with parameters to obtain the service risk control rules.

[0063] For example, in the process of risk identification, the risk identification model first extracts risk-related features from the service processing data through a feature extractor, and converts them into feature vectors through one-hot encoding and normalization. Then, the text framework of the rule template is converted into a semantic vector through a Transformer encoder, and fused with the feature vector to obtain fused features. After that, the classifier identifies the corresponding core risk items based on the fused features and outputs a set of structured risk items.

[0064] Here, risk item identification refers to the process of identifying service-type related risk items based on service processing data and rule templates. Specifically, risk item identification may include at least one of the following: identifying financial risk items, identifying data security risk items, identifying compliance risk items, and identifying operation permission risk items. For example, in a recruitment service scenario, risk item identification may be the process of identifying risk items related to recruitment positions based on job data and initialization templates.

[0065] Continuing with the previous example, in a recruitment service scenario, during the process of obtaining service risk control rules, if the blockchain response to a rule query request carrying a job posting returns an empty result, the job posting data and rule template can be input into a large language model to identify risk items and obtain risk items. Based on this, service risk control rules can be configured. For example, the service risk control rules corresponding to the service authorization request from the service system include: if the query returns an empty result, inputting the job posting data and rule template into a large language model to identify risk items, and constructing a risk detection model based on the identified risk items and rule template. For example, the risk detection template can be configured with risk parameters to obtain service risk control rules; or, for another example, the service risk control rules corresponding to the risk control request can be obtained based on the service authorization of the service system, including: if the query returns empty, the risk identification model of the recruitment service system is verified for permissions, and after the verification is passed, the risk identification model and the rule template read from the data sandbox are passed into the privacy computing space, so as to input the job data of the recruitment position and the rule template into the risk identification model to identify risk items and obtain risk items; risk control rules are constructed according to risk items and rule templates, and risk parameters are configured for the risk control rules to obtain service risk control rules.

[0066] It should be added that, in the process of obtaining service risk control rules corresponding to service authorization requests based on the service system described above, the risk items corresponding to different service types may differ. In this case, to ensure the accuracy of subsequent service risk detection, service processing data corresponding to the service type tag and an initial risk detection template can be obtained. In this case, the implementation method of obtaining service risk control rules corresponding to service authorization requests based on the service system described above can also be replaced by the following: the service risk control rules corresponding to service authorization requests based on the service system include: service processing data corresponding to the service type tag carried in the service authorization request and an initial risk detection template. The service processing data is transmitted to the privacy computing space in encrypted form. The risk identification model of the service system undergoes permission verification. After successful verification, the risk identification model and the rule template read from the data sandbox are transmitted to the privacy computing space. The service processing data and rule template are then input into the risk identification model to identify risk items. Risk control rules are constructed based on the risk items and rule templates, and parameters are configured to obtain service risk control rules. The process of configuring the replaced service risk control rules is similar to the process described above. Refer to the process described above for configuring service risk control rules and make adaptive modifications, changes, or deletions as needed during actual execution.

[0067] In the process of configuring parameters to obtain service risk control rules, in order to achieve fine-grained configuration of service risk control rules and improve the accuracy and efficiency of subsequent service risk detection, risk parameters can be determined based on service processing data and / or user service data, and service risk control rules can be further determined based on risk parameters. In one optional implementation of this embodiment, the service risk control rules are obtained by configuring parameters of the risk control rules, including: Based on the service processing data corresponding to the risk control request, perform risk prediction for the service type, perform anomaly detection for the adaptation between service processing data and user service data, and / or perform authenticity detection on key data in user service data; Risk parameters are determined based on risk prediction results, adaptation anomaly detection results, and / or authenticity detection results, and service risk control rules are obtained by configuring risk control rules according to risk parameters.

[0068] Here, in the context of a recruitment service scenario, the configuration of service risk control rules can specifically be based on determining risk parameters using job data and / or user resume data, and further determining service risk control rules based on these risk parameters. In this case, the process of configuring risk control rules to obtain service risk control rules described above can be replaced by: performing risk prediction on the job data corresponding to the risk control request, performing anomaly detection on the matching between job data and user resume data, and / or performing authenticity checks on key data in the user resume data; determining risk parameters based on the risk prediction results, anomaly detection results, and / or authenticity check results, and configuring risk control rules according to the risk parameters to obtain service risk control rules. For example, in recruitment service scenarios, during the process of determining risk parameters based on job data and / or user resume data, and further determining service risk control rules based on these risk parameters, risk prediction can be performed on job data to obtain risk prediction results. For instance, by analyzing job data such as job title, industry, job description, and scope of authority, the types of risks that the job may face can be predicted. Alternatively, anomaly detection can be performed between job data and user resume data to obtain anomaly detection results. For example, comparing the matching degree between the job requirements of the recruitment position and the applicant's resume to see if there are logical conflicts or abnormal patterns. Furthermore, authenticity checks can be performed on key data in the user resume data to obtain authenticity check results. For instance, verifying the authenticity of key data such as the applicant's education, work experience, job title, and / or project achievements. For example, in the process of detecting anomalies in the matching of job data and user resume data, an anomaly may be that the job applicant fabricated false user resume data based on the job requirements of the job posting. In this case, the anomaly detection result may be an anomaly detection result of excessive matching or doubt about the resume.

[0069] It should be noted that, in the process of configuring parameters to obtain service risk control rules, the three specific methods provided above—risk prediction based on service processing data, anomaly detection of the adaptation between service processing data and user service data, and authenticity detection of key data in user service data—can be selected for execution in practice, depending on the actual needs. Furthermore, other specific methods for configuring risk parameters can be introduced and combined with the above methods, or the combined implementation can be adapted to form new methods for configuring risk parameters. Similarly, the three specific methods provided above—risk prediction based on job data, anomaly detection of the adaptation between job data and user resume data, and authenticity detection of key data in user resume data—can be implemented in practice by selecting at least one, at least two, or all three methods as needed. Alternatively, other specific methods for configuring risk parameters can be introduced and combined with the above methods, or the combined implementation can be adapted to form a new method for configuring risk parameters.

[0070] In specific implementation, based on the above-mentioned acquisition of service risk control rules and the input of service risk control rules and risk control models into the privacy computing space, de-identified credit data and service risk control rules are input into the risk control model for service risk detection. Specifically, in the process of service risk detection, service risk detection can be based on credit risk control rules, service risk detection can be based on a dual matching verification mechanism of structured rule matching and semantic matching, attribution analysis can be performed to improve the interpretability of service risk detection results, and relevant risk trend prediction can be performed through network retrieval analysis to achieve service risk detection. The following provides four implementation methods for service risk detection, and provides a detailed explanation of each of the four implementation methods.

[0071] (1) Implementation method one: In the specific implementation process, to improve the processing efficiency of service risk detection, service risk detection can be based on credit risk control rules. In one optional implementation method provided in this embodiment, service risk detection is implemented in the following way: The user credit characteristics of the de-identified credit data are matched with the credit risk control rules included in the service risk control rules, and the service processing data and user service data are matched for anomaly detection. Risk detection results are generated based on credit matching results, anomaly detection results, and risk weight parameters included in service risk control rules.

[0072] Specifically, during service risk detection, when anonymized credit data and service risk control rules are input into the risk control model, the anonymized credit data can be parsed to obtain user credit characteristics. The risk control model then performs credit matching based on the credit risk control rules included in the service risk control rules to obtain credit matching results. Additionally, it performs anomaly detection on the service processing data and the user's service data to obtain anomaly detection results. Based on the obtained credit matching results and anomaly detection results, the risk weight parameters included in the service risk control rules are further combined to generate risk detection results. The risk detection results can be risk level and / or risk score.

[0073] In this process, during service risk detection by the risk control model, the service risk control rules are first parsed using a rule parser to extract structured credit risk control rules and their corresponding risk weight parameters. Then, a field mapper aligns the fields of the de-identified credit data with the fields of the credit risk control rules, and matches and maps the associated fields of service processing data with user service data to ensure data dimensional consistency. The result is the output of the field-aligned de-identified credit data and a set of structured credit risk control rules. Next, a structured feature extractor processes the de-identified credit data to generate user credit feature vectors. Finally, a rule matching engine uses a pattern matching algorithm to match the user credit feature vectors with each credit risk control rule. The system performs a dimensional comparison and outputs the matching result for each rule. Further, the similarity calculation module calculates the fit between the service processing data and the user service data, and uses the Isolation Forest algorithm to determine anomalies in the fit, outputting an anomaly score. Subsequently, a weighted summation calculator, based on risk weight parameters, weights and sums the confidence levels of the credit rule matching results to generate a credit risk score. A fusion neural network then non-linearly fuses the credit risk score and the anomaly score to output a comprehensive risk score. Finally, a risk grader maps the comprehensive risk score to the corresponding risk level, and a structured result generator integrates the credit matching results, anomaly score, comprehensive risk score, and risk level to generate and output the risk detection result.

[0074] For example, in a recruitment service scenario, credit matching results can be obtained by performing credit matching on user credit characteristics based on the job credit rules included in the service risk control rules. Anomaly detection results can be obtained by performing adaptation anomaly detection on the job data and user resume data. Finally, a risk detection result is generated based on the credit matching result, the anomaly detection result, and the risk weight parameters. Therefore, in a recruitment service scenario, service risk detection is implemented as follows: credit matching is performed on user credit characteristics based on the job credit rules included in the service risk control rules, and adaptation anomaly detection is performed on the job data and user resume data. A risk detection result is generated based on the credit matching result, the anomaly detection result, and the risk weight parameters included in the service risk control rules. Similarly, in a recruitment service scenario, the service risk detection process performed by the risk control model is similar to the service risk detection process performed by the risk control model described above, and will not be elaborated further in this embodiment.

[0075] (2) Implementation method two: In the specific implementation process, to improve the accuracy of service risk detection and the reliability of the generated user risk rating during service risk detection, sub-data features can be introduced to avoid one-dimensional judgments, and dual matching verification can be performed to reduce the false judgment rate. In one optional implementation method provided in this embodiment, service risk detection includes: The service risk control rules are parsed to obtain risk rules and risk weight parameters, and the risk weight parameters are mapped to the corresponding risk rules. Extract sub-data features from the sub-data under the data dimension corresponding to the risk rules in the de-identified credit data, and fuse the sub-data features to obtain fused features; Structured rule matching is performed on risk rules and fusion features, and semantic matching is also performed on risk rules and fusion features. User risk ratings are generated based on the rule matching results, semantic matching results, and risk weight parameters.

[0076] The sub-data refers to the smallest unit of information with independent semantics extracted from user service data, credit data, and / or other third-party data. For example, in a recruitment service scenario, sub-data can be the smallest unit of information with independent semantics extracted from user resume data, credit data, or other third-party data. For instance, in work experience data, "company name," "employment start and end dates," "job title," and "main responsibilities" can be different sub-data items; in credit data, "current total debt" and "longest overdue days" can also exist as sub-data items. The sub-data features refer to the feature vectors that can be used for model recognition after standardizing, numerically converting, or semantically analyzing the sub-data; for example, converting "job title" into "job level (junior / intermediate / senior)" and "management authority" into "have / have no management authority".

[0077] Specifically, in the process of service risk detection, the risk detection model first parses the input service risk control rules to obtain risk rules and risk weight parameters, and maps the risk weight parameters to the corresponding risk rules. Then, it determines the data dimensions corresponding to each risk rule in the de-identified credit data, extracts specific sub-data from it, further processes the sub-data to obtain sub-data features, and on this basis, performs feature fusion on multiple sub-data features to obtain fused features. Subsequently, a matching mechanism can be executed in parallel: on the one hand, the numerical or classification results in the fused features are matched with the thresholds or conditions in the risk rules using structured rules to obtain rule matching results; on the other hand, semantic matching is performed on the risk rules and fused features to analyze the semantic consistency between user service data and service processing data, and obtain semantic matching results. Based on the rule matching results, semantic matching results, and risk weight parameters, a user risk rating is generated.

[0078] For example, in the process of service risk detection, the risk control model first uses a rule syntax parser with a structured parsing algorithm to break down the logical judgment conditions in the service risk control rules, extract independent risk rules and their corresponding risk weight parameters, and binds the risk weight parameters to the corresponding risk rules based on a key-value association mechanism using a weight parameter mapper. Secondly, it uses a data dimension filter to select corresponding dimension sub-data from the de-identified credit data, and uses a feature extractor to perform label encoding on discrete sub-data and normalization on continuous sub-data, converting them into standardized sub-data feature vectors and outputting a multi-dimensional sub-data feature vector set. Then, it uses the attention mechanism layer of the feature fusion neural network to calculate the association weight between each sub-data feature vector and the risk assessment target, strengthening the core... The system identifies risk dimensions and uses a fully connected layer to concatenate, compress, and non-linearly transform the weighted feature vectors, generating a fused feature vector containing information from all risk dimensions. Based on this, a structured matching engine employs a pattern matching algorithm to compare the sub-data features in the fused feature vector with the logical judgment conditions of the risk rules dimension by dimension, outputting the rule matching result for each rule. A Transformer semantic encoder converts text-based risk rules into semantic vectors and uses a semantic similarity calculator to calculate the similarity between the fused feature vector and the rule semantic vectors, outputting a semantic matching score. Finally, a weighted fusion calculator calculates the comprehensive risk score, and a risk grader maps the comprehensive risk score to a user risk rating based on a preset threshold and outputs the rating.

[0079] (3) Implementation method three: In the specific implementation process, to improve the understandability and traceability of the risk detection results obtained during service risk detection, attribution analysis can also be performed; in an optional implementation method provided in this embodiment, service risk detection further includes: Risk decision rules are obtained by extracting decision rules from the risk detection network in the risk control model. Perform attribution analysis on user risk ratings and generate a visual attribution chain; Structured detection results are generated based on user risk ratings, risk decision-making rules, and visualized attribution links.

[0080] Specifically, decision rules are extracted from the risk detection network in the risk control model that performs service risk detection, and the internal judgment logic of the risk detection network is restored to obtain risk decision rules. At the same time, attribution analysis is performed on user risk ratings, and a visual attribution link is generated based on the attribution analysis results to show the reasoning process from data input to risk output. On this basis, structured risk result data is generated based on user risk ratings, risk decision rules, and visual attribution links, and the structured risk result data is used as the risk detection result.

[0081] For example, in the process of service risk detection, the risk control model first extracts decision rules from the risk detection network in the risk control model to obtain risk decision rules. Then, based on the SHAP / LIME interpreter, it quantifies the contribution of sub-data features and fusion features of sub-data in each data dimension to risk rating. The feature importance calculator sorts out the causal relationship between core features, risk decision rules and user risk rating. The visualization renderer transforms the causal relationship into a visual attribution link. Finally, the result assembler integrates the data according to preset fields to form a standardized data structure. The natural language generator transforms the core features and decision rule matching logic into explanatory text, and finally outputs the structured detection results.

[0082] Similarly, the service risk detection process described above can also be applied to recruitment service scenarios. In this case, service risk detection for job applicants in recruitment service scenarios includes: extracting decision rules from the risk detection network to obtain risk decision rules; performing attribution analysis on user risk ratings and job risk trend data, and generating a visualized attribution chain based on the attribution analysis results; and generating structured risk result data as the risk detection result based on user risk ratings, risk decision rules, and the visualized attribution chain. Similarly, the service risk detection process performed by the risk control model in recruitment service scenarios is similar to the service risk detection process performed by the risk control model described above, and will not be elaborated further in this embodiment.

[0083] Furthermore, during service risk detection, feature importance quantification can be performed. For example, the feature importance of each input factor affecting the risk detection results can be quantified to assess the contribution of each input factor to the risk detection results, thereby obtaining quantitative feature importance data. In this case, with the introduction of a feature importance quantification mechanism, the aforementioned service risk detection includes: quantifying the feature importance of user risk ratings corresponding to job analysis task fields, and job risk trend data corresponding to job analysis fields and topic role fields to obtain quantitative feature importance data; extracting decision rules from the risk detection network used for service risk detection to obtain risk decision rules; performing attribution analysis on user risk ratings and job risk trend data, and generating a visualized attribution chain based on the attribution analysis results; and generating structured risk result data as the risk detection result based on the quantitative feature importance data, risk decision rules, and visualized attribution chain.

[0084] It should be noted that, during the service risk detection process, the aforementioned decision rule extraction and attribution analysis processes can be performed in practice by selecting at least one of them as needed. When at least one of the decision rule extraction and attribution analysis processes is performed, at least one of user risk rating, risk decision rules, and visualized attribution links can be used to generate structured risk result data. That is, service risk detection includes: extracting decision rules from the risk detection network in the risk control model to obtain risk decision rules, and / or performing attribution analysis on user risk ratings and generating visualized attribution links; and generating structured risk result data based on user risk ratings, risk decision rules, and / or visualized attribution links as the risk detection result. Alternatively, other specific methods for service risk detection can be introduced and combined with the above methods, or the combined implementation can be adapted to form a new processing method for service risk detection. For example, feature importance quantification can be performed on the type analysis field for service type analysis, the user risk rating corresponding to the type analysis field, and / or the service risk trend data corresponding to the topic role field to obtain feature importance quantification data; decision rules can be extracted from the risk detection network for user risk detection to obtain risk decision rules; and structured detection results can be generated based on feature importance quantification data and risk decision rules.

[0085] (4) Implementation method four: In the specific execution process, if the above-mentioned service risk control rules record at least one of the following: a type analysis field for service type tag analysis, a detection task field for service risk detection, an interpretable output field, and a subject role field, then service risk trend prediction can be performed based on service-related data to obtain service risk trend data. This allows for analysis of relevant service trends via network retrieval, and further service risk detection can be performed in conjunction with these trends. In one optional implementation of this embodiment, service risk detection includes: Extract service element tags from the type analysis field and the topic role field, and retrieve service-related data from multiple data retrieval dimensions based on the service element tags; A service data association network is constructed based on the retrieved service-related data, and service risk trend prediction is performed based on the service data association network to obtain service risk trend data.

[0086] Optionally, the service risk control rules may record at least one of the following: a type analysis field for service type analysis, a detection task field for service risk detection, an interpretable output field, and a subject role field.

[0087] Specifically, firstly, element extraction is performed on the type analysis field and the topic role field to identify service element tags that can characterize the core content and / or risk characteristics of the service type. Then, based on the service element tags, service-related data related to the service type is retrieved from multiple data retrieval dimensions. Based on this, while focusing on individual user service data, the potential risk indicators in historical or real-time data that are potentially related to the current service are assessed. Further, a service data association network is constructed based on the retrieved service-related data. Subsequently, service risk trend prediction is performed based on the service data association network to obtain service risk trend data, which serves as the risk detection result.

[0088] For example, in the process of service risk detection, the risk control model first uses a Transformer encoder to semantically encode the text information of two types of fields, extracts key entities through a named entity recognition module, and matches the key entities with a preset service element tag library through a tag mapper to generate service element tags. Secondly, a multi-dimensional retrieval engine uses service element tags as search keywords to retrieve relevant data from various data indexes across sources, and a data filtering module filters the search results according to data validity thresholds to obtain service-related data. Then, an entity association tool identifies core entities and their relationships in different service-related data based on a data association rule library, and a weighted service data association network is constructed using a Graph Neural Network (GNN) with core entities as nodes and relationships as edges. Subsequently, a Graph Convolutional Network (GCN) is used to propagate and aggregate features of the service data association network, extracting global risk association features at the network level. Finally, a time-series prediction module combines the user's current credit characteristics and corresponding risk weights to perform time-series modeling based on historical risk trend data and predict service risk trends, outputting service risk trend data.

[0089] Similarly, the service risk detection process described above can also be applied to recruitment service scenarios. In this case, the service risk detection for job applicants in the recruitment service scenario includes: extracting job element tags from the job analysis field and the topic role field; retrieving job-related data from multiple data retrieval dimensions based on the job element tags; constructing a job data association network based on the retrieved job-related data; and predicting job risk trends based on the job data association network to obtain job risk trend data. Correspondingly, the service risk control rules record at least one of the following: the job analysis field for job analysis, the detection task field for user risk detection, the interpretable output field, and the topic role field. Similarly, in the recruitment service scenario, the service risk detection process performed by the risk control model is similar to the service risk detection process performed by the risk control model described above, and will not be elaborated further in this embodiment.

[0090] Correspondingly, in recruitment service scenarios, during the process of service risk detection for job applicants, the process begins by extracting elements from the job analysis field and thematic role field to identify job element tags that can characterize the core functions and / or risk features of the recruited position. Then, based on these job element tags, relevant job data is retrieved from multiple data retrieval dimensions. This involves assessing whether there are any risk indicators in the industry environment of the recruited position, while also considering the individual applicant's resume data. Furthermore, a job data association network is constructed based on the retrieved job-related data. For example, using the current recruited position as the core node, multiple dimensions of entities related to the recruited position, such as industry, function, and / or law, are connected, and the strength of their associations and their influence paths are established. Subsequently, job risk trend prediction is performed based on the job data association network to obtain job risk trend data.

[0091] It should be noted that the four methods for service risk detection provided above can be combined in any way as needed during actual implementation, or can be adapted, modified, changed, or deleted before being combined in any way. For example, service risk detection includes: parsing service risk control rules to obtain risk rules and risk weight parameters, and mapping the risk weight parameters to the corresponding risk rules; extracting sub-data features from the de-identified credit data under the data dimension corresponding to the risk rules, and fusing the sub-data features to obtain fused features; performing structured rule matching between risk rules and fused features, and then matching the risk rules with the fused features. The system performs semantic matching on features and generates user risk ratings based on rule matching results, semantic matching results, and risk weight parameters. It extracts service element tags from type analysis fields and topic role fields, and retrieves service-related data from multiple data retrieval dimensions based on these tags. A service data association network is constructed based on the retrieved service-related data, and service risk trend prediction is performed based on this network to obtain service risk trend data. Attribution analysis is performed on user risk ratings and service risk trend data, and a visualized attribution chain is generated based on the attribution analysis results. Finally, structured risk result data is generated based on user risk ratings and the visualized attribution chain as the risk detection result.

[0092] Step S208: Synchronize the risk detection results output by the privacy computing space to the service system.

[0093] In specific implementation, based on the risk detection results output by the risk control model for service risk detection, i.e., based on the risk detection results output by the privacy computing space, the risk detection results output by the privacy computing space are synchronized to the service system.

[0094] It should be noted that the credit-based risk control processing implementation method provided in this embodiment can be applied to different service scenarios, such as recruitment service scenarios. In this case, when applying the credit-based risk control processing implementation method to the recruitment service scenario, steps S202 to S208 can be adapted or changed according to the actual execution needs. For example, in the recruitment service scenario, steps S202 to S208 can be replaced as follows: based on the risk control request sent by the recruitment service system carrying the applicant's credit authorization certificate, call the credit reporting platform to collect credit data; obtain the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space; obtain the service risk control rules corresponding to the risk control request based on the service authorization of the recruitment service system, and input the service risk control rules and risk control model into the privacy computing space to input the de-identified credit data and service risk control rules into the risk control model for recruitment risk detection; synchronize the risk detection results output by the privacy computing space to the recruitment service system. Similarly, when the above-mentioned credit-based risk control processing is applied to a recruitment service scenario, the various optional implementation methods and feasible execution methods in steps S202 to S208 can be replaced with the implementation methods in the recruitment service scenario, or the various optional implementation methods and feasible execution methods in steps S202 to S208 can be adapted and replaced with the implementation methods in the recruitment service scenario. Likewise, the various optional implementation methods and feasible execution methods in the recruitment service scenario can also be combined in any form according to the actual execution needs, or can be combined in any form after being adapted or changed according to the actual execution needs.

[0095] It should be added that each optional implementation method and each feasible execution method in steps S202 to S208 provided in this embodiment can be executed independently as needed, or they can be combined and referenced with each other. At the same time, each specific execution step in each optional implementation method or each feasible execution method can also be executed independently or combined as needed. The execution conditions of "if" or "under what circumstances" involved in each step or operation can be directly deleted, and the subsequent operations after the execution conditions are determined. Similarly, when the above-mentioned credit-based risk control processing is applied to a recruitment service scenario, after steps S202 to S208 are replaced with the implementation method under the recruitment service scenario, each optional implementation method and each feasible execution method in the replaced steps S202 to S208 can be executed independently as needed, or they can be combined and referenced together. At the same time, each specific execution step in each optional implementation method or each feasible execution method can also be executed independently or combined as needed. The execution conditions of "if" or "under what circumstances" involved in each step or operation can be directly deleted. The subsequent operations after the execution conditions are not specifically limited in this embodiment.

[0096] In summary, the credit-based risk control method provided in this embodiment, during the risk control process for users, firstly, based on the risk control request sent by the service system carrying the user's credit authorization certificate, calls the credit reporting platform to collect credit data and obtains the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space. This de-identified credit data is then sent to the privacy computing space. Further, based on the service authorization of the service system, the corresponding service risk control rules are obtained, and the service risk control rules and risk control model are input into the privacy computing space. The de-identified credit data and service risk control rules are then input into the risk control model for service risk detection. Finally, the risk detection results output by the privacy computing space are synchronized to the service system. This approach achieves user-based risk control from the user's credit data, and improves the flexibility and comprehensiveness of user-based risk control.

[0097] The following example uses a credit-based risk control method provided in this embodiment to illustrate its application in a risk control scenario. Figure 3 The credit-based risk control method provided in this embodiment will be further explained below. Figure 3 The credit-based risk control method applied to risk control scenarios includes the following steps.

[0098] Step S302: Based on the risk control request sent by the service system carrying the user's credit authorization certificate, call the credit reporting platform to collect credit data.

[0099] Optional, the risk control request is sent after the service system performs service risk control detection.

[0100] Step S304: Based on the access authorization identifier of the data sandbox synchronized by the service system, read the service processing data corresponding to the service type tag carried in the risk control request and the user's user service data from the data sandbox.

[0101] Step S306: Transfer the service processing data and user service data into the privacy computing space in encrypted form.

[0102] Step S308: Send a service authorization verification request to the trusted management node of the trusted data space according to the service authorization identifier carried in the risk control request to perform service authorization verification; if the verification is successful, proceed to step S310 below.

[0103] Step S310: Obtain the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space.

[0104] Step S312: Submit a rule query request carrying the service type tag to the blockchain based on the service type tag carried in the risk control request; if the query returns empty, proceed to step S314 below.

[0105] Step S314: Perform permission verification on the risk identification model of the service system, and after the verification is passed, transfer the risk identification model and the rule template read from the data sandbox into the privacy computing space.

[0106] Step S316: Input the service processing data and rule template into the risk identification model in the privacy computing space to identify risk items and obtain risk items.

[0107] Step S318: Construct risk control rules based on risk items and rule templates, configure parameters for the risk control rules to obtain service risk control rules, and input the service risk control rules and risk control model into the privacy computing space.

[0108] Step S320: Input the de-identified credit data and service risk control rules into the risk control model in the privacy computing space to perform service risk detection.

[0109] Service risk detection includes: matching the user credit characteristics of de-identified credit data with the credit risk control rules included in the service risk control rules, and performing anomaly detection on the service processing data and user service data; generating risk detection results based on the credit matching results, anomaly detection results, and risk weight parameters included in the service risk control rules.

[0110] Step S322: Synchronize the risk detection results output by the privacy computing space to the service system.

[0111] It should be noted that any one or more steps in steps S302 to S322 can be combined with any one or more steps in steps S202 to S208 to form a new implementation method according to the needs of implementation and deployment. In addition, any one or more technical features in steps S302 to S322 can be selected and combined with any one or more technical features provided in steps S202 to S208 to form a new implementation method according to the actual deployment needs. Alternatively, any one or more technical features in steps S302 to S322 can be replaced with any one or more technical features provided in steps S202 to S208 to form a new implementation method according to the actual deployment needs. These will not be elaborated on here.

[0112] The following example uses a credit-based risk control method provided in this embodiment in a recruitment service scenario as an example, combined with... Figure 4 The credit-based risk control method provided in this embodiment will be further explained below. Figure 4 The credit-based risk control method applied to recruitment service scenarios includes the following steps.

[0113] Step S402: Based on the risk control request sent by the recruitment service system carrying the credit authorization certificate of the applicant, call the credit reporting platform to collect credit data.

[0114] Optionally, the risk control request is sent after the recruitment service system performs recruitment risk detection; recruitment risk detection includes: when the recruitment process node of the recruitment position is detected to be at a preset process node, detecting whether the recruitment position triggers recruitment risk control detection; if so, performing credit authorization processing on the applicant user to obtain credit authorization certificate.

[0115] Step S404: Obtain the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space.

[0116] Step S406: Based on the access authorization identifier of the data sandbox synchronized by the recruitment service system, read the job data corresponding to the recruitment position carried in the risk control request and the user resume data of the applicant from the data sandbox, and pass the job data and user resume data into the privacy computing space.

[0117] Step S408: Perform permission verification on the risk identification model of the recruitment service system, and after the verification is passed, transfer the risk identification model and the rule template read from the data sandbox into the privacy computing space.

[0118] Step S410: Input the service processing data and rule template into the risk identification model in the privacy computing space to identify risk items and obtain risk items, and build risk control rules based on the risk items and rule template.

[0119] Step S412: Based on the job data, perform risk prediction for the job postings, detect anomalies in the matching between the job data and the user resume data, and verify the authenticity of key data in the user resume data.

[0120] Step S414: Based on the risk prediction results, adaptation anomaly detection results and authenticity detection results, determine the risk parameters, configure the risk control rules according to the risk parameters to obtain the service risk control rules, and input the service risk control rules and risk control model into the privacy computing space.

[0121] Step S416: Perform feature importance quantification on the user risk rating corresponding to the job analysis field, the job risk trend data corresponding to the job analysis field and the topic role field, and obtain feature importance quantification data.

[0122] The user risk rating is generated as follows: The service risk control rules are parsed to obtain risk rules and risk weight parameters, and the risk weight parameters are mapped to the corresponding risk rules; sub-data features of sub-data under the data dimension corresponding to the risk rules are extracted from the de-identified credit data, and the sub-data features are fused to obtain fused features; structured rule matching is performed between the risk rules and the fused features, and semantic matching is also performed between the risk rules and the fused features; the user risk rating is generated based on the rule matching results, semantic matching results, and risk weight parameters.

[0123] Step S418 involves extracting decision rules from the risk detection network in the risk control model to obtain risk decision rules, and performing attribution analysis on user risk ratings and job risk trend data to generate a visualized attribution chain.

[0124] Step S420: Based on the feature importance quantification data, risk decision rules, and visualized attribution links, structured risk outcome data is generated as the risk detection result.

[0125] Step S422: Synchronize the risk detection results output by the privacy computing space to the recruitment service system.

[0126] It should be noted that any one or more steps in steps S402 to S422 can be combined with any one or more steps in steps S202 to S208 to form a new implementation method according to the needs of implementation and deployment. In addition, any one or more technical features in steps S402 to S422 can be selected and combined with any one or more technical features provided in steps S202 to S208 to form a new implementation method according to the actual deployment needs. Alternatively, any one or more technical features in steps S402 to S422 can be replaced with any one or more technical features provided in steps S202 to S208 to form a new implementation method according to the actual deployment needs. These will not be elaborated on here.

[0127] The following is an example of a credit-based risk control processing device provided in this specification: In the above embodiments, a credit-based risk control processing method is provided, and correspondingly, a credit-based risk control processing device is also provided, which will be described below with reference to the accompanying drawings.

[0128] Reference Figure 5 The diagram illustrates an embodiment of a credit-based risk control processing device provided in this embodiment.

[0129] Since the apparatus embodiments correspond to the method embodiments, the descriptions are relatively simple. For relevant parts, please refer to the corresponding descriptions of the method embodiments provided above. The apparatus embodiments described below are merely illustrative.

[0130] This embodiment provides a credit-based risk control processing device that runs on a risk control engine. The device includes: The data acquisition module 502 is configured to call the credit reporting platform to collect credit data based on the risk control request sent by the service system carrying the user's credit authorization certificate; The data sending module 504 is configured to obtain the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space; The risk detection module 506 is configured to obtain the service risk control rules corresponding to the risk control request based on the service authorization of the service system, and to input the service risk control rules and the risk control model into the privacy computing space, so as to input the de-identified credit data and the service risk control rules into the risk control model for service risk detection; The result synchronization module 508 is configured to synchronize the risk detection results output by the privacy computing space to the service system.

[0131] For ease of description, the above devices are described by dividing them into various modules or units based on their functions. Of course, when implementing one or more of these specifications, the functions of each module or unit can be implemented in one or more software and / or hardware components, or a module that performs the same function can be implemented by a combination of multiple sub-modules or sub-units. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0132] The following is an example of a credit-based risk control processing device provided in this specification: Corresponding to the credit-based risk control processing method described above, and based on the same technical concept, one or more embodiments of this specification also provide a credit-based risk control processing device, which is used to execute the credit-based risk control processing method provided above. Figure 6 This is a schematic diagram of a credit-based risk control processing device provided for one or more embodiments of this specification.

[0133] This embodiment provides a credit-based risk control processing device, comprising: like Figure 6As shown, device 600 mainly consists of a communication interface 602, a user interface 604, a processor 606, and a data storage 608. These components are interconnected and communicate with each other via a system bus, network, or other connection mechanism 610. The communication interface 602 enables device 600 to communicate with other devices, access networks, and transmission networks via analog or digital modulation. For example, the communication interface 602 may include a chipset and antenna for wireless communication with a radio access network or access point. Furthermore, the communication interface 602 can be a wired interface such as Ethernet, Token Ring, or a USB port, or a wireless interface such as Wi-Fi, Bluetooth, Global Positioning System (GPS), or a wide-area wireless interface (e.g., WiMAX or LTE). Of course, the communication interface 602 can also support other forms of physical layer interfaces and standard or proprietary communication protocols. The communication interface 602 may also include multiple physical communication interfaces, such as Wi-Fi, Bluetooth, and wide-area wireless interfaces. The user interface 604 includes receiving user input and providing output to the user. Therefore, user interface 604 may include input components such as a keypad, keyboard, touch-sensitive or presence-sensitive panel, computer mouse, trackball, joystick, microphone, still camera, and video camera, and output components such as a display screen (which may be combined with a touch-sensitive panel), CRT, LCD, LED, display using DLP technology, printer, and other similar devices known or developed in the future. User interface 604 may also generate auditory output via speakers, speaker jacks, audio output ports, audio output devices, headphones, and other similar devices known or developed in the future. In some embodiments, user interface 604 may include software, circuitry, or other forms of logic capable of transmitting and receiving data from external user input / output devices. Additionally or alternatively, device 600 may support remote access from other devices via communication interface 602 or another physical interface (not shown). User interface 604 may be configured to receive user input, the position and movement of which may be indicated by indicators or cursors described herein. User interface 604 may also be configured as a display device for rendering or displaying text fragments.

[0134] Processor 606 may include one or more general-purpose processors and / or special-purpose processors. Data storage 608 may include one or more volatile and / or non-volatile storage components and may be integrated wholly or partially with processor 606. Data storage 608 may include removable and non-removable components.

[0135] Processor 606 is capable of executing program instructions 618 (e.g., compiled or uncompiled program logic and / or machine code) stored in data storage 608 to perform the various functions described herein. Data storage 608 may comprise a non-transitory computer-readable medium on which program instructions are stored, which, when executed by device 600, enable device 600 to perform any methods, processes, or functions disclosed in this specification and / or the accompanying drawings. Execution of program instructions 618 by processor 606 may result in processor 606 using data 612. For example, program instructions 618 may include an operating system 622 (e.g., an operating system kernel, device drivers, and / or other modules) installed on device 600 and one or more application programs 620 (e.g., a browser, social application, or game application). Similarly, data 612 may include operating system data 616 and application data 614. Operating system data 616 is primarily accessible to operating system 622, while application data 614 is primarily accessible to one or more application programs 620. Application data 614 may reside in a file system visible or hidden from the user of device 600. Application 620 can communicate with operating system 622 through one or more application programming interfaces (APIs). These APIs facilitate application 620 in reading and / or writing application data 614, transmitting or receiving information via communication interface 602, and receiving or displaying information on user interface 604. In some terms, application 620 may be simply referred to as "app". Furthermore, application 620 can be downloaded to device 600 through one or more online app stores or app markets. However, applications can also be installed on device 600 in other ways, such as through a web browser or a physical interface on device 600 (e.g., a USB port).

[0136] In one specific embodiment, the credit-based risk control processing device includes a memory and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for the credit-based risk control processing device, and is configured to be executed by one or more processors. The one or more programs include computer-executable instructions for performing the following: Based on the risk control request sent by the service system carrying the user's credit authorization certificate, the credit reporting platform is invoked to collect credit data; Obtain de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space; Based on the service authorization of the service system, the service risk control rules corresponding to the risk control request are obtained, and the service risk control rules and risk control model are transmitted into the privacy computing space so as to input the de-identified credit data and the service risk control rules into the risk control model for service risk detection; The risk detection results output by the privacy computing space are synchronized to the service system.

[0137] This specification provides an embodiment of a computer-readable storage medium as follows: Corresponding to the credit-based risk control processing method described above, and based on the same technical concept, one or more embodiments of this specification also provide a computer-readable storage medium.

[0138] The computer-readable storage medium provided in this embodiment is used to store computer-executable instructions, which, when executed, implement the following process: Based on the risk control request sent by the service system carrying the user's credit authorization certificate, the credit reporting platform is invoked to collect credit data; Obtain de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space; Based on the service authorization of the service system, the service risk control rules corresponding to the risk control request are obtained, and the service risk control rules and risk control model are transmitted into the privacy computing space so as to input the de-identified credit data and the service risk control rules into the risk control model for service risk detection; The risk detection results output by the privacy computing space are synchronized to the service system.

[0139] It should be noted that the embodiments of a computer-readable storage medium described in this specification and the embodiments of a credit-based risk control processing method described in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can be referred to the implementation of the corresponding method described above, and the repeated parts will not be described again.

[0140] This specification provides an example of a computer program product as follows: Corresponding to the credit-based risk control processing method described above, and based on the same technical concept, one or more embodiments of this specification also provide a computer program product.

[0141] A computer program product includes a computer program / instructions that, when executed by a processor, perform the following steps: Based on the risk control request sent by the service system carrying the user's credit authorization certificate, the credit reporting platform is invoked to collect credit data; Obtain de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space; Based on the service authorization of the service system, the service risk control rules corresponding to the risk control request are obtained, and the service risk control rules and risk control model are transmitted into the privacy computing space so as to input the de-identified credit data and the service risk control rules into the risk control model for service risk detection; The risk detection results output by the privacy computing space are synchronized to the service system.

[0142] It should be noted that the embodiments of a computer program product described in this specification and the embodiments of a credit-based risk control processing method described in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can be referred to the implementation of the corresponding method described above, and the repeated parts will not be described again.

[0143] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments. For example, the device embodiment, equipment embodiment and computer-readable storage medium embodiment are all similar to the method embodiment, so the description is relatively simple. When reading the relevant content of the device embodiment, equipment embodiment and computer-readable storage medium embodiment, please refer to the description of the method embodiment.

[0144] While one or more embodiments of this specification provide method steps as described in the embodiments or flowcharts, it is understood that the order of steps listed in the embodiments or flowcharts is merely one possible execution order among many steps, and does not represent the only execution order. Therefore, when the claims involve method steps, any changes or adjustments to the order of such steps, or the parallelism between steps, are also within the scope of protection of the claims. This specification uses specific terms to describe embodiments of this specification. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different locations in this specification do not necessarily refer to the same embodiment. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0145] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0146] In the 1930s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many improvements to the methodology today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that an improvement to the methodology cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must also be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, 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, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0147] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0148] The systems, devices, modules, or units described in 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, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0149] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0150] Those skilled in the art will understand that one or more embodiments of this specification can be provided as a method, system, or computer program product. Therefore, one or more embodiments of this specification may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this specification may take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0151] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0152] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0153] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0154] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0155] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0156] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer-readable 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, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0157] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising at least one…" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0158] One or more embodiments of this specification can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a particular task or implement a particular abstract data type. One or more embodiments of this specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0159] The above description is merely an embodiment of this document and is not intended to limit the scope of this document. Various modifications and variations can be made to this document by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this document should be included within the scope of the claims of this document.

Claims

1. A credit-based risk control processing method, applied to a risk control engine, the method comprising: Based on the risk control request sent by the service system carrying the user's credit authorization certificate, the credit reporting platform is invoked to collect credit data; Obtain de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space; Based on the service authorization of the service system, the service risk control rules corresponding to the risk control request are obtained, and the service risk control rules and risk control model are transmitted into the privacy computing space so as to input the de-identified credit data and the service risk control rules into the risk control model for service risk detection; The risk detection results output by the privacy computing space are synchronized to the service system.

2. The credit-based risk control processing method according to claim 1, after the step of calling the credit reporting platform to collect credit data based on the risk control request carrying the user's credit authorization certificate sent by the service system, and before the step of transmitting the service risk control rules and risk control model into the privacy computing space for operation execution, further includes: Based on the access authorization identifier of the data sandbox synchronized by the service system, the service processing data corresponding to the service type tag carried by the risk control request and the user's user service data are read from the data sandbox. The service processing data and the user service data are transmitted to the privacy computing space in encrypted form.

3. The credit-based risk control method according to claim 2, wherein the service risk detection is implemented in the following manner: The user credit characteristics of the de-identified credit data are matched with the credit risk control rules contained in the service risk control rules, and the service processing data and the user service data are subjected to anomaly detection. Risk detection results are generated based on credit matching results, anomaly detection results, and risk weight parameters included in the service risk control rules.

4. The credit-based risk control processing method according to claim 1, wherein obtaining the service risk control rules corresponding to the risk control request based on the service authorization of the service system includes: Based on the service type tag carried in the risk control request, a rule query request carrying the service type tag is submitted to the blockchain, and the service risk control rule returned by the query is obtained. The service risk control rules are configured by the service system for the service type tags and stored on the blockchain.

5. The credit-based risk control processing method according to claim 4, wherein obtaining the service risk control rule corresponding to the risk control request based on the service authorization of the service system further includes: If the query returns empty, the risk identification model of the service system is subject to permission verification. After the verification is successful, the risk identification model and the rule template read from the data sandbox are transmitted to the privacy computing space so that the service processing data and the rule template are input into the risk identification model to identify risk items and obtain risk items. Risk control rules are constructed based on the risk items and the rule template, and the service risk control rules are obtained by configuring the parameters of the risk control rules.

6. The credit-based risk control processing method according to claim 5, wherein the step of configuring parameters of the risk control rules to obtain the service risk control rules includes: Based on the job data of the recruitment position corresponding to the risk control request, perform risk prediction for the recruitment position, perform anomaly detection of the adaptation between job data and user resume data, and / or perform authenticity detection on key data in user resume data. Risk parameters are determined based on risk prediction results, adaptation anomaly detection results, and / or authenticity detection results, and the risk control rules are configured according to the risk parameters to obtain the service risk control rules.

7. The credit-based risk control processing method according to claim 1, after the step of calling the credit reporting platform to collect credit data based on the risk control request carrying the user's credit authorization certificate sent by the service system, and before the step of obtaining the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, further includes: Based on the service authorization identifier carried in the risk control request, a service authorization verification request is sent to the trusted management node of the trusted data space to perform service authorization verification; If the verification is successful, the operation of obtaining the de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space will be performed; The service authorization verification includes: querying the contract records between the user and the institution from the blockchain and parsing the contract to obtain the scope of contract authorization; detecting whether the service authorization identifier matches the scope of contract authorization; and if so, confirming that the verification is successful.

8. The credit-based risk control processing method according to claim 1, wherein the risk control request is sent after the service system performs service risk control detection; The service risk control detection includes: If a service process node is detected to be at a preset process node, it is checked whether the service processing performed on the user triggers a service risk control detection. If so, the user's credit authorization processing is performed to obtain the credit authorization certificate.

9. The credit-based risk control processing method according to claim 8, wherein detecting whether the service processing performed on the user triggers service risk control detection includes: The user's service data, service processing data, and service evaluation task text are input into the risk control trigger detection model to perform service risk assessment. If the service assessment result indicates that risk detection is required, then service risk control detection is triggered.

10. The credit-based risk control method according to claim 9, wherein the service risk assessment includes: Based on the job data of the job posting, risk prediction is performed for the job posting, anomaly detection is performed between the job data and the user resume data, and / or authenticity detection is performed on key data in the user resume data, and the job evaluation result is determined based on the risk prediction results, anomaly detection results and / or authenticity detection results.

11. The credit-based risk control method according to claim 1, wherein the service risk detection includes: The service risk control rules are parsed to obtain risk rules and risk weight parameters, and the risk weight parameters are mapped to the corresponding risk rules; Extract sub-data features from the sub-data under the data dimension corresponding to the risk rule in the de-identified credit data, and fuse the sub-data features to obtain fused features; The risk rules and the fusion features are matched using structured rules, and semantic matching is performed on the risk rules and the fusion features. A user risk rating is generated based on the rule matching results, the semantic matching results, and the risk weight parameters.

12. The credit-based risk control processing method according to claim 11, wherein the service risk detection further includes: Risk decision rules are obtained by extracting decision rules from the risk detection network in the risk control model. Attribution analysis is performed on the user risk rating and a visual attribution chain is generated; Structured detection results are generated based on the user risk rating, the risk decision rules, and the visualized attribution link.

13. The credit-based risk control processing method according to claim 11, wherein the service risk detection further includes: The job analysis field and the topic role field are used to extract elements to obtain job element tags, and job-related data is retrieved from multiple data retrieval dimensions based on the job element tags. A job data association network is constructed based on the retrieved job-related data, and job risk trend prediction is performed based on the job data association network to obtain job risk trend data.

14. A credit-based risk control processing device, operating within a risk control engine, the device comprising: The data collection module is configured to call the credit reporting platform to collect credit data based on the risk control request sent by the service system, which carries the user's credit authorization certificate. The data sending module is configured to obtain de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space; The risk detection module is configured to obtain the service risk control rules corresponding to the risk control request based on the service authorization of the service system, and to input the service risk control rules and the risk control model into the privacy computing space, so as to input the de-identified credit data and the service risk control rules into the risk control model for service risk detection; The result synchronization module is configured to synchronize the risk detection results output by the privacy computing space to the service system.

15. A credit-based risk control processing device, comprising: processor; And, a memory configured to store computer-executable instructions, which, when executed, cause the processor to: Based on the risk control request sent by the service system carrying the user's credit authorization certificate, the credit reporting platform is invoked to collect credit data; Obtain de-identified credit data uploaded by the credit reporting platform from the data isolation space of the trusted data space, and send the de-identified credit data to the privacy computing space; Based on the service authorization of the service system, the service risk control rules corresponding to the risk control request are obtained, and the service risk control rules and risk control model are transmitted into the privacy computing space so as to input the de-identified credit data and the service risk control rules into the risk control model for service risk detection; The risk detection results output by the privacy computing space are synchronized to the service system.

16. A computer-readable storage medium for storing computer-executable instructions that, when executed, implement the steps of the method of claim 1.

Images