Method and device for analyzing regulatory identification of associated party

Abstract

The application discloses a kind of associated party's regulatory identification analysis method and device, it is related to financial regulatory technical field, the method includes: obtaining the multiple associated party identity parameters corresponding to target associated party;Wherein, each associated party identity parameter corresponds one or more regulatory identification parameters;Regulatory identification parameter includes the time effectiveness parameter and validity threshold parameter corresponding to different existing regulatory identification;Multiple associated party identity parameters corresponding to the target associated party are input into identity parameter transmission regulatory algorithm model, and the target regulatory identification corresponding to the target associated party in different time periods is obtained;The target regulatory identification is used to describe the existing regulatory identification in effect in each time period;The identity parameter transmission regulatory algorithm model is based on gradient descent algorithm, and is trained to the pre-set machine learning model using associated party regulatory identification training sample set to obtain.The application is used to improve the accuracy, timeliness and analysis efficiency of associated party's regulatory identification analysis.

Description

Technical Field

[0001] The present invention relates to the field of financial supervision technology, and in particular to a method and device for analyzing supervision identification of related parties. Background Art

[0002] This section is intended to provide a background or context for the presented embodiments of the invention. No admission is made that the description herein is prior art by virtue of its inclusion in this section.

[0003] Accurate and timely management of related-party regulatory identification is crucial in bank operations. Due to the diverse regulatory frameworks under which banks are subject, the varying definitions of related parties within each framework, the constant adjustments to regulatory rules, and the constant changes in related-party identities due to changes in information such as departures and shareholdings, banks face significant challenges in the refined management of related parties.

[0004] Currently, existing solutions for analyzing related-party regulatory identification generally implement related functions by establishing a customer identity relationship set based on identity parameters, gradually establishing connections between all customer groups to form a relationship graph, and then calculating and transmitting the regulatory identification group value for each node in the graph based on the relationship graph. However, this existing technology has certain limitations.

[0005] The existing automatic configuration function of related party regulatory identification does not take into account the timeliness of regulatory rules. Adjustments to changes in regulatory rules rely on developers to promptly adjust the system's fixed model. If the model is not adjusted in time on the day the rules take effect, the regulatory identification of the related parties under the regulatory caliber cannot be updated in time, affecting the integrity and accuracy of the data reported by the related parties under the relevant regulatory caliber. Subsequent maintenance requires a lot of manpower and is prone to errors.

[0006] There are differences in the scope of related parties defined by different regulatory rules, and the judgment criteria for some specific identities are different. The existing related party calculation model does not distinguish the differences under different regulatory rules, resulting in inaccurate calculated related party regulatory identification.

[0007] The existing automatic configuration function of related party supervision identification relies on the timely maintenance of related party identities by users in the system to ensure timely and accurate updates of related party supervision identification due to changes in related party identities. If maintenance is delayed, the supervision identification will be incorrect during that period and the update of the supervision identification will be delayed. Summary of the Invention

[0008] An embodiment of the present invention provides a method for analyzing regulatory identification of related parties, which is used to improve the accuracy, timeliness, and efficiency of regulatory identification analysis of related parties. The method includes:

[0009] obtain a plurality of association identity parameters corresponding to the target association party; each association identity parameter corresponds to one or more supervision identifier parameters; the supervision identifier parameters include time-effect parameters and validity threshold parameters corresponding to different existing supervision identifiers; the time-effect parameters are used to describe effective time and invalid time of the existing supervision identifier; the validity threshold parameters are used to describe supervision conditions and condition thresholds for the existing supervision identifier to take effect;

[0010] input the plurality of association identity parameters corresponding to the target association party into an identity parameter conduction supervision algorithm model to obtain a target supervision identifier corresponding to the target association party at different time periods; the target supervision identifier is used to describe the existing supervision identifier taking effect at each time period; the identity parameter conduction supervision algorithm model is obtained by training a preset machine learning model based on a gradient descent algorithm using an association supervision identifier training sample set; the association supervision identifier training sample set includes historical association identity parameters of different historical association parties and historical target supervision identifiers corresponding to the different historical association parties.

[0011] The embodiment of the present application also provides an association supervision identifier analysis device to improve the accuracy, timeliness and analysis efficiency of the association supervision identifier analysis, and the device comprises:

[0012] an association identity parameter obtaining module configured to obtain a plurality of association identity parameters corresponding to the target association party; each association identity parameter corresponds to one or more supervision identifier parameters; the supervision identifier parameters include time-effect parameters and validity threshold parameters corresponding to different existing supervision identifiers; the time-effect parameters are used to describe effective time and invalid time of the existing supervision identifier; the validity threshold parameters are used to describe supervision conditions and condition thresholds for the existing supervision identifier to take effect;

[0013] a target supervision identifier output module configured to input the plurality of association identity parameters corresponding to the target association party into an identity parameter conduction supervision algorithm model to obtain a target supervision identifier corresponding to the target association party at different time periods; the target supervision identifier is used to describe the existing supervision identifier taking effect at each time period; the identity parameter conduction supervision algorithm model is obtained by training a preset machine learning model based on a gradient descent algorithm using an association supervision identifier training sample set; the association supervision identifier training sample set includes historical association identity parameters of different historical association parties and historical target supervision identifiers corresponding to the different historical association parties.

[0014] The embodiment of the present application also provides a computer device comprising a memory, a processor and a computer program stored in the memory and capable of running on the processor; when the processor executes the computer program, the above-mentioned association supervision identifier analysis method is realized.

[0015] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, which implements the above-mentioned method for analyzing the supervisory identification of related parties when executed by a processor.

[0016] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the above-mentioned method for analyzing the regulatory identification of related parties.

[0017] In an embodiment of the present invention, multiple related party identity parameters corresponding to a target related party are obtained; wherein each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include time validity parameters and validity threshold parameters corresponding to different existing regulatory identifications; the time validity parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification;

[0018] The multiple related party identity parameters corresponding to the target related party are input into the identity parameter transmission supervision algorithm model to obtain the target supervision identifier corresponding to the target related party in different time periods; the target supervision identifier is used to describe the existing supervision identifier effective in each time period; the identity parameter transmission supervision algorithm model is based on the gradient descent algorithm and is obtained by training a pre-set machine learning model with a related party supervision identifier training sample set; the related party supervision identifier training sample set includes: historical related party identity parameters of different historical related parties, and historical target supervision identifiers corresponding to different historical related parties. The embodiment of the present invention clarifies the effective time and expiration time of the existing supervision identifier by setting a timeliness parameter. When the supervision rules change, it can be timely determined whether the supervision identifier needs to be updated based on the timeliness parameter instead of relying on manual intervention. The identity parameter transmission supervision algorithm model can flexibly adapt to changes in supervision rules, ensuring that the supervision identifier of the target related party can be quickly updated when the supervision rules change, thereby solving the problem of untimely supervision identifier updates; by setting the validity threshold parameter, the supervision conditions and condition thresholds for the effectiveness of the existing supervision identifier are limited, so that when calculating the target supervision identifier, it can accurately determine the differentiated regulations of different regulators. The machine learning model can learn the accurate calculation logic under different regulatory rules, thereby improving the accuracy of regulatory identification calculation and overcoming the defects of inaccurate calculation in existing technologies; in addition, by constructing an identity parameter transmission regulatory algorithm model, it can automatically calculate the target regulatory identification corresponding to different time periods based on the input related party identity parameters, reducing the dependence on users to maintain the related party identity in the system in a timely manner. Even if the related party identity changes, the regulatory identification can be quickly recalculated based on the new related party identity parameters and algorithm model, avoiding regulatory identification errors or untimely updates due to user maintenance delays, and effectively solving the problem that related party identity maintenance affects the accuracy of regulatory identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0020] Figure 1 Schematic diagram of a flow chart of a method for analyzing regulatory identification of related parties in an embodiment of the present invention;

[0021] Figure 2 This is an example diagram of the relationship between an associated party identity parameter and a supervisory identifier in an embodiment of the present invention;

[0022] Figure 3This is a specific example diagram of the transmission logic of an identity parameter transmission supervision algorithm model in an embodiment of the present invention;

[0023] Figure 4 This is a specific example diagram of obtaining multiple associated party identity parameters corresponding to a target associated party in an embodiment of the present invention;

[0024] Figure 5 This is a specific schematic diagram of an identity parameter conduction supervision algorithm model obtained through training in an embodiment of the present invention;

[0025] Figure 6 This is a schematic structural diagram of a device for analyzing supervision identification of an associated party according to an embodiment of the present invention;

[0026] Figure 7 Schematic diagram of a computer device used for analyzing regulatory identification of related parties in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0028] The term "and / or" herein simply describes an association relationship, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, the simultaneous existence of A and B, and the existence of B alone. In addition, the term "at least one" herein refers to any combination of at least two of any one or more of a plurality of items. For example, "at least one of A, B, and C" can represent any one or more elements selected from the set consisting of A, B, and C.

[0029] In the description of this specification, the terms "include", "including", "have", "contain", etc. are all open terms, which mean including but not limited to. The descriptions with reference to the terms "one embodiment", "a specific embodiment", "some embodiments", "for example", etc. mean that the specific features, structures or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. The order of steps involved in each embodiment is used to schematically illustrate the implementation of the present application, and the order of steps therein is not limited and can be appropriately adjusted as needed.

[0030] The acquisition, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. The information collected in this application is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data are in compliance with the relevant laws, regulations and standards of relevant countries and regions, and necessary confidentiality measures are taken, which do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse. In addition, this application provides users with corresponding operation entrances for users to choose to agree or refuse the results of automated decision-making. If the user chooses to refuse, they can enter the expert decision-making process.

[0031] It should be noted that in the embodiments of the present application, some software, components, models and other existing solutions in the industry may be mentioned. For example, some existing software tools, components, algorithm models or other well-known solutions in other technical fields may be cited. They should be regarded as exemplary, and their purpose is only to illustrate the feasibility of the implementation of the technical solution of the present application. These references should be understood as typical examples, and their core purpose is to explain and verify the rationality and feasibility of the implementation of the technical solution proposed in the present application. However, it does not mean that the applicant has or will necessarily use the solution. Such references do not imply that the applicant has actually adopted these existing solutions, or will inevitably adopt these methods in the future during its technical implementation. In other words, these references only serve to illustrate the nature of the invention, to help understand the relationship between the innovation of the present application and the prior art and its transcendence, and do not constitute an endorsement or reliance statement on a specific prior art product.

[0032] The embodiments of the present invention involve the following terms, which are explained as follows:

[0033] Related party: a natural person, legal person or unincorporated organization that has an associated relationship with a bank as defined in the regulatory rules that the bank needs to comply with;

[0034] Related-party transactions refer to the transfer of resources, services, obligations or other interests between a bank and its related parties;

[0035] Related party identity: The relationship between superior and subordinate related parties as defined by regulators.

[0036] Since banks are subject to multiple regulatory frameworks and the definitions of related parties under each framework vary, and the rules are constantly being adjusted, in addition, due to changes in information such as the resignation or shareholding of related parties, their identities are also constantly changing, requiring a high level of refined management and timely user maintenance. Therefore, there is a problem of delayed related party maintenance or changes in regulatory rules leading to untimely and inaccurate updates of regulatory identification.

[0037] (1) The existing automatic transmission function of related party regulatory identification does not take into account the timeliness of regulatory rules. If the rules of a certain regulatory party change, if the system does not promptly amend the existing regulatory identification calculation model when the new regulatory rules take effect, the regulatory identification of the related parties in the system will not be updated in a timely manner. The automatic configuration function of related party regulatory identification does not take into account the impact of three factors on regulatory identification calculation: the timeliness of regulatory rules, the differences in the definition of related parties with the same identity under different regulatory standards, and the timeliness of related party identity maintenance. This will lead to problems such as inaccurate calculation of regulatory identification and untimely updates;

[0038] (2) The existing automatic configuration function of related party regulatory identification does not make special treatment for the differentiated provisions of different regulators. When different regulatory standards have different definitions of related parties with the same identity, the regulatory identification of the related parties is not calculated accurately. The scope of related parties defined by different regulatory rules varies, and the judgment standards for some specific identities are different. For example, regulator "A" stipulates that only natural persons with a shareholding ratio of more than 5% are considered related parties, but regulator "B" stipulates that only natural persons with a shareholding ratio of more than 10% are considered related parties. Under the existing related party regulatory calculation model, the differences under different regulatory rules are not distinguished, resulting in inaccurate calculated related party regulatory identification.

[0039] (3) The existing automatic configuration function of the regulatory identification of related parties depends on the timely maintenance of the related party identity by the user in the system to ensure that the regulatory identification of related parties is updated in a timely and accurate manner due to changes in the related party identity. If the maintenance is delayed, the regulatory identification will be incorrect during that period and the update of the regulatory identification will be delayed. The relationship between a related party and its superior related party is constantly changing, and different relationships may transmit different regulatory identifications. Under the existing technology, it is impossible to set the validity period of the relationship. There are only two ways to deal with the relationship that once existed and has been entered: retain or delete. If retained, the regulatory identification transmitted by the relationship will always exist; if deleted, the regulatory identification that should have been possessed in the corresponding historical stage will be lost. Both treatments will lead to errors in the calculation of the regulatory identification, which is inconsistent with the actual situation.

[0040] In order to solve the above problems, first, a time dimension and specific regulatory judgment rules are added to the identity parameters, and when entering related parties, users are required to enter the valid time interval of the related party identity to add judgment factors. At the same time, a set of identity parameter regulatory identification transmission model calculation is established to realize the measurement of regulatory identity parameter transmission, thereby improving the accuracy and timeliness of related party regulatory identification calculation, reducing the dependence on the timeliness of user operations, further improving the refined management of related parties, and improving the quality of regulatory data submitted by banks. Specifically, the embodiment of the present invention provides a method for analyzing the regulatory identification of related parties to improve the accuracy, timeliness and analysis efficiency of the regulatory identification analysis of related parties, see Figure 1, the method may include:

[0041] Step 101: Acquire multiple related party identity parameters corresponding to the target related party; wherein each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include timeliness parameters and validity threshold parameters corresponding to different existing regulatory identifications; the timeliness parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification;

[0042] Step 102: Input multiple related party identity parameters corresponding to the target related party into the identity parameter transmission supervision algorithm model to obtain the target supervision identifier corresponding to the target related party in different time periods; the target supervision identifier is used to describe the existing supervision identifier effective in each time period; the identity parameter transmission supervision algorithm model is based on the gradient descent algorithm, and is obtained by training a preset machine learning model with a related party supervision identifier training sample set; the related party supervision identifier training sample set includes: historical related party identity parameters of different historical related parties, and historical target supervision identifiers corresponding to different historical related parties.

[0043] In an embodiment of the present invention, multiple related party identity parameters corresponding to a target related party are obtained; wherein each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include time validity parameters and validity threshold parameters corresponding to different existing regulatory identifications; the time validity parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification;

[0044] The multiple related party identity parameters corresponding to the target related party are input into the identity parameter transmission supervision algorithm model to obtain the target supervision identifier corresponding to the target related party in different time periods; the target supervision identifier is used to describe the existing supervision identifier effective in each time period; the identity parameter transmission supervision algorithm model is based on the gradient descent algorithm and is obtained by training a pre-set machine learning model with a related party supervision identifier training sample set; the related party supervision identifier training sample set includes: historical related party identity parameters of different historical related parties, and historical target supervision identifiers corresponding to different historical related parties. The embodiment of the present invention clarifies the effective time and expiration time of the existing supervision identifier by setting a timeliness parameter. When the supervision rules change, it can be timely determined whether the supervision identifier needs to be updated based on the timeliness parameter instead of relying on manual intervention. The identity parameter transmission supervision algorithm model can flexibly adapt to changes in supervision rules, ensuring that the supervision identifier of the target related party can be quickly updated when the supervision rules change, thereby solving the problem of untimely supervision identifier updates; by setting the validity threshold parameter, the supervision conditions and condition thresholds for the effectiveness of the existing supervision identifier are limited, so that when calculating the target supervision identifier, it can accurately determine the differentiated regulations of different regulators. The machine learning model can learn the accurate calculation logic under different regulatory rules, thereby improving the accuracy of regulatory identification calculation and overcoming the defects of inaccurate calculation in existing technologies; in addition, by constructing an identity parameter transmission regulatory algorithm model, it can automatically calculate the target regulatory identification corresponding to different time periods based on the input related party identity parameters, reducing the dependence on users to maintain the related party identity in the system in a timely manner. Even if the related party identity changes, the regulatory identification can be quickly recalculated based on the new related party identity parameters and algorithm model, avoiding regulatory identification errors or untimely updates due to user maintenance delays, and effectively solving the problem that related party identity maintenance affects the accuracy of regulatory identification.

[0045] During specific implementation, step 101 is first performed: multiple related party identity parameters corresponding to the target related party are obtained; wherein, each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include timeliness parameters and validity threshold parameters corresponding to different existing regulatory identifications; the timeliness parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification.

[0046] In the embodiment, first, it is necessary to comprehensively and accurately extract the definition information formulated by each regulator for the related parties from the preset regulatory rule library.

[0047] The extracted definitions of related parties from various regulatory bodies are meticulously broken down to derive the identity parameters for each related party. These related party identity parameters include, but are not limited to, shareholder identity, positional relationship identity, and kinship relationship identity. For example, shareholder identity parameters can reflect the related party's shareholding in the enterprise, including information such as shareholding percentage and duration of shareholding; positional relationship identity parameters can reflect the related party's position hierarchy and scope of responsibilities within the enterprise or with related enterprises; and kinship relationship identity parameters clarify the ties between the related party and other related parties formed through kinship.

[0048] For each related party identity parameter obtained through splitting, it is necessary to further determine the timeliness parameters in the corresponding regulatory identification parameters based on the effective time of the transaction management-related systems involving the related parties of each regulatory party. The timeliness parameters are specifically used to accurately describe the effective time and expiration time of the existing regulatory identification. Taking a certain regulatory identification as an example, its timeliness parameters may clearly stipulate that it will take effect from X month X day of XX year and will expire on AA month AA day of AA year. Through such a clear timeliness definition, it can be ensured that at different time points, the regulatory identification applicable to the related parties can maintain a high degree of consistency with the timeliness of the regulatory rules, avoiding the incorrect or outdated application of the regulatory identification due to time factors.

[0049] When mapping related party identity parameters to regulatory identifiers, if different regulators define a related party identity parameter with the same name and meaning, it is necessary to determine timeliness parameters and validity threshold parameters for multiple regulatory identifiers for that related party identity parameter. For example, the related party identity parameter "major shareholder" may correspond to different regulatory identifiers under different regulators' regulations, and each regulatory identifier has its own timeliness parameters and validity threshold parameters. The validity threshold parameters are used to describe in detail the regulatory conditions and condition thresholds for the effectiveness of existing regulatory identifiers. For shareholder identity parameters, the validity threshold parameters may include a shareholding ratio threshold, such as a regulatory identifier specifying that the regulatory identifier takes effect when the shareholding ratio reaches 5% or above. For position relationship identity parameters, the validity threshold parameters may reflect a position level threshold, such as specifying that a specific regulatory identifier applies to senior management personnel and above. The validity threshold parameters for kinship identity parameters may be the type of kinship relationship, such as specifying that a specific regulatory identifier applies to kinship relationships.

[0050] In one embodiment, the related party identity parameters include but are not limited to shareholder identity, job relationship identity and kinship identity; the validity threshold parameters of each related party identity parameter include but are not limited to shareholding ratio threshold, job level threshold, and kinship type.

[0051] In a specific embodiment of the related party regulatory identification analysis method of the present invention, the related party identity parameters cover multiple types, including but not limited to shareholder identity, position relationship identity and kinship identity. These different types of related party identity parameters each have unique characteristics, and their corresponding validity threshold parameters also have clear definitions and diverse forms of expression.

[0052] The shareholder identity parameter primarily characterizes the shareholding of related parties in a company and plays a key role in determining the regulatory designation of related parties. Its validity threshold parameters include shareholding ratio thresholds, with different shareholding ratio thresholds corresponding to different regulatory designations. For example, under certain regulatory requirements, when a shareholding ratio reaches or exceeds 5%, a related party may be deemed to have a certain level of influence, triggering a series of corresponding regulatory designations. This means that the shareholder may have a certain degree of influence in corporate decision-making and related-party transactions, and thus be subject to stricter oversight. In other regulatory contexts, a shareholding ratio of 10% or higher may be required for specific regulatory designations to apply. This reflects the differentiated criteria used by different regulators to assess shareholder influence based on factors such as company size and industry characteristics. By using clear shareholding ratio thresholds as validity threshold parameters, it is possible to precisely define the regulatory designations corresponding to shareholder identity parameters in different regulatory environments, thereby achieving refined oversight of shareholder-related parties.

[0053] The position-based identity parameter focuses on reflecting the positional hierarchy and scope of responsibilities of related parties within an enterprise or with related enterprises. Its effectiveness threshold parameter is based on a position-level threshold. For example, in financial institutions, senior management (such as presidents, vice presidents, and department general managers) hold high positions, and their decisions and actions may have a significant impact on the enterprise's operations and related-party transactions. When a related party's position level reaches senior management, regulatory requirements trigger specific supervisory designations, requiring more rigorous review and disclosure of related-party transactions. Conversely, for mid-level managers or grassroots employees, whose positions are lower, the applicable supervisory designations and requirements are relatively relaxed. This position-level-based effectiveness threshold parameter allocation allows for the rational allocation of supervisory resources based on the related party's positional influence within the enterprise, ensuring targeted and effective supervision and preventing potential profiteering or risks arising from position-based relationships.

[0054] The kinship identity parameter focuses on the ties of connection formed between related parties and other related parties based on kinship. Its effectiveness threshold parameter is mainly the kinship type. In business operations, certain kinship relationships may give rise to potential conflicts of interest or related-party transaction risks. For example, kinship relationships (such as parents, children, spouses, etc.) are generally considered to be relatively close kinship types and may receive special attention in supervision. When related parties are related, transactions involving related parties may be subject to more stringent approval procedures and information disclosure requirements, and will be assigned specific regulatory labels accordingly. For other kinship relationships (such as siblings, uncles and nephews), regulatory requirements and corresponding regulatory labels will also vary depending on the closeness of the relationship and the potential impact. By clarifying the kinship type as the effectiveness threshold parameter, it is possible to effectively identify and manage the related-party transaction risks caused by kinship relationships, and maintain market fairness and transparency.

[0055] In practical applications, by accurately identifying and applying these diverse related-party identity parameters and their validity threshold parameters, combined with the subsequent identity parameter transmission supervision algorithm model, it is possible to achieve precise analysis and dynamic management of related-party supervision identification, adapt to the complex and changing regulatory environment and business conditions, and improve the scientific nature and effectiveness of related-party supervision.

[0056] Figure 4 FIG is a specific example diagram of obtaining multiple associated party identity parameters corresponding to a target associated party in an embodiment of the present invention. In one embodiment, Figure 4 As shown, multiple related party identity parameters corresponding to the target related party are obtained, including:

[0057] Step 401: Select corresponding affiliate identity parameters for the target affiliate based on the actual relationship between the target affiliate and the affiliate management entity or the superior affiliate;

[0058] Step 402: extracting association-related information of different supervisors with the target related party from a preset supervisory rule library;

[0059] Step 403: Split the association-related information of each supervisory party to the target related party into related party identity parameters corresponding to the supervisory structure;

[0060] Step 404: for each existing supervisory identifier in the identity parameter of the associated party, determine the timeliness parameters and validity threshold parameters of different existing supervisory identifiers according to the associated related information of the supervisor.

[0061] In a specific embodiment of the method for analyzing related party supervision identification of the present invention, the process of obtaining multiple related party identity parameters corresponding to the target related party involves multiple key steps, which will be described in detail below.

[0062] First, it is necessary to thoroughly explore the actual relationship between the target related party and its management entity or superior related party. This actual relationship covers a variety of possible forms, such as equity relationship, business cooperation relationship, management and control relationship, etc.

[0063] Based on a precise understanding of these relationships, appropriate related party identity parameters are selected for the target related party. If the target related party is a shareholder of a company, holding a significant stake and exerting influence on corporate decision-making, then based on this specific situation, related party identity parameters related to shareholder status, such as shareholder identity parameters, should be selected. If the target related party is an internal manager who plays a key role in business processes or resource allocation, position-based identity parameters should be selected to accurately reflect their position and role within the corporate structure. If the target related party has a family relationship with the company, and this relationship could influence the company's operational decisions or involve related party transactions, family-based identity parameters are appropriate. By carefully selecting related party identity parameters based on the actual relationship, the accuracy and relevance of subsequent analysis can be ensured.

[0064] Next, comprehensive information on the relationships between different regulators and the target related parties is extracted from a pre-defined regulatory rules library. This library integrates various rules, regulations, and guidance issued by numerous authoritative regulators, including those from various financial regulators. This information source provides a solid foundation for accurately defining the identity parameters of the target related parties. The information extracted from the regulatory rules library includes, but is not limited to, the regulator's definitions of related parties, classification standards, regulatory requirements, and conditions associated with specific identities.

[0065] The related-party information obtained from each regulatory body on the target related party is meticulously broken down to derive related party identity parameters corresponding to the regulatory structure. For example, the Financial Regulatory Administration's regulations on related parties of banking and insurance institutions may include a variety of definitions of related parties, such as shareholders holding a certain percentage of shares, legal or natural persons with controlling relationships, key management personnel, and their immediate family members. During the breakdown process, these different scenarios are converted into specific related party identity parameters, such as shareholder identity parameters (corresponding to shareholders holding a certain percentage of shares), position relationship identity parameters (corresponding to key management personnel), and kinship relationship identity parameters (corresponding to immediate family relationships). This precise breakdown transforms abstract regulatory provisions into actionable related party identity parameters, providing a clear foundation for subsequent analysis and calculations.

[0066] For each existing regulatory identifier within the related party's identity parameters, the timeliness parameters and validity threshold parameters for each existing regulatory identifier are determined based on the relevant information about the related party. For timeliness parameters, strict reference must be made to the regulatory authority's specified effective and expiration dates. For example, if a regulatory identifier is specified to be effective from a specific date and expire at another date, this information will be accurately recorded as the regulatory identifier's timeliness parameter. For validity threshold parameters, these parameters are determined based on the conditions set by the regulatory authority for each identity parameter. For example, for shareholder identity parameters, the regulator may specify an 8% shareholding threshold. When the shareholding threshold exceeds this threshold, the specific regulatory identifier becomes effective. In this case, 8% serves as the regulatory identifier's validity threshold parameter (shareholding threshold). For position-based identity parameters, if the regulator specifies that a regulatory identifier applies to department managers and above, the "department manager" level becomes the corresponding validity threshold parameter (position-based threshold). For kinship identity parameters, if kinship triggers a specific regulatory identifier, the "kinship" level serves as the validity threshold parameter (kinship type). Through such a rigorous determination process, we ensure that the regulatory identification corresponding to each related party's identity parameters can be accurately defined in terms of time validity and conditional applicability, thereby providing strong support for accurate regulatory identification analysis.

[0067] In the entire process of obtaining multiple related party identity parameters corresponding to the target related parties, each step is closely linked and interdependent, together forming a complete and rigorous system, which lays a solid foundation for the subsequent use of the identity parameter transmission supervision algorithm model to accurately calculate the target supervision identification corresponding to the target related parties in different time periods, and helps to achieve refined and scientific management of related party supervision, and effectively respond to the complex and changing regulatory environment and business conditions.

[0068] During specific implementation, after performing step 101: obtaining multiple related party identity parameters corresponding to the target related party; wherein each related party identity parameter corresponds to one or more regulatory identification parameters, performing step 102: inputting the multiple related party identity parameters corresponding to the target related party into the identity parameter transmission regulatory algorithm model to obtain the target regulatory identification corresponding to the target related party in different time periods; the target regulatory identification is used to describe the existing regulatory identification effective in each time period; the identity parameter transmission regulatory algorithm model is based on the gradient descent algorithm, and is obtained by training a preset machine learning model with a related party regulatory identification training sample set; the related party regulatory identification training sample set includes: historical related party identity parameters of different historical related parties, and historical target regulatory identifications corresponding to different historical related parties.

[0069] In the embodiment, the identity parameter transmission supervision algorithm model is constructed based on the gradient descent algorithm, which plays an important role in optimizing model parameters. By training the pre-set machine learning model with the related party supervision identification training sample set, the model is able to learn the complex relationship between the related party identity parameters and the supervision identification. The related party supervision identification training sample set contains rich historical information, among which the historical related party identity parameters of different historical related parties and the historical target supervision identification corresponding to different historical related parties are key components. These historical data are derived from actual supervision scenarios and related party management records during the operation of the enterprise, and are highly authentic and representative. For example, the historical related party identity parameters may cover the detailed identity information of various related parties (such as major shareholders, senior management, family-related related parties, etc.) in different types of enterprises (such as financial institutions, manufacturing enterprises, service enterprises, etc.), including specific data such as shareholding ratio, job level, and family relationship type. The historical target supervision identification clearly records the accurate supervision identification applicable to these related parties in the corresponding historical period. By integrating and analyzing a large amount of such historical data, a solid data foundation is provided for model training, enabling it to accurately predict the corresponding target regulatory identification based on the input related party identity parameters when facing new target related parties.

[0070] Figure 5 Detailed schematic diagram of a training-based identity parameter conduction supervision algorithm model according to an embodiment of the present invention. In one embodiment, Figure 5 As shown, it also includes: obtaining the identity parameter conduction supervision algorithm model by training as follows:

[0071] Step 501: Obtain a training sample set of related party regulatory identifications; randomly initialize the weights and bias parameters of a preset machine learning model;

[0072] Step 502: Preset a loss function for the machine learning model; the loss function is used to measure the difference between the target regulatory identifier predicted by the model and the actual regulatory identifier in the sample set;

[0073] Step 503: Using the related party regulatory identification training sample set, adopting the gradient descent algorithm as the optimization algorithm, and iteratively updating the randomly initialized machine learning model to gradually reduce the value of the loss function;

[0074] Step 504: Repeat the above-mentioned iterative update operation until the iteratively updated model converges or reaches a predetermined number of iterations; the model obtained by stopping the iterative update is used as the trained identity parameter conduction supervision algorithm model.

[0075] For example, the training process of the identity parameter transmission supervision algorithm model includes the following steps:

[0076] 1. Acquisition of related party supervision identification training sample set

[0077] To construct an effective algorithmic model for the transmission of regulatory identity parameters, the primary task is to obtain a training set of related-party regulatory identifiers. This data set primarily draws from long-term, accumulated historical regulatory data. This data covers enterprises across a wide range of industries, including but not limited to various entities in finance, manufacturing, and services. From these enterprise regulatory records, detailed information on various historical related parties is mined and filtered to generate historical related-party identity parameters. These parameters comprehensively capture the various characteristics of related-party relationships within the enterprise. For example, for shareholder identity, information such as shareholding percentage, nature of shares, and duration of shareholding is accurately recorded; for positional relationships, key elements such as position level, department, and scope of responsibilities are clearly documented; and for kinship relationships, detailed information on the type of kinship relationship, the relationship between the kin and the enterprise, and the potential influence on corporate decision-making is provided. Furthermore, for each historical related party, the historical target regulatory identifiers that were actually applied at the time are accurately obtained. These identifiers reflect the decisions made by regulators based on the actual circumstances of the enterprise and regulatory rules. By systematically integrating and meticulously curating this vast, multi-dimensional historical data, a rich and representative training set of related-party regulatory identifiers is constructed, providing a solid data foundation for subsequent model training.

[0078] 2. Initialization of machine learning model parameters

[0079] After obtaining the training sample set, the pre-set machine learning model is initialized. Random initialization is used for the model's weights and bias parameters. This process assigns initial values ​​to these parameters within a specific range. This operation aims to break the symmetry of the model's initial state and prevent the model from being trapped in a local optimum at the beginning of training. By introducing randomness, the model has ample room for exploration during subsequent training, enabling it to fully learn the complex relationships and underlying patterns within the data. Different initial weight and bias values ​​allow the model to learn and adjust in diverse ways when presented with the same training data, increasing the likelihood of finding a global optimal solution or a near-optimal solution, laying the foundation for subsequent, accurate model training.

[0080] 3. Presetting of loss function

[0081] Next, the loss function for the machine learning model is pre-set. This loss function plays a crucial role in model training. Its core function is to measure the degree of discrepancy between the target regulatory identifier predicted by the model and the actual regulatory identifier in the sample set. Specifically, the loss function quantifies this discrepancy through precise mathematical calculations. A common construction method is based on a combination of a mean square error term and a weight decay term. The formula for calculating the mean square error term typically involves summing and averaging the squared differences between the model's predicted value and the actual value. It intuitively reflects the degree of deviation between the predicted result and the actual situation. During training, the model continuously adjusts its parameters to minimize the mean square error term, thereby improving prediction accuracy. The weight decay term is a regularization method introduced to prevent model overfitting. It constrains the model weights, preventing excessive weights from leading to excessive model complexity and loss of generalization ability to new data. By appropriately setting the coefficient of the weight decay term, a balance can be achieved between model complexity and generalization ability, ensuring that the model not only performs well on the training samples but also accurately predicts regulatory identifiers when faced with new related-party data.

[0082] 4. Model iterative update based on gradient descent algorithm

[0083] Based on the obtained training sample set of related-party regulatory identifications, the gradient descent algorithm is used as the optimization algorithm to iteratively update the randomly initialized machine learning model. During each iteration, the model first calculates the related-party identity parameters in the training sample set based on the current weights and bias parameters to obtain a predicted target regulatory identification. These predicted values ​​and the actual regulatory identifications in the sample set are then substituted into the loss function to calculate the value of the loss function. Next, the gradient of the loss function with respect to the model weights and bias parameters is calculated to determine the direction and magnitude of parameter adjustments. The gradient descent algorithm updates the model parameters at a pre-set learning rate, following the inverse direction of the gradient. The choice of learning rate is crucial: a too high a rate may prevent the model from converging or even diverging, while a too low a rate may lengthen the training process. In this way, the model gradually adjusts its parameters with each iteration, gradually reducing the value of the loss function. By repeating this iterative update process, the model continuously optimizes its parameters to better fit the relationship between the related-party identity parameters and regulatory identifications in the training data.

[0084] 5. Model convergence judgment and final determination

[0085] In the process of continuous iterative updates, it is necessary to constantly judge the status of the model. Repeat the above iterative update operation until one of the following two conditions is met: First, the model converges after iterative update, that is, the loss function value of the model no longer changes significantly in multiple consecutive iterations, or the change amplitude is less than the pre-set threshold, indicating that the model has learned the stable pattern in the data and reached a relatively optimal state; second, the predetermined number of iterations is reached. Even if the model has not yet fully converged, in order to avoid over-training or too long training time, training is stopped when the predetermined number of iterations is reached. When one of the above conditions is met, the model obtained by stopping the iterative update is determined as the trained identity parameter conduction supervision algorithm model. After a rigorous training process, the model has the ability to accurately predict the target regulatory identification based on the input related party identity parameters, and can provide reliable technical support for subsequent related party regulatory identification analysis, and effectively respond to complex and changing related party regulatory scenarios.

[0086] In one embodiment, the identity parameter transmission supervision algorithm model includes:

[0087] An input layer, for receiving multiple related party identity parameters corresponding to the target related party;

[0088] The transmission layer is used to perform nonlinear transformation on the input related party identity parameters, and perform mapping and translation changes through weights and offsets to transmit the regulatory identification of the related party identity parameters to the target related party, and output the target regulatory identification corresponding to the target related party in different time periods through the output layer; the transmission layer includes multiple transmission layer nodes, each transmission layer node represents a state in the transmission process of the regulatory identification of the related party identity parameters;

[0089] The output layer is used to output the target regulatory identification corresponding to the target related party in different time periods.

[0090] In the above embodiment, in the method for analyzing related party regulatory identification of the present invention, the identity parameter transmission regulatory algorithm model is the core component for realizing accurate regulatory identification analysis, and its specific structure and function show a high degree of systematicity and logic in one embodiment.

[0091] The input layer in the model plays a key role in receiving multiple related party identity parameters corresponding to the target related party. These related party identity parameters are the basic data for subsequent analysis and contain a wealth of information.

[0092] For example, when the target related party is a shareholder of a certain enterprise, the related party identity parameters received by the input layer may include parameters related to the shareholder's identity, such as shareholding ratio, nature of shares (such as common stock, preferred stock, etc.), and length of shareholding. If the target related party is an internal manager of the enterprise, the input layer will receive job relationship identity parameters such as job level, department, and scope of responsibilities. For related parties with family relationships, the parameters received by the input layer include family relationship identity parameters such as the type of family relationship (such as relatives, other relatives, etc.), the way relatives are related in the enterprise (such as indirect connection through equity, direct participation in business management, etc.). The input layer has a high degree of compatibility and data processing capabilities, and can accurately identify and receive related party identity parameters of various types and formats, ensuring the integrity and accuracy of the data, and providing a basis for subsequent processing in the transmission layer.

[0093] The transmission layer is the core computational area of ​​the model, responsible for performing complex nonlinear transformations on the input related party identity parameters and mapping and translating them through weights and offsets, thereby achieving the key function of transmitting the regulatory identification of the related party identity parameters to the target related party. The transmission layer contains multiple transmission layer nodes, each of which represents a state in the regulatory identification transmission process of a related party identity parameter. During the transmission process, the related party identity parameters received from the input layer first enter the first node of the transmission layer. Each node performs a weighted calculation on the input parameters based on preset weights. The weight value reflects the relative importance of the related party identity parameters represented by the node in the overall regulatory identification transmission process.

[0094] For example, in certain regulatory scenarios, shareholding ratios may carry a higher weight in determining regulatory designations, while kinship type may carry a relatively lower weight. After weighted calculations, nodes perform nonlinear transformations on the results using activation functions (such as the sigmoid function). This nonlinear transformation capability enables the model to handle complex nonlinear relationships between related-party identity parameters and regulatory designations, enhancing the model's adaptability to real-world regulatory situations. The nonlinearly transformed results are then shifted based on an offset to further adjust the data distribution and characteristics. The processed data is then passed to the next node using the connection weights between nodes, and this process continues layer by layer, with multiple rounds of computation and transmission occurring in the transmission layer. When processing data, each node considers not only its own inputs and parameters but also the outputs of preceding nodes, which in turn influence the inputs of subsequent nodes, forming a dynamic, interconnected transmission network. In this way, the transmission layer gradually integrates and optimizes the information contained in the related-party identity parameters, transforming them into intermediate results that are closely linked to regulatory designations, ultimately transmitting these results to the output layer.

[0095] The output layer's primary function is to output the target regulatory identifiers for the target related parties at different time periods. After complex processing by the transmission layer, the regulatory identifier information related to the target related parties is delivered to the output layer in a highly refined and accurate form. Based on the final results of the transmission layer, the output layer outputs the regulatory identifiers applicable to the target related parties at different time periods according to predetermined rules and formats.

[0096] For example, the output may be presented in a timeline format, clearly listing the regulatory identifiers corresponding to the target related parties in various time periods. These target regulatory identifiers provide regulators and enterprises with a clear regulatory basis, helping to accurately determine the regulatory status of target related parties at different times, thereby implementing corresponding regulatory measures and internal enterprise management strategies to ensure the compliance and controllability of related-party transactions and promote fair and orderly market operations. The entire identity parameter transmission regulatory algorithm model achieves accurate transmission and analysis from related-party identity parameters to target regulatory identifiers through the collaborative work of the input layer, transmission layer, and output layer, providing an efficient and reliable technical solution for related-party supervision.

[0097] In one embodiment, the objective function of the identity parameter conduction supervision algorithm model is constructed based on the mean square error term and the weight decay term; the mean square error term is used to measure the difference between the predicted value and the true value; the weight decay term is used to avoid overfitting caused by excessive weight, and the gradient descent method is used to solve the objective function to optimize the model parameters.

[0098] In the embodiment, the objective function of the identity parameter transmission supervision algorithm model is constructed based on the mean square error term and the weight decay term. The mean square error term is a key indicator to measure the difference between the model prediction value and the true value. In the context of related party supervision identification analysis, the predicted value is the target supervision identification predicted by the model based on the input related party identity parameters, and the true value is the supervision identification actually applicable to the corresponding related party in a specific time period obtained from historical supervision data. The calculation formula of the mean square error term is usually to sum the squares of the difference between the predicted supervision identification and the true supervision identification in each sample data, and then divide it by the number of samples.

[0099] The constructed objective function is solved using the gradient descent method to optimize the model parameters. The basic principle of the gradient descent method is to gradually adjust the parameter values ​​based on the gradient direction of the objective function at the current parameter value, so that the objective function value decreases towards the minimum value. During each iteration, the gradient of the objective function with respect to each model parameter (including weights and bias parameters) is first calculated. For the mean square error term, the calculation of its gradient with respect to the weight parameter involves complex mathematical derivation based on the difference between the predicted value and the true value and the characteristics of the input data.

[0100] In one embodiment, it further includes:

[0101] Update related party identity parameters and corresponding regulatory identification parameters in real time;

[0102] Retrain or fine-tune the identity parameter transmission supervision algorithm model based on the updated related party identity parameters and regulatory identification parameters.

[0103] In one embodiment of the related party regulatory identification analysis method of the present invention, in order to ensure the timeliness and accuracy of the regulatory identification analysis, it involves real-time updating of the related party identity parameters and the corresponding regulatory identification parameters, and corresponding retraining or fine-tuning of the identity parameter transmission regulatory algorithm model based on the updated data.

[0104] As a company's operations continue and the regulatory environment evolves, the actual circumstances of related parties may change, requiring real-time updates to related party identity parameters and corresponding regulatory identification parameters. Updates to related party identity parameters can arise from a wide range of sources. For example, changes in a company's equity structure can lead to changes in shareholder identity parameters, including increases or decreases in shareholding ratios, the entry of new shareholders, or the exit of existing shareholders. When a company makes internal personnel adjustments, position relationship identity parameters require corresponding updates, such as manager promotions or the redivision of responsibilities. Family relationship identity parameters may also change due to changes in related party relationships, such as the establishment or dissolution of family relationships. These changes directly impact the related party's position and potential influence within the company, and thus are linked to its corresponding regulatory identification parameters.

[0105] The update of regulatory identification parameters closely follows the adjustment of regulatory rules and changes in the actual situation of related parties. Regulators may revise regulatory rules based on market development and risk management needs, such as adjusting shareholding ratio thresholds, job level definition standards, or regulatory focus on kinship types, which will prompt changes in the corresponding regulatory identification parameters of related parties. At the same time, the optimization of the company's internal risk assessment mechanism may also lead to changes in regulatory requirements for related parties, thereby triggering the update of regulatory identification parameters. For example, if a company finds that the business transaction risks of a related party have increased, it may proactively adjust its regulatory identification parameters and strengthen its supervision over it. To achieve real-time updates, companies need to establish an efficient information monitoring and collection mechanism to promptly obtain various changes in related party-related information and accurately convert them into updateable related party identity parameters and regulatory identification parameter formats to ensure the timeliness and accuracy of the data.

[0106] After the related party identity parameters and regulatory identification parameters are updated, the identity parameter transmission supervision algorithm model needs to be retrained or fine-tuned based on the new data to ensure that the model's predictive ability matches the actual situation. If the amount of updated data is large and has a significant impact on the judgment logic of the related party regulatory identification, such as major changes in regulatory rules leading to a comprehensive change in the calculation rules of regulatory identification, or large-scale equity restructuring or business transformation of the enterprise, the model needs to be retrained. The retraining process is similar to the initial training process of the model, that is, the training sample set containing the updated related party identity parameters and regulatory identification parameters is re-obtained, the weights and bias parameters of the preset machine learning model are randomly initialized again, and a new loss function is pre-set (taking into account the updated regulatory environment and data characteristics). Then, based on the new training sample set, the gradient descent algorithm is used for iterative updates until the model converges or reaches the predetermined number of iterations, thereby obtaining a brand new identity parameter transmission supervision algorithm model that adapts to the new data characteristics and regulatory requirements.

[0107] However, if the updated data volume is relatively small, or the impact on the regulatory identification logic is relatively localized, such as a minor adjustment to the shareholding ratio of a specific related party or a small change in a manager's position, fine-tuning the model can be used. The fine-tuning process primarily involves locally adjusting model parameters based on the existing model using a small amount of updated data. First, the updated related party identity parameters are input into the existing identity parameter transmission regulatory algorithm model to calculate the model's predictions. Then, a loss function is calculated based on the difference between the updated regulatory identification parameters (true values) and the predictions. Next, the gradient descent algorithm is used to calculate the gradient of the loss function with respect to the model parameters, and the model parameters are fine-tuned and updated at a low learning rate. This process is repeated until the model achieves satisfactory predictive performance on the updated data, for example, the loss function value falls within an acceptable range. This approach avoids the high computational cost and time associated with large-scale retraining while ensuring that the model can promptly adapt to small changes in related party identity and regulatory identification parameters, maintaining the accuracy and timeliness of related party regulatory identification analysis. This dynamic update and model adjustment mechanism enables the related party regulatory identification analysis method of the present invention to flexibly respond to complex and changing business operations and regulatory environments, and continuously provide reliable regulatory identification analysis services.

[0108] In one embodiment, it further includes:

[0109] Verify and review the calculated target regulatory identifier; the verification and review are used to determine that the target regulatory identifier complies with regulatory requirements and business logic;

[0110] If verification and review reveal errors or inconsistencies in the target regulatory identification, the input process of the related party identity parameters and regulatory identification parameters will be traced back to locate and correct the errors.

[0111] In one embodiment of the related party regulatory identification analysis method of the present invention, in order to ensure the accuracy and reliability of the regulatory identification analysis results, a strict verification and review process is implemented after the target regulatory identification is calculated, and an effective backtracking and correction mechanism is established for possible errors or inconsistencies.

[0112] The verification process is primarily based on pre-set regulatory requirements and business logic rules. Regulatory requirements are derived from laws, regulations, and guidelines issued by various regulatory bodies, such as the detailed regulations on related-party supervision formulated by the Financial Regulatory Administration and the China Securities Regulatory Commission. These regulations clarify the regulatory identification standards that should apply to different types of related parties in different situations. Business logic is a logical system for determining related-party regulatory identification based on factors such as the company's own operating characteristics, internal management rules, and industry practices. For example, in some industries, there may be special regulatory identification requirements for related-party transactions in specific business areas. These requirements are formulated based on industry risk characteristics and business process specifications.

[0113] During the verification process, the first step is to check whether the target regulatory identification conforms to the regulator's definition and classification standards for related party identities. For example, for related parties with shareholder status, the verification process involves verifying whether their regulatory identification is consistent with factors such as shareholding ratio and share type, and whether they meet the regulatory identification requirements set by the regulator for different shareholding levels. For related parties with position-based relationships, the verification process involves reviewing whether their regulatory identification is consistent with factors such as job level and scope of responsibilities, and whether they comply with the regulator's regulations on the supervision of internal management personnel.

[0114] At the same time, based on business logic, the rationality of the target regulatory identification within the enterprise's internal related party management system is checked. For example, within a corporate group, transactions between related parties may be subject to internal risk control rules. The target regulatory identification should be consistent with these rules to ensure the compliance and risk controllability of related party transactions within the enterprise. Through comprehensive verification and review of the target regulatory identification, its accuracy and consistency across regulatory requirements and business logic are ensured.

[0115] If errors or inconsistencies are found in the target regulatory identification after verification and review, the input process of the related party identity parameters and regulatory identification parameters will be immediately traced back to accurately locate and correct the errors. The traceback process starts from the calculation results of the target regulatory identification and reversely traces the processing flow of the data in the identity parameter transmission regulatory algorithm model. First, check whether the output results of the output layer are consistent with the transmission logic within the model, and confirm whether there are problems such as data conversion or output format errors in the output stage. Then, go deep into the transmission layer and check the calculation process and parameter settings of each transmission layer node one by one. In the transmission layer, focus on whether the weight connection between nodes is correct, whether the application of the activation function is as expected, and whether the offset setting is reasonable. By comparing the input data and expected output data of each node, combined with regulatory requirements and business logic, determine whether there is a regulatory identification transmission deviation caused by node calculation errors or parameter anomalies.

[0116] Continue to trace back to the input layer and carefully review the accuracy and completeness of the input related party identity parameters and regulatory identification parameters. Check whether the source of the related party identity parameters is reliable, and whether there are omissions, errors or data format mismatches during the data entry process. For regulatory identification parameters, confirm whether they are correctly set according to the latest regulatory rules and internal corporate regulations. Once the error is located, make targeted corrections based on the error type and specific circumstances. If it is a data entry error, correct the erroneous data in a timely manner; if the parameter setting is improper, readjust the parameter value; if it involves a logical error within the algorithm model, repair or optimize the relevant part of the model. After correcting the error, re-run the identity parameter transmission regulatory algorithm model, recalculate and verify the target regulatory identification to ensure that the corrected results are accurate and meet regulatory requirements and business logic. Through this rigorous verification, review and error handling mechanism, the related party regulatory identification analysis method of the present invention can effectively improve the accuracy and reliability of regulatory identification analysis, timely discover and correct potential errors, provide solid technical support for related party supervision, ensure that enterprises operate in compliance under complex regulatory environments, and reduce related party transaction risks.

[0117] In one embodiment, it further includes:

[0118] Monitor and manage transactions involving related parties of target related parties based on the target regulatory identification corresponding to the target related parties in different time periods;

[0119] Based on the target regulatory identification, mark the regulatory scope to which the related party is subject to in transactions on different transaction dates.

[0120] In the above embodiment, a dynamic transaction monitoring mechanism is established based on the target regulatory identification of the target related party in different time periods. For each time period covered by the target regulatory identification, all transaction activities involving related parties in which the target related party participates are closely monitored. In the real-time or near real-time stage of the transaction, detailed information of the transaction is collected, including key elements such as the transaction amount, transaction type (such as equity transaction, fund lending, commodity trading, etc.), and counterparty. The regulatory requirements and risk warnings contained in the target regulatory identification are used to conduct preliminary screening and risk assessment of the transaction. For example, transactions within the time period marked as high-risk regulatory identification are subject to more stringent review, focusing on whether the transaction contains potential interest transfer, insider trading, or violations of regulatory provisions.

[0121] Develop corresponding transaction management strategies based on the regulatory rules set by the target regulatory identification. If the target regulatory identification indicates that the related party is subject to strict capital flow supervision within a certain period of time, then in transaction management, the related party's large-scale capital transfers will be strictly restricted, or it will be required to obtain additional approval before conducting specific types of transactions. At the same time, based on the changing trends of the target regulatory identification, the related party's transaction behavior will be dynamically adjusted and intervened. If it is found that the regulatory identification of the related party has changed from low risk to high risk, the monitoring of its transaction activities will be strengthened in a timely manner. It may be required to increase the frequency and level of detail of transaction information disclosure, or suspend certain high-risk transaction types to prevent the further expansion of potential risks and ensure the stability and compliance of the company's overall operations.

[0122] Based on the target regulatory identifier, the regulatory scope of transactions involving related parties on different transaction dates is accurately marked. This marking process is based on the regulator's detailed definition of the regulatory rules and requirements corresponding to different regulatory identifiers. For example, a regulator stipulates that transactions involving specific types of related parties within a specific shareholding ratio should be classified under a specific regulatory scope, which involves specific reporting requirements, approval processes, and information disclosure standards. When the target related party conducts transactions within this shareholding ratio, the transaction is clearly marked as subject to the relevant provisions of this specific regulatory scope based on its corresponding target regulatory identifier.

[0123] In the transaction record system of an enterprise, for each transaction involving related parties, the corresponding regulatory scope information is clearly marked in its transaction record. This not only helps the enterprise to manage and analyze related party transactions internally, but also facilitates the provision of transaction information that meets regulatory requirements quickly and accurately when responding to regulatory inspections. At the same time, by accurately marking the regulatory scope, the enterprise can better communicate and cooperate with the regulatory authorities, ensuring that its related party transactions are conducted within the regulatory framework, avoiding compliance risks due to unclear regulatory scope. This transaction monitoring, management and regulatory scope marking mechanism based on target regulatory identification forms a complete closed-loop management system, effectively improving the enterprise's control ability over related party transactions and enhancing the enterprise's adaptability and compliance level in a complex regulatory environment.

[0124] Two specific embodiments are given below to illustrate the specific application of the method of the present application.

[0125] The first specific embodiment is as follows:

[0126] Suppose there is a large enterprise group that includes multiple subsidiaries, involving numerous related party transactions in the process of enterprise operation, which requires effective regulatory identification analysis of related parties, involving the following steps:

[0127] 1. Obtain target related party identity parameters

[0128] First, determine the target related party, for example, an important shareholder A of a subsidiary. According to the actual relationship between A and the enterprise group (the main body of related party management), A as a shareholder has a shareholding ratio of 12%, and has a certain influence in some major decisions of the enterprise group, so the shareholder identity parameter is selected. At the same time, A holds a senior consultant position in a subsidiary of the group, which also exists as a job relationship identity parameter.

[0129] Extract the related information of A from the pre-set regulatory rule library for different regulatory authorities (such as the Financial Supervision Administration and the Securities Regulatory Commission). For the regulations of the Financial Supervision Administration, shareholders with a shareholding ratio of more than 10% are considered important related parties, and related parties holding a senior consultant position have specific regulatory requirements in related party transactions. These regulations are divided into corresponding related party identity parameters, and for A's shareholder identity parameter, according to the regulations of the Financial Supervision Administration, it belongs to the category of important related parties.

[0130] For the existing regulatory identifier in A's shareholder identity parameters, according to the regulations of the Financial Regulatory Administration, the validity period is determined to be effective from the date A's shareholding reaches 10%, with no clear expiration time (unless there is a significant change in the shareholding ratio, etc.). The validity threshold parameter is the shareholding ratio threshold, that is, a shareholding of more than 10% triggers the regulatory identifier. For his position relationship identity parameters, the validity period is effective from the date he assumes the position of senior advisor, and there is also no clear expiration time (unless there is a change in position). The validity threshold parameter is the position level threshold, that is, senior advisor and above positions trigger the corresponding regulatory identifier.

[0131] 2. Training Identity Parameter Transmission Supervision Algorithm Model

[0132] To build an algorithmic model for the transmission of identity parameters, a training sample set of related-party regulatory identifications was obtained. Information on different historical related parties was screened from the enterprise group's multi-year historical regulatory data. This included the historical related-party identity parameters of various shareholders, managers, and their relatives over the past decade, as well as the historical target regulatory identifications that were actually applied to them at the time. For example, there was a shareholder named B, who held an 8% stake and held no position. Under the regulatory environment at the time, the regulatory identification that applied to him was a standard related-party identification. This information formed part of the training sample set.

[0133] The weights and bias parameters of a pre-set machine learning model (e.g., a neural network model) are randomly initialized and assigned initial values. A loss function for the pre-set machine learning model is constructed using a mean square error term and a weight decay term. The mean square error term is used to measure the difference between the target regulatory identification predicted by the model and the actual regulatory identification in the sample set, for example, the difference between the regulatory identification that should apply to B under the current circumstances and the actual ordinary related party regulatory identification. The weight decay term is used to avoid model overfitting and prevent excessive weights from causing the model to become too complex and lose its generalization ability.

[0134] Based on the training sample set of related-party regulatory identification, the gradient descent algorithm is used as the optimization algorithm. In each iteration, the historical related-party identity parameters in the sample set are input into the model to obtain the predicted target regulatory identification and calculate the loss function value. Then, based on the gradient of the loss function to the model parameters, the model parameters are updated at a certain learning rate so that the loss function value gradually decreases. The iterative update operation is repeated until the model converges (such as the loss function value no longer changes significantly) or the predetermined number of iterations (such as 1000 times) is reached. Finally, the trained identity parameter conduction regulatory algorithm model is obtained.

[0135] 3. Calculate the target regulatory identification

[0136] A's shareholder identity parameters (12% stake, effective date, etc.) and position-relationship identity parameters (Senior Advisor position, effective date, etc.) are input into the trained identity parameter transmission and supervision algorithm model. The model's input layer receives these parameters, and the multiple transmission layer nodes in the transmission layer begin to operate. Each node performs a nonlinear transformation on the input parameters, mapping and translating them using weights and offsets. For example, for the shareholder identity parameters, based on the weights learned during model training, information is exchanged and calculated with other nodes, gradually transmitting information related to their regulatory identification.

[0137] After complex calculations at the transmission layer, the output layer ultimately outputs the target regulatory designations for A at different time periods. Suppose the model output indicates that in the current time period, because A holds over 10% of the shares and holds a senior advisory position, the target regulatory designation for A is a high-risk related-party designation, indicating that transactions involving related parties are subject to stricter oversight and scrutiny.

[0138] 4. Real-time updates and model adjustments

[0139] Suppose a corporate group undergoes an equity restructuring, increasing A's shareholding to 15% and simultaneously changing his position to a member of the Supervisory Board. In this case, A's related party identity parameters are updated in real time, incorporating the new shareholding and position information into the system. Based on the latest regulatory provisions in the regulatory rule library, corresponding regulatory identification parameter changes are determined. For example, under the new regulations, related parties with a shareholding exceeding 15% and serving as members of the Supervisory Board are subject to key regulatory oversight. The validity period of the regulatory identification parameters is updated to take effect from the date of the change, and the validity threshold parameters are adjusted accordingly.

[0140] Based on the updated related party identity and regulatory identification parameters, and due to significant data changes, it was decided to retrain the identity parameter transmission supervision algorithm model. A new related party regulatory identification training sample set containing the latest information on A and updated information on other related parties was obtained. The model was initialized again, the loss function was set, and gradient descent iterative updates were performed until a new optimized model was obtained to adapt to the new regulatory requirements and the actual situation of the enterprise.

[0141] 5. Verification, review and error handling

[0142] Verify and review the calculated target regulatory designation for A. Based on the latest regulations from regulators such as the Financial Regulatory Administration and the internal business logic of the enterprise group, check whether the regulatory designation for high-risk related parties meets the requirements. For example, check whether the application of this regulatory designation to a related party with a 15% shareholding and a supervisory board member complies with the regulatory principles for risk management of such related parties and is consistent with other related party management rules within the enterprise group.

[0143] If problems are discovered during the verification and audit process, such as discovering that according to the new regulations, related parties with a shareholding ratio of 15% and serving as members of the Supervisory Board should be subject to a higher level of regulatory identification, but the model output is a high-risk related party regulatory identification, which is inconsistent. In this case, the input process of A's related party identity parameters and regulatory identification parameters is traced back. Starting from the output layer, check whether the output result has errors in the model transmission process, and then check the calculation process and weight setting of the transmission layer nodes in turn. Finally, check whether the parameter input of the input layer is accurate. After detailed investigation, it was found that a small error occurred during the weight update process of a node in the transmission layer, resulting in a deviation in the calculation result. Correct the error, recalculate A's target regulatory identification, and conduct another verification and audit to ensure the accuracy of the results.

[0144] 6. Monitoring and management of related-party transactions and regulatory caliber marking

[0145] Based on the target regulatory designation assigned to A at different time periods (currently a high-risk related-party regulatory designation), related-party transactions involving A are monitored and managed. For each related-party transaction in which A participates, such as a fund borrowing transaction with another subsidiary within the corporate group, detailed information is recorded, including the transaction amount (e.g., 5 million yuan), transaction type (fund borrowing), and counterparty (subsidiary name). Because A is a high-risk related-party, the transaction is subject to strict scrutiny, requiring a detailed explanation of the funds' use and a risk assessment report.

[0146] Based on A's target regulatory identifier, the regulatory scope of the fund lending transaction on the date it occurred is marked. According to regulatory requirements, fund lending transactions involving high-risk related parties should be categorized under a specific regulatory scope, requiring regular reporting of transaction details to regulators and meeting certain capital adequacy requirements. Within the enterprise group's transaction record system, the transaction should be clearly marked as falling under this regulatory scope to facilitate subsequent statistical analysis, regulatory reporting, and internal management. This ensures that A's related-party transactions are conducted within the correct regulatory framework, effectively reducing the enterprise group's related-party transaction risks and improving the compliance and stability of its overall operations.

[0147] The second embodiment first adds a time dimension and regulatory-specific judgment rules to the identity parameters, and requires users to enter the valid time period of the related party identity when entering the related party, adding a judgment factor. At the same time, a set of identity parameter regulatory identification transmission model calculation inversion is established to realize the measurement of regulatory identity parameter transmission, thereby improving the accuracy and timeliness of the calculation of the related party regulatory identification, reducing the dependence on the timeliness of user operations, further enhancing the refined management of related parties, and improving the quality of regulatory data submitted by banks.

[0148] The second embodiment involves the following steps:

[0149] (1) Related party identity parameter settings:

[0150] (1.1) Each regulator’s definition of related parties is split into individual related party identity parameters A (each related party identity parameter A represents a set of related parties), and each is assigned a corresponding regulatory identifier; if it is found that a related party identity parameter defined by different regulators has the same name and meaning, then multiple corresponding regulatory identifiers are assigned to this related party identity parameter.

[0151] (1.2) Based on the effective date of each regulatory party's transaction management regulations, further add effective and expiration date information to each regulatory identifier attached to the related party identity parameter A in (1.1). This establishes the related party identity parameter A(T) (the related party identity parameter A(T) represents a set of related party parameters).

[0152] (2) Identity threshold parameter settings corresponding to the related party identity:

[0153] (2.1) If the regulator specifies comparison conditions for some of the related party identity parameters A(T) in (1), threshold items and threshold values ​​are further set under the related party identity parameters; there can be multiple combinations of threshold items and threshold values. If different regulators define the same related party identity but specify different comparison conditions, multiple combinations of threshold items and threshold values ​​are set under the related party identity parameter A(T) and are assigned corresponding regulatory identifiers. This establishes the related party identity parameter A(T)B (the related party identity parameter A(T)B represents a set of related party parameters).

[0154] (2.2) Based on the effective date of the transaction management regulations for each regulatory party's related parties, further add effective and expiration dates to each regulatory identifier corresponding to the related party identity parameter A(T)B in (2.1). This establishes the related party identity parameter A(T)B(T) (the related party identity parameter A(T)B(T) represents a related party parameter set), which is referred to as the related party identity parameter E (the related party identity parameter E represents a related party parameter set).

[0155] (3) Related party entry:

[0156] (3.1) In the course of daily operations and management, enterprises shall proactively check and confirm related parties and enter them into the IT system for management. When entering, it is necessary to select the relationship between the related party and the related party management entity, or select a related party that has been stored in the database as its superior related party and select the relationship with the superior related party, that is, a related party identity parameter e in the related party identity parameter E. A group of information composed of the above-mentioned related parties, related party identity parameters e, related party management entities or superior related parties, etc., constitutes the related party identity F, which represents the relationship between the related party and the related party management entity or superior related party; according to the actual situation, set the effective and expiration time for the selected related party identity F. If there are multiple related party identities, select and enter the effective and expiration time respectively according to the above steps; the effective and expiration times of different related party identities are allowed to overlap. The above-mentioned related party identity F with the effective and expiration time attached can be named related party identity N. From this, we can see that if an affiliate has multiple superior affiliates, or has multiple association relationships with the affiliate management entity or superior affiliate, then the affiliate can have multiple affiliate identities N, which can be named affiliate identity N1, affiliate identity N2, to affiliate identity NM.

[0157] (3.2) If the associated party identity N selected in (3.1) has corresponding identity threshold parameters, further select the threshold item and enter the actual threshold value, as well as the corresponding effective and ineffective time. Depending on the actual situation, multiple threshold items can be selected and the corresponding actual threshold values ​​can be entered, along with the corresponding effective and ineffective time. The effective and ineffective time of different threshold items can overlap.

[0158] (4) Constructing an identity parameter regulatory transmission model

[0159] (4.1) Figure 2 FIG. 1 is an example diagram of the relationship between an associated party identity parameter and a supervisory identifier in an embodiment of the present invention. Figure 2 As shown in the figure, a training sample set of identity parameter associated party supervision identification is established, involving the following parameters:

[0160] {(x 1 ,y 1 ),(x 2 ,y 2 ),L,(xn,y n )}, construct a nonlinear model M w,b (x), parameters w, b, fit y through these parameters.

[0161] (4.2) The construction and transmission method in the identity parameter transmission supervision algorithm model is as follows:

[0162]

[0163] Among them, the Lth layer has kl nodes, conduction layer is the activation value of the i-th node in the l-th layer, is the input value of the lth layer. From the J node of the lth layer to the i node of the l+1th layer through the weight connection, f(q) is the activation function, and the nonlinear transformation is performed through the activation function. The result of the transformation is passed through the weight Map to the new space and pass the offset The leftmost layer of the transmission model is the input of the regulatory identity parameters of the related party involved in the transaction, the rightmost layer is the final output result after the regulatory identity transmission, and the middle layer is the transmission process layer. This application uses the sigmoid function as the activation function, as follows:

[0164]

[0165] Figure 3 This is a specific example diagram of the transmission logic of an identity parameter transmission supervision algorithm model in an embodiment of the present invention. The construction transmission method in the identity parameter transmission supervision algorithm model is as follows: Figure 3 shown.

[0166] (4.3) Constructing the objective function

[0167] If there is only one sample, the cost function is:

[0168]

[0169] There are n data samples, and the entire cost function is:

[0170]

[0171] The first part of the above formula J(w,b) is the mean square error term; the second part is the weight attenuation term, which is used to avoid overfitting caused by excessive weights.

[0172] The entire conduction model algorithm is ultimately transformed into solving the minimum value of the target cost function:

[0173]

[0174] To this end, the objective function of the entire conduction model can be solved using the gradient descent method:

[0175]

[0176] In the above formula, α is the iterative learning rate.

[0177] (5) According to the identity parameter transmission supervision algorithm model, the supervision identification of the related party and the identity parameter transmission relationship are established to achieve the measurement of the supervision identity identification of the related party. The implementation example is as follows:

[0178] Table 1 shows the correspondence between related party identity parameters and various information such as regulatory identification. Table 1 lists in detail the related party identity parameters, their corresponding regulatory identification, the validity period of the regulatory identification (including the effective date and expiration date), threshold items, thresholds, and the validity period of the threshold combination (including the effective date and expiration date). For example, it shows the various regulatory identifications corresponding to different related party identity parameters (such as related party identity parameters, related party identity parameters, etc.) and their related time limits and threshold conditions, which helps to understand the complex correspondence between related party identity parameters and regulatory identifications, as well as the specific settings of time limits and threshold parameters. An example table of related party identity basic parameters E is shown in Table 1:

[0179] Table 1

[0180]

[0181] Table 2 shows the basic information table of the related parties to be reported, showing the related party identity, the validity period of the related party identity, the threshold situation, and the corresponding validity period of the threshold. Through the example of a specific related party, the changes in its related party identity and the corresponding threshold situation in different time periods are shown, which provides a reference for understanding how to determine its identity parameters and threshold parameters based on the actual situation of the related party. It also provides a basic data example for the subsequent calculation of the target regulatory identification. Table 2 shows the identity, validity period, threshold, etc. of the related parties to be reported:

[0182] Table 2

[0183]

[0184] Table 3 is an example table of target regulatory identifications corresponding to target related parties in different time periods, which clearly shows the regulatory identifications corresponding to related parties at different start and end dates. Through the division of specific time intervals, it shows the regulatory identifications that should be effective in each time period based on the related party's related party identity parameters, threshold conditions, and the regulatory identification transmission algorithm model. It intuitively reflects the application results of the method of the present invention in determining the regulatory identification of related parties in different time periods, which helps to understand the final output and actual effect of the entire regulatory identification analysis method. The final regulatory identification calculation results of related party Y calculated by the gradient descent transmission model algorithm are shown in Table 3:

[0185] Table 3

[0186]

[0187]

[0188] Through this transmission model, related party Y can promptly obtain the regulatory identification required at each point in time. At the same time, the system can automatically monitor the validity of each regulatory identification of related party Y on a daily basis, ensuring timely and effective capture of corresponding related party-involved transactions, accurately marking the regulatory caliber of the related party-involved transactions on the date of the transaction, participating in the calculation of the regulatory business model for related party-involved transactions, various report statistics, and accurate reporting to regulators, thereby rapidly improving the quality of related party-involved transactions and meeting regulatory requirements. This invention not only optimizes the configuration of the newly added time function parameters for related party regulatory identification, but also constructs an overall transmission model method to improve the accuracy, flexibility, and scalability of regulatory identification of related party-involved transactions.

[0189] In response to changes in regulatory rules, this specific embodiment adds a time element to the identity parameters as a basic configuration of the identity parameters, improving the timeliness of the control of the time element by the transmission model's automatic transmission of its regulatory identification calculation function through the identity relationship of the related parties;

[0190] Based on changes in regulatory rules, an identity transmission network model algorithm is constructed to effectively transmit regulatory identity parameters to related party identities. This improves the flexibility and scalability of the system's automatic transmission of regulatory identification functions through identity relationships;

[0191] By updating and adjusting the parameters of the related-party regulatory identification transmission model in the same direction through the gradient descent algorithm, the transmission model process convergence is achieved, the model is quickly fitted, and the measurement of regulatory identity parameters is quickly completed, so as to achieve the timeliness and accuracy of the measurement of subsequent related-party transactions.

[0192] Of course, it is understandable that the above detailed process may have other variations, and all relevant variations should fall within the scope of protection of the present invention.

[0193] In an embodiment of the present invention, multiple related party identity parameters corresponding to the target related party are obtained; wherein, each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include timeliness parameters and validity threshold parameters corresponding to different existing regulatory identifications; the timeliness parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification; the multiple related party identity parameters corresponding to the target related party are input into the identity parameter transmission supervision algorithm model to obtain the target regulatory identification corresponding to the target related party in different time periods; the target regulatory identification is used to describe the existing regulatory identification that is effective in each time period; the identity parameter transmission supervision algorithm model is based on the gradient descent algorithm, and is obtained by training a preset machine learning model with a related party regulatory identification training sample set; the related party regulatory identification training sample set includes: historical related party identity parameters of different historical related parties, and historical target regulatory identifications corresponding to different historical related parties. The embodiment of the present invention clarifies the effective time and expiration time of the existing regulatory identification by setting the timeliness parameter. When the regulatory rules change, it can be timely judged whether the regulatory identification needs to be updated based on the timeliness parameter, rather than relying on manual intervention. The regulatory algorithm model transmitted by the identity parameter can flexibly adapt to the changes in regulatory rules, ensuring that the regulatory identification of the target related party can be quickly updated when the regulatory rules change, thereby solving the problem of untimely update of the regulatory identification; by setting the validity threshold parameter, the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification are limited, so that when calculating the target regulatory identification, accurate judgment can be made based on the differentiated regulations of different regulators, and the machine The machine learning model can learn the accurate calculation logic under different regulatory rules, thereby improving the accuracy of regulatory identification calculation and overcoming the defects of inaccurate calculation in existing technologies; in addition, by constructing an identity parameter transmission regulatory algorithm model, it can automatically calculate the target regulatory identification corresponding to different time periods based on the input related party identity parameters, reducing the dependence on users to maintain the related party identity in the system in a timely manner. Even if the related party identity changes, the regulatory identification can be quickly recalculated based on the new related party identity parameters and algorithm model, avoiding regulatory identification errors or untimely updates due to user maintenance delays, and effectively solving the problem that related party identity maintenance affects the accuracy of regulatory identification.

[0194] As described above, this embodiment of the present invention adds a time dimension to identity parameters to avoid the time dimension not being transmitted, which can lead to delayed updates to the time identity parameters during the calculation of the related party's supervisory identifier and the inability to quickly respond to changes in the time of the related party's supervisory identifier. This improves the timeliness of the transmission model's supervisory identifier calculation function, which automatically transmits the time element through the related party identity relationship.

[0195] The embodiment of the present invention builds a set of identity transmission network model algorithms, which instantiate and measure the transaction identity parameters of the entire related party and the final regulatory identity mark through an effective transmission model.

[0196] The embodiment of the present invention updates and adjusts the parameters of the related party regulatory identification transmission model in the same direction through the gradient descent algorithm to achieve the minimization of the cost function; at the same time, it adjusts the iterative learning rate to control the step size of the parameter update, so as to achieve rapid convergence of the transmission model process and then achieve rapid model fitting, thereby quickly completing the measurement of regulatory identity parameters and achieving better regulatory identity identification transmission effect.

[0197] This embodiment of the present invention adds time and regulatory-specific judgment rule elements to the identity parameters, building a related-party regulatory identification transmission model. Through this transmission algorithm model, the regulatory identification information of related parties can be calculated promptly and quickly, meeting regulatory and internal control management requirements. This not only enhances the refinement of related-party management, but also improves the quality of regulatory data submitted.

[0198] In the prior art, since the automatic configuration function of the regulatory identification of related parties does not take into account the timeliness of regulatory rules, it relies on developers to manually adjust the system solidification model, which easily leads to a delay in the update of regulatory identification. In the embodiment of the present invention, when obtaining multiple related party identity parameters corresponding to the target related party, the regulatory identification parameters corresponding to each related party identity parameter include a timeliness parameter, which specifies the effective time and expiration time of the existing regulatory identification. When the regulatory rules change, the system can judge whether the regulatory identification needs to be updated in a timely manner based on these timeliness parameters, rather than relying on manual intervention. At the same time, the identity parameter transmission supervision algorithm model is trained based on the gradient descent algorithm and the related party regulatory identification training sample set containing rich historical information. It can adapt to changes in regulatory rules more flexibly, adjust the calculation logic in time, and ensure that the regulatory identification of the target related party can be quickly updated when the rules change, thereby solving the problem of untimely update of regulatory identification.

[0199] Different regulatory rules have different definitions of related parties and specific identity judgment standards. The existing related party regulator calculation model does not effectively distinguish these differences, resulting in inaccurate calculations. In an embodiment of the present invention, the corresponding validity threshold parameter is determined when the related party identity parameter is obtained. The parameter describes in detail the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification, such as shareholding ratio thresholds, position level thresholds, etc. This makes it possible to make accurate judgments based on the differentiated regulations of different regulators when calculating the target regulatory identification. In addition, the identity parameter transmission supervision algorithm model has been trained on a large number of historical related party identity parameters and historical target regulatory identifications of different historical related parties, and can learn the accurate calculation logic under different regulatory rules, thereby improving the accuracy of regulatory identification calculations and overcoming the defects of inaccurate calculations in the existing technology.

[0200] The existing automatic configuration function of related party regulatory identification relies on timely maintenance by users for regulatory identification updates caused by changes in related party identities, which poses a risk of delays and errors. The method in the embodiment of the present invention constructs a complete identity parameter transmission regulatory algorithm model. The model is trained based on historical data and can automatically calculate the target regulatory identification corresponding to different time periods based on the input related party identity parameters. This reduces the reliance on users to maintain related party identities in the system in a timely manner. Even if the related party identity changes, such as changes in shareholding ratios, job adjustments, etc., the system can quickly recalculate the regulatory identification based on the new related party identity parameters and algorithm model, avoiding regulatory identification errors or untimely updates due to user maintenance delays, and effectively solving the problem of related party identity maintenance affecting the accuracy of regulatory identification.

[0201] The present invention also provides a device for analyzing the supervisory identification of related parties, as described in the following embodiments. Since the principle of solving the problem of this device is similar to that of the method for analyzing the supervisory identification of related parties, the implementation of this device can refer to the implementation of the method for analyzing the supervisory identification of related parties, and the repeated parts will not be repeated.

[0202] Figure 6 The present invention also provides a related party supervision identification analysis device for improving the accuracy, timeliness and efficiency of related party supervision identification analysis. Figure 6 As shown, the device includes:

[0203] A related party identity parameter acquisition module 601 is used to acquire multiple related party identity parameters corresponding to the target related party; each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include timeliness parameters and validity threshold parameters corresponding to different existing regulatory identifications; the timeliness parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification;

[0204] The target supervision identifier output module 602 is used to input multiple related party identity parameters corresponding to the target related party into the identity parameter transmission supervision algorithm model to obtain the target supervision identifier corresponding to the target related party in different time periods; the target supervision identifier is used to describe the existing supervision identifier effective in each time period; the identity parameter transmission supervision algorithm model is based on the gradient descent algorithm, and is obtained by training a preset machine learning model with a related party supervision identifier training sample set; the related party supervision identifier training sample set includes: historical related party identity parameters of different historical related parties, and historical target supervision identifiers corresponding to different historical related parties.

[0205] In one embodiment, the related party identity parameters include but are not limited to shareholder identity, job relationship identity and kinship identity; the validity threshold parameters of each related party identity parameter include but are not limited to shareholding ratio threshold, job level threshold, and kinship type.

[0206] In one embodiment, obtaining multiple related party identity parameters corresponding to the target related party includes:

[0207] Select corresponding related party identity parameters for the target related party based on the actual relationship between the target related party and the related party management entity or the superior related party;

[0208] Extract relevant information about the association of different regulators with the target related party from the preset regulatory rule library;

[0209] Split the related information of each supervisory party on the target related party into related party identity parameters corresponding to the supervisory structure;

[0210] For each existing supervisory identifier in the identity parameter of the associated party, the timeliness parameters and validity threshold parameters of different existing supervisory identifiers are determined based on the associated relevant information of the supervisor.

[0211] In one embodiment, the method further includes: obtaining an identity parameter conduction supervision algorithm model through training as follows:

[0212] Obtain a training sample set of related party regulatory identifications; randomly initialize the weights and bias parameters of the preset machine learning model;

[0213] A loss function is pre-set for the machine learning model; the loss function is used to measure the difference between the target regulatory identifier predicted by the model and the actual regulatory identifier in the sample set;

[0214] Using the related party regulatory identification training sample set, the gradient descent algorithm is used as the optimization algorithm to iteratively update the randomly initialized machine learning model to gradually reduce the value of the loss function;

[0215] Repeat the above iterative update operation until the iteratively updated model converges or reaches a predetermined number of iterations; the model obtained by stopping the iterative update is used as the trained identity parameter conduction supervision algorithm model.

[0216] In one embodiment, the identity parameter transmission supervision algorithm model includes:

[0217] An input layer, for receiving multiple related party identity parameters corresponding to the target related party;

[0218] The conduction layer is used for nonlinear conversion of the input associated party identity parameter, mapping and translation change through weight and offset, conduction of the regulatory identifier of the associated party identity parameter to the target associated party, and output of the target regulatory identifier corresponding to the target associated party at different time periods through the output layer; the conduction layer comprises a plurality of conduction layer nodes, and each conduction layer node represents a state in the conduction process of the regulatory identifier of the associated party identity parameter.

[0219] The output layer is used for output of the target regulatory identifier corresponding to the target associated party at different time periods.

[0220] In one embodiment, the method further comprises:

[0221] The associated party identity parameter and the corresponding regulatory identifier parameter are updated in real time.

[0222] According to the updated associated party identity parameter and the regulatory identifier parameter, the identity parameter conduction regulatory algorithm model is retrained or fine-tuned.

[0223] In one embodiment, the method further comprises:

[0224] The calculated target regulatory identifier is verified and audited; the verification and audit are used to determine whether the target regulatory identifier meets the regulatory requirements and business logic.

[0225] If the target regulatory identifier is found to have errors or inconsistencies after verification and audit, the input process of the associated party identity parameter and the regulatory identifier parameter is traced back, and the error is located and corrected.

[0226] In one embodiment, the objective function of the identity parameter conduction regulatory algorithm model is constructed based on a mean square error term and a weight decay term; the mean square error term is used to measure the difference between the predicted value and the true value; the weight decay term is used to avoid overfitting phenomenon caused by too large weight, and the objective function is solved by using the gradient descent method to optimize the model parameters.

[0227] In one embodiment, the method further comprises:

[0228] According to the target regulatory identifier corresponding to the target associated party at different time periods, the transaction of the target associated party involving the associated party is monitored and managed.

[0229] Based on the target regulatory identifier, the regulatory scope to which the transaction involving the associated party belongs under different transaction dates is marked.

[0230] The embodiment of the computer device for implementing all or part of the contents of the above-mentioned associated party regulatory identifier analysis method is provided.

[0231] A processor, a memory, a communications interface, and a bus; wherein the processor, memory, and communications interface communicate with each other via the bus; the communications interface is used to implement information transmission between related devices; the computer device can be a desktop computer, a tablet computer, a mobile terminal, etc., but this embodiment is not limited thereto. In this embodiment, the computer device can be implemented with reference to the embodiment for implementing the method for analyzing the regulatory identification of an associated party and the embodiment for implementing the apparatus for analyzing the regulatory identification of an associated party, the contents of which are incorporated herein and repeated parts are not repeated.

[0232] Figure 7 1 is a schematic block diagram of the system structure of the computer device 1000 according to an embodiment of the present application. Figure 7 As shown, the computer device 1000 may include a central processor 1001 and a memory 1002; the memory 1002 is coupled to the central processor 1001. Figure 7 is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0233] In one embodiment, the analysis function of the supervisory identification of the related party may be integrated into the central processing unit 1001. The central processing unit 1001 may be configured to perform the following control:

[0234] Obtain multiple related party identity parameters corresponding to the target related party; wherein each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include timeliness parameters and validity threshold parameters corresponding to different existing regulatory identifications; the timeliness parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification;

[0235] The multiple related party identity parameters corresponding to the target related party are input into the identity parameter transmission supervision algorithm model to obtain the target supervision identifier corresponding to the target related party in different time periods; the target supervision identifier is used to describe the existing supervision identifier effective in each time period; the identity parameter transmission supervision algorithm model is based on the gradient descent algorithm, and is obtained by training a preset machine learning model with a related party supervision identifier training sample set; the related party supervision identifier training sample set includes: historical related party identity parameters of different historical related parties, and historical target supervision identifiers corresponding to different historical related parties.

[0236] In another embodiment, the regulatory identification analysis device of the associated party can be configured separately from the central processor 1001. For example, the regulatory identification analysis device of the associated party can be configured as a chip connected to the central processor 1001, and the regulatory identification analysis function of the associated party can be realized through the control of the central processor.

[0237] like Figure 7 As shown, the computer device 1000 may further include: a communication module 1003, an input unit 1004, an audio processor 1005, a display 1006, and a power supply 1007. It is worth noting that the computer device 1000 does not necessarily have to include Figure 7 In addition, the computer device 1000 may also include all components shown in Figure 7 For components not shown, reference may be made to the prior art.

[0238] like Figure 7 As shown, the central processing unit 1001 is sometimes also referred to as a controller or an operation control unit, and may include a microprocessor or other processor device and / or logic device. The central processing unit 1001 receives inputs and controls the operations of various components of the computer device 1000 .

[0239] Memory 1002 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned device-related information and may also store programs that execute the relevant information. The CPU 1001 may execute the programs stored in memory 1002 to implement information storage or processing.

[0240] Input unit 1004 provides input to CPU 1001. Input unit 1004 may be, for example, a keypad or touch input device. Power supply 1007 is used to provide power to computer device 1000. Display 1006 is used to display objects such as images and text. This display may be, for example, an LCD display, but is not limited thereto.

[0241] The memory 1002 may be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), or SIM card. Alternatively, it may be a memory that retains information even when power is off, can be selectively erased, and is provided with more data. Examples of such memory are sometimes referred to as EPROMs. The memory 1002 may also be some other type of device. The memory 1002 includes a buffer memory 1021 (sometimes referred to as a buffer). The memory 1002 may include an application / function storage unit 1022 for storing application programs and function programs or processes used by the central processing unit 1001 to execute the operations of the computer device 1000.

[0242] The memory 1002 may also include a data storage unit 1023 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the computer device. The driver storage unit 1024 of the memory 1002 may include various driver programs for the computer device for communication functions and / or for executing other functions of the computer device (such as messaging applications, address book applications, etc.).

[0243] The communication module 1003 is a transmitter / receiver that sends and receives signals via the antenna 1008. The communication module (transmitter / receiver) 1003 is coupled to the central processor 1001 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0244] Based on different communication technologies, multiple communication modules 1003 can be provided in the same computer device, such as a cellular network module, a Bluetooth module, and / or a wireless local area network module. The communication module (transmitter / receiver) 1003 is also coupled to a speaker 1009 and a microphone 1010 via an audio processor 1005 to provide audio output via the speaker 1009 and receive audio input from the microphone 1010, thereby implementing common telecommunication functions. The audio processor 1005 may include any suitable buffer, decoder, amplifier, etc. Furthermore, the audio processor 1005 is coupled to the central processing unit 1001, enabling local recording via the microphone 1010 and playback of stored audio via the speaker 1009.

[0245] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, which implements the above-mentioned method for analyzing the supervisory identification of related parties when executed by a processor.

[0246] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the above-mentioned method for analyzing the regulatory identification of related parties.

[0247] In an embodiment of the present invention, multiple related party identity parameters corresponding to a target related party are obtained; wherein each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include time validity parameters and validity threshold parameters corresponding to different existing regulatory identifications; the time validity parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification;

[0248] The multiple related party identity parameters corresponding to the target related party are input into the identity parameter transmission supervision algorithm model to obtain the target supervision identifier corresponding to the target related party in different time periods; the target supervision identifier is used to describe the existing supervision identifier effective in each time period; the identity parameter transmission supervision algorithm model is based on the gradient descent algorithm and is obtained by training a pre-set machine learning model with a related party supervision identifier training sample set; the related party supervision identifier training sample set includes: historical related party identity parameters of different historical related parties, and historical target supervision identifiers corresponding to different historical related parties. The embodiment of the present invention clarifies the effective time and expiration time of the existing supervision identifier by setting a timeliness parameter. When the supervision rules change, it can be timely determined whether the supervision identifier needs to be updated based on the timeliness parameter instead of relying on manual intervention. The identity parameter transmission supervision algorithm model can flexibly adapt to changes in supervision rules, ensuring that the supervision identifier of the target related party can be quickly updated when the supervision rules change, thereby solving the problem of untimely supervision identifier updates; by setting the validity threshold parameter, the supervision conditions and condition thresholds for the effectiveness of the existing supervision identifier are limited, so that when calculating the target supervision identifier, it can accurately determine the differentiated regulations of different regulators. The machine learning model can learn the accurate calculation logic under different regulatory rules, thereby improving the accuracy of regulatory identification calculation and overcoming the defects of inaccurate calculation in existing technologies; in addition, by constructing an identity parameter transmission regulatory algorithm model, it can automatically calculate the target regulatory identification corresponding to different time periods based on the input related party identity parameters, reducing the dependence on users to maintain the related party identity in the system in a timely manner. Even if the related party identity changes, the regulatory identification can be quickly recalculated based on the new related party identity parameters and algorithm model, avoiding regulatory identification errors or untimely updates due to user maintenance delays, and effectively solving the problem that related party identity maintenance affects the accuracy of regulatory identification.

[0249] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0250] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0251] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0252] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0253] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for analyzing the regulatory identification of related parties, characterized in that: include: Obtain multiple related party identity parameters corresponding to the target related party; wherein each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include timeliness parameters and validity threshold parameters corresponding to different existing regulatory identifications; the timeliness parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification; Inputting multiple related party identity parameters corresponding to the target related party into the identity parameter transmission supervision algorithm model to obtain the target supervision identifier corresponding to the target related party in different time periods; the target supervision identifier is used to describe the existing supervision identifier effective in each time period; the identity parameter transmission supervision algorithm model is based on a gradient descent algorithm and is obtained by training a preset machine learning model with a related party supervision identifier training sample set; the related party supervision identifier training sample set includes: historical related party identity parameters of different historical related parties, and historical target supervision identifiers corresponding to different historical related parties; It also includes: obtaining an identity parameter transmission supervision algorithm model through training as follows: obtaining a training sample set of related party supervision identification; randomly initializing the weights and bias parameters of a preset machine learning model; presetting a loss function of the machine learning model; the loss function is used to measure the difference between the target supervision identification obtained by the model prediction and the actual supervision identification in the sample set; using the related party supervision identification training sample set, adopting a gradient descent algorithm as an optimization algorithm, and iteratively updating the randomly initialized machine learning model to gradually reduce the value of the loss function; repeating the above iterative update operation until the iteratively updated model converges or reaches a predetermined number of iterations; and using the model obtained by stopping the iterative update as the trained identity parameter transmission supervision algorithm model; The objective function of the identity parameter conduction supervision algorithm model is constructed based on the mean square error term and the weight decay term; the mean square error term is used to measure the difference between the predicted value and the true value; the weight decay term is used to avoid overfitting caused by excessive weight, and the gradient descent method is used to solve the objective function to optimize the model parameters.

2. The method according to claim 1, wherein The related party identity parameters include but are not limited to shareholder identity, job relationship identity and kinship identity; the validity threshold parameters of each related party identity parameter include but are not limited to shareholding ratio threshold, job level threshold, and kinship type.

3. The method according to claim 1, wherein Get multiple related party identity parameters corresponding to the target related party, including: Select corresponding related party identity parameters for the target related party based on the actual relationship between the target related party and the related party management entity or the superior related party; Extract relevant information about the association of different regulators with the target related party from the preset regulatory rule library; Split the related information of each supervisory party on the target related party into related party identity parameters corresponding to the supervisory structure; For each existing supervisory identifier in the identity parameter of the associated party, the timeliness parameters and validity threshold parameters of different existing supervisory identifiers are determined based on the associated relevant information of the supervisor.

4. The method according to claim 1, wherein The identity parameter transmission supervision algorithm model includes: An input layer, for receiving multiple related party identity parameters corresponding to the target related party; The transmission layer is used to perform nonlinear transformation on the input related party identity parameters, and perform mapping and translation changes through weights and offsets to transmit the regulatory identification of the related party identity parameters to the target related party, and output the target regulatory identification corresponding to the target related party in different time periods through the output layer; the transmission layer includes multiple transmission layer nodes, each transmission layer node represents a state in the transmission process of the regulatory identification of the related party identity parameters; The output layer is used to output the target regulatory identification corresponding to the target related party in different time periods.

5. The method according to claim 1, wherein Also includes: Update related party identity parameters and corresponding regulatory identification parameters in real time; Retrain or fine-tune the identity parameter transmission supervision algorithm model based on the updated related party identity parameters and regulatory identification parameters.

6. The method according to claim 1, wherein Also includes: Verify and review the calculated target regulatory identification; The verification and review are used to determine that the target regulatory identification complies with regulatory requirements and business logic; If verification and review reveal errors or inconsistencies in the target regulatory identification, the input process of the related party identity parameters and regulatory identification parameters will be traced back to locate and correct the errors.

7. The method according to claim 1, wherein Also includes: Monitor and manage transactions involving related parties of target related parties based on the target regulatory identification corresponding to the target related parties in different time periods; Based on the target regulatory identification, mark the regulatory scope to which the related party is subject to in transactions on different transaction dates.

8. A device for analyzing the supervision mark of an associated party, characterized in that: include: A related party identity parameter acquisition module is used to obtain multiple related party identity parameters corresponding to the target related party; each related party identity parameter corresponds to one or more regulatory identification parameters; the regulatory identification parameters include timeliness parameters and validity threshold parameters corresponding to different existing regulatory identifications; the timeliness parameters are used to describe the effective time and expiration time of the existing regulatory identification; the validity threshold parameters are used to describe the regulatory conditions and condition thresholds for the effectiveness of the existing regulatory identification; A target supervision identifier output module is configured to input multiple related party identity parameters corresponding to the target related party into an identity parameter transmission supervision algorithm model to obtain target supervision identifiers corresponding to the target related party in different time periods; the target supervision identifiers are used to describe existing supervision identifiers effective in each time period; the identity parameter transmission supervision algorithm model is based on a gradient descent algorithm and is obtained by training a preset machine learning model using a related party supervision identifier training sample set; the related party supervision identifier training sample set includes: historical related party identity parameters of different historical related parties, and historical target supervision identifiers corresponding to different historical related parties; The target regulatory identification output module is further used to: obtain an identity parameter transmission regulatory algorithm model through training as follows: obtain a training sample set of related party regulatory identification; randomly initialize the weights and bias parameters of a preset machine learning model; pre-set the loss function of the machine learning model; the loss function is used to measure the difference between the target regulatory identification predicted by the model and the actual regulatory identification in the sample set; using the related party regulatory identification training sample set, adopting the gradient descent algorithm as the optimization algorithm, and iteratively updating the randomly initialized machine learning model to gradually reduce the value of the loss function; repeating the above iterative update operation until the iteratively updated model converges or reaches a predetermined number of iterations; and using the model obtained by stopping the iterative update as the trained identity parameter transmission regulatory algorithm model; The objective function of the identity parameter conduction supervision algorithm model is constructed based on the mean square error term and the weight decay term; the mean square error term is used to measure the difference between the predicted value and the true value; the weight decay term is used to avoid overfitting caused by excessive weight, and the gradient descent method is used to solve the objective function to optimize the model parameters.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

11. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images