A dynamic trust decision notarization method and system based on Bayesian reasoning

Abstract

The present invention belongs to the field of blockchain technology, and specifically relates to a dynamic trust decision notarization method and system based on Bayesian reasoning. The present invention prevents illegal transactions by designing a reward and punishment mechanism of a regulatory agency, prompting notaries and users to dynamically adjust transaction strategies through Bayesian reasoning. Notaries derive posterior beliefs based on prior beliefs to make strategy selections. The sender derives the current posterior beliefs and the queue length of each notary based on prior beliefs to select which notary to process the transaction. The receiver also updates the posterior beliefs through Bayesian reasoning to determine the transaction strategy. After the transaction is completed, the notary synchronously updates the blockchain status and sends confirmation to both parties. The present invention combines notary cross-chain and Bayesian reasoning to better ensure the transaction quality under blockchain cross-chain, and provide strong support for improving transaction quality and reducing transaction violations.

Description

Technical Field

[0001] The present invention belongs to the technical field of blockchain, and specifically relates to a dynamic trust decision notarization method and system based on Bayesian reasoning. Background Art

[0002] Blockchain is widely used in various fields because of its characteristics of immutability, open and transparent operation, and good protection of message storage in real scenarios. However, in the application process of blockchain, since blockchain cannot directly interoperate between different systems, information between different systems cannot be interacted, and value transfer between blockchains has become an urgent problem to be solved. Therefore, cross-chain technology came into being to solve the problem of asset transfer between different blockchains and realize data sharing and business collaboration between different blockchains, which has important theoretical significance and practical value. Among them, the notary mechanism is an important implementation method in cross-chain technology. Through one or more notaries as trusted third parties, they continuously monitor the information on the chain, and verify and forward the obtained event information to the target chain. This mechanism simplifies the process of cross-chain asset exchange and transfer through a centralized processing mode, supports the free transaction of cross-chain contracts and asset mortgages, and makes interoperability between chains simple.

[0003] With the rapid development of notaries across chains, the trust of transaction users in notaries has led to the problem of notaries' violations of regulations, which has seriously undermined the overall trust of the system. In order to effectively solve the problem of notaries' violations of regulations, it is necessary to improve the detection and evaluation methods, continuously promote and adjust the supervision process, and provide a powerful supervision method for the platform. However, due to the many shortcomings of the supervision method, some notaries are lucky enough to handle the transaction process lazily. These people often lead to omissions in transaction review, illegal transactions are carried out through transactions, and there may even be risks of notaries violating regulations. How to effectively prevent notaries from violating regulations and improve the accuracy and efficiency of notaries in handling transaction processes has also become a challenge. The existing methods need to improve the investment in dealing with such notaries, and it is difficult to support the enthusiasm of notaries to standardize transaction processing in a cross-chain environment. Summary of the invention

[0004] In order to solve the trust problem of notaries under cross-chain and realize more reliable and efficient cross-chain transaction processing, the purpose of the present invention is to propose a dynamic trust decision notarization method and system based on Bayesian reasoning, by integrating the idea of ​​game theory into the notary cross-chain, constructing the game of mutual competition between notaries and mutual trust between transaction users and notaries, and allowing regulatory agencies to supervise the system as a whole to achieve multi-faceted stability and optimization. In the game, the system evaluates the notary's reputation value in order to better observe the notary's behavioral information. The strategy of each participant in the game is optimal in all subsequent games starting from the information set, or in other words, all participants are sequentially rational. After multiple rounds of games, a refined Bayesian equilibrium is finally reached, the notaries improve their work, and the transaction users trust the notaries' improved strategies.

[0005] The present invention is achieved through the following technical solutions.

[0006] A dynamic trust decision notarization method based on Bayesian reasoning, the steps are as follows:

[0007] Step 1: Build a gaming environment by designing the regulatory agency’s regulatory strategy and reward and punishment behaviors, and use the regulatory strategy to take targeted precautions against notaries’ different types of illegal trading strategy choices, and guide notaries and trading users to update their own trading strategies to meet system requirements through Bayesian reasoning;

[0008] Step 2: Set the initial notary types and type ratios. Each type of notary has a corresponding trading strategy. The notary derives the current posterior beliefs based on the prior beliefs to choose its own trading strategy.

[0009] Step 3: The sender derives the current posterior belief and the queue length of each notary based on the prior belief to select which notary to process the transaction; Step 4: Through the Bayesian reasoning and strategy adjustment model, the receiver derives the current posterior belief based on the prior belief and determines the transaction strategy of its choice based on the posterior belief;

[0010] Step 5: The notary ensures that the transaction is complete, the relevant blockchains are modified accordingly, and sends a transaction confirmation notification to the sender and receiver, waiting for the transaction confirmation;

[0011] Step 6: Simulate a cross-chain environment where multiple notaries process transactions. Set a notary to only process one transaction at a time. At the same time, control the total number of transactions initiated by transaction users in the system to remain within the set range, simulating the cross-chain transaction scenario of notaries in actual situations.

[0012] Further preferably, in step one, guiding the notary and the transaction users to update their own transaction strategies through Bayesian reasoning meets the system requirements, which means that the notary and the transaction users judge the probability of each regulatory strategy through current information, and based on the probability distribution of the regulatory strategy, use Bayesian reasoning to update the probability of their own transaction strategies being discovered under each type of regulatory strategy, thereby selecting the optimal transaction strategy; and as the transaction situation is updated, the regulatory agency will update the regulatory strategy based on the detected violations to ensure the stability of the cross-chain environment.

[0013] For further optimization, step two specifically refers to: the notary will have a certain prior knowledge of the market environment, transaction user behavior and regulatory strategy in the initial stage; in the observation of transactions, by observing the specific circumstances of the transactions, use Bayesian reasoning to update his own beliefs, and then choose the trading strategy that best suits his expectations based on the updated posterior beliefs; when choosing a trading strategy, model his own expectations by constructing a strategy selection model in game theory. The notary needs to consider the behavior and trading strategies of other participants, as well as his own expectations, and determine the optimal trading strategy by analyzing the expectation matrix under different strategy combinations.

[0014] Further optimization, step three is as follows: the sender first forms a priori trust probability distribution for different notaries based on previous understanding of notaries and market conditions, and then, in each transaction, takes into account the currently observed information and updates the trust in each notary through Bayesian reasoning; finally, using the decision tree method, comprehensively consider factors such as trust and queue length to select the optimal notary for transaction processing.

[0015] Further optimization, step five is specifically as follows: based on the blockchain technology principles and consistency algorithm, the notary ensures that the transaction is modified accordingly on the relevant blockchain; in cross-chain transactions, the notary coordinates the interaction between different blockchains to ensure that the transaction can be correctly confirmed and recorded on both the source blockchain and the target blockchain, and finally sends a message to the transaction user to confirm the transaction. The transaction user verifies the transaction and confirms the transaction only after obtaining the correct record.

[0016] The present invention also provides a dynamic trust decision notarization system based on Bayesian reasoning, which is composed of a regulatory agency decision layer, a notary decision layer, a sender decision layer, a receiver decision layer, and a game decision theory layer;

[0017] The regulatory agency decision-making layer: The regulatory agency completes the initialization operation before the cross-chain, and continuously updates the regulatory strategy based on the prior beliefs and the overall information collected by the system;

[0018] Notary decision-making layer: In the initial stage, the notary needs to prepare to start cross-chain transaction processing, wait for transactions to be received in the system, and observe the situations of other participants and their own situations, constantly update their own transaction strategy selection according to prior beliefs, and then execute the updated transaction strategy in transaction processing;

[0019] The sender decision-making layer: In the initial stage, the sender has transactions that need to be processed. It enters the system, initiates cross-chain transactions, obtains the information of the notary and regulatory agency currently observed, and calculates the notary selection priority of the current information based on its own prior beliefs. Then, it generates standardized cross-chain transaction information and passes it to the designated notary, waiting for subsequent related operations.

[0020] The decision-making layer of the receiver: The receiver waits for transactions in the system based on the transactions learned from the sender in advance, monitors the transaction processing process, calculates the probability of selecting the notary type and the probability of the selected transaction strategy based on its own prior beliefs, and determines whether to agree to the transaction based on the decision result, and then proceeds with the subsequent transaction processing;

[0021] At the game decision theory level, all participants are in a cross-chain environment and can only obtain incomplete information in the system. Each participant follows his or her own optimal trading strategy based on the incomplete information he or she obtains, and uses Bayesian reasoning to update posterior beliefs based on the observed information to select the current trading strategy.

[0022] Further optimization, the supervision strategy update method is as follows:

[0023] (1);

[0024] (2);

[0025] in, represents one of the supervision strategies, Regulatory strategies The new and old probabilities of is the inertia coefficient, For regulatory strategies The number of violations detected under is the total number of violations, represents the number of notaries, Indicates A collection of successful transaction records of notaries, It is an indicator function, which is used to check the number of violations in the notary's transaction records under the current regulatory policy and determine whether the transaction records are suspected of violations.

[0026] Further optimization, in the notary decision-making layer, the notary selects the transaction strategy based on the reputation model theory, updates the reputation value by considering the notary's past reputation value and recent transaction behavior, and introduces weights of data accuracy, response timeliness, data completeness, and normalized transaction time in the reputation value calculation; the reputation value guides the behavioral strategies executed by other participants, and the notary analyzes and calculates the expectations brought about by different transaction strategies of its own, and adjusts its own decision-making sensitivity.

[0027] Further optimization, in the decision-making layer of the sender, when calculating the notary selection priority, first based on the expected utility theory, through the observed recent transaction situation and reputation value of the notary, plus the proportion of notary types and all transaction strategies of the notary known to oneself, a decision is made based on the current cross-chain environment state, trying to achieve one's own expected results, and combining Bayesian reasoning to calculate one's own trust probability for each notary in the cross-chain; then observe the number of waiting transactions for each notary in the current cross-chain to obtain the idleness index of the queue, as a consideration of the actual situation of the transaction waiting time to meet the transaction; finally, add a comprehensive adjustment item, set the trust probability, the idleness index of the queue, and the weight of the comprehensive adjustment item, and calculate the priority of each notary.

[0028] Further optimization, in the decision-making layer of the receiver, first, use Bayesian reasoning to obtain the posterior probability of the type of honest notary in the current environment, the probability of lazy notary in the current situation, and the probability of malicious notary in the current situation; then, considering the transaction cost, the impact of refusing to trade, and the result of successful transaction, calculate the expected evaluation under different decisions; finally, convert the expectation into trust probability to update the trust probability of the notary, and derive the decision-making behavior pattern of the receiver under uncertain conditions.

[0029] The present invention adopts the Bayesian dynamic game theory to model the notary mechanism. The Bayesian dynamic game environment is simulated. The notary cross-chain environment satisfies the condition that all participants cannot obtain all the information of other participants. Participants can only obtain the prior beliefs of other participants based on the partial information that can be known. Then, given the strategies of other participants and their own posterior probabilities, the strategy of each participant is optimal in all subsequent games starting from the information set, or in other words, all participants are sequentially rational. To achieve the Bayesian equilibrium process, all participants execute their own behavioral strategies in turn according to the cross-chain process, and update their own posterior beliefs with the new information set generated by each transaction. Based on the updated beliefs, the participants continue to adjust their strategies and strive to achieve their own optimal strategies in the cross-chain interactive environment. After multiple rounds of such strategy adjustments and interactions, all participants are within their respective information constraints and capabilities. , Achieve a high degree of fit with the strategies of the cross-chain environment and other participants, and reach a stable equilibrium state. In this state, the strategic choices of all parties restrict and promote each other, and jointly promote the efficient and orderly operation of the cross-chain ecosystem. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a flowchart of a dynamic trust decision notarization method based on Bayesian reasoning;

[0031] Figure 2 It is a schematic diagram of the overall structure of dynamic trust decision notarization based on Bayesian reasoning;

[0032] Figure 3 This is a graph showing the changing trend of the reputation value of each type of notary under 100 transactions;

[0033] Figure 4 This is the probability distribution diagram of the malicious notary’s choice of transaction strategy under 100 transactions. DETAILED DESCRIPTION

[0034] The present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0035] Reference Figure 1 , a dynamic trust decision notarization method based on Bayesian reasoning, the steps are as follows:

[0036] Step 1: Build a gaming environment by designing the regulatory agency's regulatory strategy and reward and punishment behaviors, and use the regulatory strategy to take targeted precautions against notaries' different types of illegal trading strategy choices, and guide notaries and trading users to update their own trading strategies through Bayesian reasoning to meet system requirements. The purpose is to make the behavior of other participants driven by rewards meet the overall goal of the system. Among them, three different regulatory strategies are designed to take targeted precautions against notaries' different types of illegal trading strategy choices, and to warn notaries to handle transactions in a standardized manner through regulatory strategies. Notaries and trading users judge the probability of each regulatory strategy through current information, and use Bayesian reasoning to update the probability of their own trading strategies being discovered under each type of regulatory strategy based on the probability distribution of regulatory strategies, so as to select the optimal trading strategy. As the transaction situation is updated, the regulatory agency will update the regulatory strategy based on the violations detected to ensure the stability of the cross-chain environment.

[0037] Step 2: Set the initial notary type and type ratio. Each type of notary has a corresponding trading strategy. The notary derives the current posterior belief based on the prior prior belief to choose its own trading strategy. The notary will have a certain prior knowledge of the market environment, transaction user behavior and regulatory strategy in the initial stage. In the process of observing transactions, by observing the specific circumstances of the transaction, use Bayesian reasoning to update your beliefs, and then choose the trading strategy that best suits your expectations based on the updated posterior belief. When choosing a trading strategy, the notary models its own expectations by building a strategy selection model in game theory. The notary needs to consider the behavior and trading strategies of other participants, as well as its own expectations, and determine the optimal trading strategy by analyzing the expectation matrix under different strategy combinations.

[0038] Step 3: The sender derives the current posterior belief and the queue length of each notary based on the prior belief to select which notary to process the transaction. The sender first forms a priori trust probability distribution for different notaries based on previous understanding of notaries and market conditions. Then, in each transaction, the sender updates the trust in each notary through Bayesian reasoning, taking into account the currently observed information. Finally, the decision tree method is used to comprehensively consider factors such as trust and queue length to select the optimal notary for transaction processing to maximize its own transaction efficiency and security. Step 4: Through the Bayesian reasoning and strategy adjustment model, the receiver derives the current posterior belief based on the prior belief and judges the transaction strategy it chooses based on the posterior belief. The receiver has a certain prior knowledge of the notary at the beginning. As the transaction process progresses, the receiver can monitor the transaction information and other related information sent by the notary and use Bayesian reasoning to update the trust in the notary. Then, based on the updated belief and its own needs, it makes a decision between accepting or rejecting the transaction.

[0039] Step 5: The notary ensures that the transaction is completed, the relevant blockchains are modified accordingly, and sends a transaction confirmation notification to the sender and receiver, waiting for the transaction confirmation. Based on the principles of blockchain technology and the consistency algorithm, the notary ensures that the transaction is modified accordingly on the relevant blockchains. In cross-chain transactions, the notary needs to coordinate the interaction between different blockchains to ensure that the transaction can be correctly confirmed and recorded on both the source blockchain and the target blockchain to ensure the atomicity and consistency of the transaction. Finally, a message is sent to the transaction user to confirm the transaction. The transaction user verifies the transaction through the simple payment verification (SPV) method and confirms the transaction only after the correct record is obtained.

[0040] Step 6: Simulate a cross-chain environment where multiple notaries handle transactions. Set the notary to only handle one transaction at a time. At the same time, control the total number of transactions initiated by transaction users in the system to remain within the set range, and simulate the cross-chain transaction scenario of the notary in actual situations. Setting the notary to only handle one transaction at a time is based on the theory of resource allocation and task scheduling. As a resource in the system, the notary has limited processing power and can only handle one transaction at a time. This is similar to the single processor scheduling strategy in the operating system, which ensures the consistency and reliability of the notary when handling transactions, and is also in line with actual application scenarios. In addition, in this cross-chain transaction scenario, the number of users who initiate transactions at the same time is limited, which involves the theory of system stability. It can test whether the system remains stable under different loads and better simulate the actual environment.

[0041] Figure 2 The basic framework of the dynamic trust decision notarization system based on Bayesian reasoning is demonstrated. The framework adopts a multi-level decision tree, which consists of five levels: the regulatory agency decision layer, the notary decision layer, the sender decision layer, the receiver decision layer, and the game decision theory layer. Among them, the sender decision layer, the notary decision layer, and the receiver decision layer together constitute the decision-making process of each transaction. The regulatory agency decision layer is responsible for the macro-control system decision, and the game decision theory layer is responsible for providing a theoretical basis for each decision layer.

[0042] The regulatory agency decision-making layer, the regulatory agency completes the initialization operation before the cross-chain, through a priori belief and data collected by regulators Conduct regulatory strategy analysis to continuously update strategies and ensure the balance and stability of the system.

[0043] Notary decision-making layer. In the initial stage, the notary needs to prepare to start cross-chain transaction processing, wait for transactions to be received in the system, and observe the situation of other participants and their own situation (information observed by the notary). ), based on prior beliefs Continuously update your own trading strategy selection, and then execute the updated trading strategy in transaction processing.

[0044] The sender decision-making layer, the sender has transactions to be processed in the initial stage, enters the system, initiates cross-chain transactions, obtains the information currently observed by the notary and the regulatory agency (the information observed by the user) ), combined with one's own prior beliefs , calculate the notary selection priority of the current information, and then generate standardized cross-chain transaction information and pass it to the designated notary, waiting for subsequent related operations.

[0045] The receiver decision-making layer, based on the transaction learned from the sender in advance, waits in the system to receive the transaction, and monitors the transaction processing process to obtain the information observed by the user , and then based on one's own prior beliefs Calculate the probability of the selected notary type and the probability of the selected transaction strategy, and get the decision result to determine whether to trust the transaction and complete the subsequent transaction processing. Otherwise, apply to cancel the transaction processing.

[0046] At the game decision-making theory level, all participants are in a cross-chain environment and have their own prior beliefs about the cross-chain process from the beginning. , including: basic information of cross-chain transactions , strategies of all participants , Information disclosed by notaries , and then obtain the incomplete information observed in the system during the cross-chain process. When making game decisions, each participant updates information and strategies. Participants obtain the latest incomplete information and follow their own optimal trading strategies. Based on the observed information, Bayesian reasoning updates the posterior beliefs to select the current trading strategy.

[0047] Figure 3 The chart shows the changing trend of the reputation value of each type of notary under 100 transactions. In the simulation of 100 transactions, there are 3 honest notaries, 1 lazy notary, and 1 malicious notary in the system to process transactions. Users choose the best transaction notary they think to conduct cross-chain transactions, and each notary also chooses his own strategy to achieve the desired results. Figure 3 The figure shows the changing trend of the reputation value of one notary of each type, reflecting the reputation value of the entire notary group. The reputation value of the notary in the system is public data information. All participants can use the reputation value as one of the credentials to judge the current notary's behavior and make corresponding strategies. The notary needs to take into account the impact of his own strategy on the reputation value to decide the strategy.

[0048] Figure 4The probability distribution of malicious notaries’ trading strategies under 100 transactions is shown. In the simulation of 100 transactions, there are 3 honest notaries, 1 lazy notary, and 1 malicious notary in the system to process transactions. Users choose the best notary to conduct cross-chain transactions, and each notary also chooses his own strategy to achieve the desired result. There are two trading strategies, namely, evil work and disguise work , the malicious notary will choose a trading strategy in the transaction. By constructing a transaction decision tree, the expected results of two trading strategies are obtained to determine the probability of choosing the corresponding trading strategy. It will greatly affect the user's strategy choice, resulting in unsatisfied expected results, and tend to choose to pretend to work , and as the number of transactions increases, choose to disguise work The expected results are getting better and better, while the evil work This process will be disrupted, and ultimately the malicious notary's strategy will choose to develop in the direction the system expects.

[0049] The game process is the core process of this system. The decision-making level of the regulatory agency, the sender, the notary, and the receiver will all conduct Bayesian reasoning (incomplete information game) based on the information they can observe.

[0050] In the decision-making level of regulatory agencies, regulatory agencies update regulatory strategies based on reinforcement learning and dynamic adjustment theory. As an institution that prevents illegal behavior in cross-chain transactions and ensures transaction security and system stability, regulatory agencies can reduce the motivation of lawbreakers to commit evil by increasing the cost of evil and the intensity of punishment, and can dynamically adjust regulatory strategies to avoid existing risks. In the process of implementing regulatory transactions, regulatory strategies are adjusted to prevent certain illegal behaviors from escaping supervision, summarize transaction conditions in the system, ensure that the transaction success rate meets normal levels and reaches a stable or rising trend, and introduce an inertia coefficient. It reflects the inertia or stability of the regulatory agency's strategy selection, that is, the degree of dependence on historical strategy selection. The regulatory strategy is adjusted through the above behavior so that the system trading situation reaches a stable upward state. The regulatory strategy update method is as follows:

[0051] (1);

[0052] (2);

[0053] in, represents one of the supervision strategies, Regulatory strategies The new and old probabilities of is the inertia coefficient, For regulatory strategies The number of violations detected under is the total number of violations, represents the number of notaries, Indicates A collection of successful transaction records of notaries, It is an indicator function, which is used to check the number of violations in the notary's transaction records under the current regulatory policy and determine whether the transaction records are suspected of violations.

[0054] In the notary decision-making level, the notary selects the trading strategy based on the reputation model theory. The reputation of the notary is often an important indicator in the market, which can well reflect the overall behavior of the notary. By considering the notary's past reputation value and recent transaction behavior to update the reputation value , which reflects the combined impact of the notary’s long-term behavior accumulation and short-term performance. , time coefficient Weight coefficient with past reputation value Together, they have a decaying effect on past reputation values, which is consistent with the phenomenon that over time, the impact of past events on current reputation gradually weakens. Reputation values ​​are affected by many factors (data accuracy , Timely response , Data Integrity , Normalized transaction time ) have different weights respectively. , in order to reflect the real notary behavior, the calculation formula is:

[0055] (3);

[0056] in, Data accuracy , Timely response , Data Integrity , Normalized transaction time The weight of , e is a natural constant.

[0057] Reputation will guide the behavior strategies of other participants. Based on this, we can predict the behavior of other participants and analyze and calculate different trading strategies. Possible expectations , adjust your decision sensitivity Ultimately, the expected outcome motivates notaries to maintain good long-term behavior, because long-term accumulated high reputation can bring more business and rewards. This strategic decision-making process can be expressed by the following formula:

[0058] (4);

[0059] (5);

[0060] in, Choose for lazy notaries The probability of a trading strategy, Choose for lazy notaries Probability of trading strategies, perfunctory work , improve work They are the two trading strategies of the notary. They are Trading strategies and Expectations of trading strategies.

[0061] In the decision-making layer of the sender, when calculating the notary selection priority, first based on the expected utility theory, through the observed recent transaction status and reputation of the notary, plus the proportion of notary types and all transaction strategies of the notary, the decision is made based on the current cross-chain environment state, trying to achieve the expected result, and combining Bayesian reasoning to calculate the trust probability of each notary in the cross-chain ; Then observe the number of waiting transactions for each notary across the current cross-chain to obtain the idleness index of the queue , as a consideration of the actual transaction waiting time to meet the transaction; finally add comprehensive adjustment items , set different weights to adjust the proportion of the three major influencing factors, and achieve Calculate the priority of each notary to meet different transaction requirements. The function expression is:

[0062] (6);

[0063] in, For notary public The priority of The trust probability , Queue idleness indicator , Comprehensive adjustment items The weight of .

[0064] In the decision-making layer of the receiver, the receiver chooses whether to accept the transaction based on the utility maximization theory and Bayesian reasoning. The receiver's decision takes into account the expectations that each type of notary may bring under the current situation to determine the strategy choice. First, Bayesian reasoning is used to obtain the posterior probability of the type of honest notary in the current environment. , the probability of a lazy notary under the current situation , the probability of a malicious notary under the current situation Then, considering the transaction cost, the impact of refusing a transaction, the result of a successful transaction, etc., the expected evaluation under different decisions is calculated; finally, based on the rational decision of the user to pursue his own expected results, the expectation is converted into a trust probability in the form of an exponential function to update the trust probability of the notary, and the decision-making behavior pattern of the recipient under uncertainty is obtained. The expression is:

[0065] (7);

[0066] (8);

[0067] in, Represents the properties of transaction records (such as accuracy, completeness, etc.), is the number of transaction records, A notary public Properties The weight of It is Transaction records in attributes The value of It’s a strategy Lower Attributes expected value. is the set of all notary types, As a notary public of integrity, For the lazy notary, For malicious notaries, Represents a notary public in good faith in the current situation score, plus the average reputation score , divided by the sum of the scores of each type of notary in the current situation , and get the probability of being a good notary in the current situation ,Finally, the probability of each type of notary in the current situation is normalized to obtain the final probability of each type of notary in the current situation.

[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic trust decision notarization method based on Bayesian reasoning, characterized in that: Here are the steps: Step 1: Build a gaming environment by designing the regulatory agency’s regulatory strategy and reward and punishment behaviors, and use the regulatory strategy to take targeted precautions against notaries’ different types of illegal trading strategy choices, and guide notaries and trading users to update their own trading strategies to meet system requirements through Bayesian reasoning; Step 2: Set the initial notary types and type ratios. Each type of notary has a corresponding trading strategy. The notary derives the current posterior beliefs based on the prior beliefs to choose its own trading strategy. Step 3: The sender derives the current posterior belief and the queue length of each notary based on the prior belief to select which notary to process the transaction; Step 4: Through the Bayesian reasoning and strategy adjustment model, the receiver derives the current posterior belief based on the prior belief and judges the trading strategy of his own choice based on the posterior belief; Step 5: The notary ensures that the transaction is complete, the relevant blockchains are modified accordingly, and sends a transaction confirmation notification to the sender and receiver, waiting for the transaction confirmation; Step 6: Simulate a cross-chain environment where multiple notaries process transactions. Set a notary to only process one transaction at a time. At the same time, control the total number of transactions initiated by transaction users in the system to remain within the set range, simulating the cross-chain transaction scenario of notaries in actual situations.

2. The dynamic trust decision notarization method based on Bayesian reasoning according to claim 1 is characterized in that: In step one, guiding notaries and transaction users to update their own transaction strategies through Bayesian reasoning in accordance with system requirements means that notaries and transaction users judge the probability of each regulatory strategy through current information, and based on the probability distribution of regulatory strategies, use Bayesian reasoning to update the probability of their own transaction strategies being discovered under each type of regulatory strategy, thereby selecting the optimal transaction strategy; and as the transaction situation is updated, the regulatory agency will update the regulatory strategy based on the violations detected to ensure the stability of the cross-chain environment.

3. The dynamic trust decision notarization method based on Bayesian reasoning according to claim 1 is characterized in that: Step two specifically means: the notary will have a certain prior knowledge of the market environment, transaction user behavior and regulatory strategy in the initial stage; in the process of observing transactions, by observing the specific circumstances of the transaction, use Bayesian reasoning to update his own beliefs, and then choose the trading strategy that best suits his expectations based on the updated posterior beliefs; when choosing a trading strategy, he models his own expectations by building a strategy selection model in game theory. The notary needs to consider the behavior and trading strategies of other participants, as well as his own expectations, and determine the optimal trading strategy by analyzing the expectation matrix under different strategy combinations.

4. The dynamic trust decision notarization method based on Bayesian reasoning according to claim 1 is characterized in that: Step three is as follows: the sender first forms a priori trust probability distribution for different notaries based on previous understanding of notaries and market conditions. Then, in each transaction, taking into account the currently observed information, the trust in each notary is updated through Bayesian reasoning. Finally, the decision tree method is used to comprehensively consider trust and queue length to select the optimal notary for transaction processing.

5. The dynamic trust decision notarization method based on Bayesian reasoning according to claim 1 is characterized in that: Step five is as follows: Based on the principles of blockchain technology and the consistency algorithm, the notary ensures that the transaction is modified accordingly on the relevant blockchain; in cross-chain transactions, the notary coordinates the interaction between different blockchains to ensure that the transaction can be correctly confirmed and recorded on both the source blockchain and the target blockchain, and finally sends a message to the transaction user to confirm the transaction. The transaction user verifies the transaction and confirms the transaction only after obtaining the correct record.

6. A dynamic trust decision notarization system based on Bayesian reasoning, characterized in that: It is composed of the management agency decision-making layer, the notary decision-making layer, the sender decision-making layer, the receiver decision-making layer, and the game decision theory layer; The regulatory agency decision-making layer: The regulatory agency completes the initialization operation before the cross-chain, and continuously updates the regulatory strategy based on the prior beliefs and the overall information collected by the system; Notary decision-making layer: In the initial stage, the notary needs to prepare to start cross-chain transaction processing, wait for transactions to be received in the system, and observe the situations of other participants and their own situations, constantly update their own transaction strategy selection according to prior beliefs, and then execute the updated transaction strategy in transaction processing; The sender decision-making layer: In the initial stage, the sender has transactions that need to be processed. It enters the system, initiates cross-chain transactions, obtains the information of the notary and regulatory agency currently observed, and calculates the notary selection priority of the current information based on its own prior beliefs. Then, it generates standardized cross-chain transaction information and passes it to the designated notary, waiting for subsequent related operations. The decision-making layer of the receiver: The receiver waits for transactions in the system based on the transactions learned from the sender in advance, monitors the transaction processing process, calculates the probability of selecting the notary type and the probability of the selected transaction strategy based on its own prior beliefs, and determines whether to agree to the transaction based on the decision result, and then proceeds with the subsequent transaction processing; At the game decision theory level, all participants are in a cross-chain environment and can only obtain incomplete information in the system. Each participant follows his or her own optimal trading strategy based on the incomplete information he or she obtains, and uses Bayesian reasoning to update posterior beliefs based on the observed information to select the current trading strategy.

7. The dynamic trust decision notarization system based on Bayesian reasoning according to claim 6 is characterized in that: The supervision strategy update method is as follows: （1）； （2）； in, represents one of the supervision strategies, Regulatory strategies The new and old probabilities of is the inertia coefficient, For regulatory strategies The number of violations detected under is the total number of violations, represents the number of notaries, Indicates A collection of successful transaction records of notaries, It is an indicator function, which is used to check the number of violations in the notary's transaction records under the current regulatory policy and determine whether the transaction records are suspected of violations.

8. The dynamic trust decision notarization system based on Bayesian reasoning according to claim 6 is characterized in that: In the notary decision-making layer, the notary selects the transaction strategy based on the reputation model theory, updates the reputation value by considering the notary's past reputation value and recent transaction behavior, and introduces the weights of data accuracy, response timeliness, data integrity, and normalized transaction time in the reputation value calculation; The reputation value guides the behavioral strategies executed by other participants. The notary analyzes and calculates the expectations brought about by different transaction strategies and adjusts its own decision-making sensitivity.

9. The dynamic trust decision notarization system based on Bayesian reasoning according to claim 6, characterized in that: In the decision-making layer of the sender, when calculating the notary selection priority, first based on the expected utility theory, through the observed recent transaction status and reputation of the notary, plus the proportion of notary types and all transaction strategies of the notary, the decision is made based on the current cross-chain environment state, trying to achieve the expected result, and combining Bayesian reasoning to calculate the trust probability of each notary in the cross-chain; Then, the number of waiting transactions for each notary across the current cross-chain is observed to derive the queue's idleness index, which takes into account the actual situation of the transaction waiting time to meet the transaction; finally, a comprehensive adjustment item is added, and the trust probability, the queue's idleness index, and the weight of the comprehensive adjustment item are set to calculate the priority of each notary.

10. The dynamic trust decision notarization system based on Bayesian reasoning according to claim 6, characterized in that: In the decision-making layer of the receiver, first, Bayesian reasoning is used to obtain the posterior probability of the type of honest notary in the current environment, the probability of lazy notary in the current situation, and the probability of malicious notary in the current situation; then, considering the transaction cost, the impact of refusing to trade, and the result of successful transaction, the expected evaluation under different decisions is calculated; finally, the expectation is converted into trust probability to update the trust probability of the notary, and the decision-making behavior pattern of the receiver under uncertainty is obtained.

Images