Business resource sharing method and business resource sharing system based on cloud computing

Abstract

The embodiments of the present application relate to the field of computer technology, and specifically provide a business resource sharing method and a business resource sharing system based on cloud computing. The embodiments of the present application have many advantages by obtaining various information and performing targeted processing. First, the cloud service interaction information, resource demand information and authority statistics of the X group are integrated, covering the behavior, demand and authority of the cloud server under various resource sharing states, and comprehensively and meticulously describing the cloud service status. This information is input into the debugged target resource sharing processing network to obtain the current collaboration weight, which can quantify the confidence score of the cloud server group in business collaboration and scientifically evaluate the collaboration potential of the cloud server. Finally, the cloud server group to be connected is determined based on the set sharing discrimination index to ensure that only qualified cloud servers participate in the business resource sharing thread, optimize resource allocation, improve the rationality, security and effectiveness of resource sharing, and ensure smooth and efficient business collaboration.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and in particular to a business resource sharing method and a business resource sharing system based on cloud computing. Background Art

[0002] In the cloud computing environment, with the widespread application of cloud services, business resource sharing faces many challenges. Traditional resource sharing methods often lack consideration of comprehensive information about cloud servers, and cannot accurately determine which cloud servers are suitable for participating in specific business resource sharing threads. For example, only considering the hardware resources of cloud servers while ignoring their business behaviors, structured data usage requirements, and permissions under different resource sharing states, leads to unreasonable resource allocation, which may cause resources to be idle or overloaded, and reduce resource utilization. At the same time, when determining whether cloud servers participate in business collaboration, there is a lack of a quantitative confidence assessment mechanism, making it difficult to accurately judge the reliability of cloud server groups in business collaboration, thereby increasing business risks and affecting the smooth progress of business collaboration. Summary of the invention

[0003] The embodiments of the present application at least provide a business resource sharing method and a business resource sharing system based on cloud computing.

[0004] The embodiment of the present application provides a business resource sharing method based on cloud computing, which is applied to a business resource sharing system. The method includes: obtaining X groups of cloud service interaction information, resource demand information and authority statistics information, wherein the X groups of cloud service interaction information include Y business behavior descriptions of each cloud server in X cloud servers, the X cloud servers are cloud servers that apply for a round of business resource sharing threads in a target resource sharing state in an open structured data network, the Y business behavior descriptions of each cloud server include operation behavior codes of each cloud server in multiple resource sharing states, the multiple resource sharing states are resource sharing states that have completed authentication in the structured data network, the resource demand information includes structured data usage requirements of each group of cloud servers in Z groups of cloud servers in the multiple resource sharing states, and the authority statistics information includes The target resource sharing state includes differentiated permission labels, the Z group of cloud servers are multiple groups of cloud servers obtained by clustering the X cloud servers according to the current induction strategy, and X, Y and Z are all positive integers not less than 2; the X group of cloud service interaction information, resource demand information and permission statistical information are input into the debugged target resource sharing processing network to obtain the current collaboration weight output by the target resource sharing processing network, wherein the current collaboration weight is used to characterize the confidence score of the Z group of cloud servers completing the business collaboration in the round of business resource sharing thread on the basis of determining the Z group of cloud servers as the groups of cloud servers connected to the round of business resource sharing thread; on the basis that the current collaboration weight meets the set sharing discrimination index, the Z group of cloud servers are determined as the groups of cloud servers connected to the round of business resource sharing thread.

[0005] An embodiment of the present application also provides a business resource sharing system, including a processor and a memory; the processor and the memory are communicatively connected, and the processor is used to read a computer program from the memory and execute it to implement the above method.

[0006] The embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored, and the computer program implements the above method when running.

[0007] The technical solution provided by the embodiment of the embodiment of the present application may include the following beneficial effects: by obtaining multi-faceted information and performing targeted processing, it has many advantages. First, the cloud service interaction information, resource demand information and permission statistics of group X are integrated to cover the behavior, demand and permission of cloud servers under various resource sharing states, and comprehensively and meticulously describe the cloud service status. This information is input into the debugged target resource sharing processing network to obtain the current collaboration weight, which can quantify the confidence score of the cloud server group in business collaboration and scientifically evaluate the collaboration potential of the cloud server. Finally, the cloud server group to be connected is determined based on the set sharing discrimination index to ensure that only qualified cloud servers participate in the business resource sharing thread, optimize resource allocation, improve the rationality, security and effectiveness of resource sharing, and ensure smooth and efficient business collaboration.

[0008] For a description of the effects of the above-mentioned business resource sharing system and computer-readable storage medium, please refer to the description of the above-mentioned method.

[0009] In order to make the above-mentioned purposes, features and advantages of the embodiments of the present application more obvious and easy to understand, the following specifically cites preferred embodiments and describes them in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments, which are incorporated into the specification and constitute a part of the specification, and these drawings illustrate embodiments that conform to the embodiments of the present application, and are used together with the specification to illustrate the technical solutions of the embodiments of the present application. It should be understood that the following drawings only illustrate certain embodiments of the embodiments of the present application, and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can also be obtained based on these drawings without paying creative work.

[0011] Figure 1 It is a block diagram of a business resource sharing system shown in an embodiment of the present application.

[0012] Figure 2 It is a flow chart of a business resource sharing method based on cloud computing shown in an embodiment of the present application. DETAILED DESCRIPTION

[0013] Figure 1The structural diagram of the business resource sharing system 10 provided in the embodiment of the present application includes a processor 102, a memory 104, and a bus 106. Among them, the memory 104 is used to store execution instructions, including a memory and an external memory. The memory can also be understood as an internal memory, which is used to temporarily store the operation data in the processor 102 and the data exchanged with the external memory such as the hard disk. The processor 102 exchanges data with the external memory through the memory. When the business resource sharing system 10 is running, the processor 102 communicates with the memory 104 through the bus 106, so that the processor 102 executes the cloud computing-based business resource sharing method of the embodiment of the present application.

[0014] Please combine Figure 2 , Figure 2 It is a flow chart of a business resource sharing method based on cloud computing provided in an embodiment of the present application, which is applied to a business resource sharing system. The method may exemplarily include the following steps 110 to 130.

[0015] Step 110: Obtain X groups of cloud service interaction information, resource demand information, and authority statistics information.

[0016] In an embodiment of the present application, the X group of cloud service interaction information includes Y business behavior descriptions of each cloud server in the X cloud servers, the X cloud servers are cloud servers that apply for a round of business resource sharing threads in a target resource sharing state in an open structured data network, the Y business behavior descriptions of each cloud server include operation behavior codes of each cloud server in multiple resource sharing states, the multiple resource sharing states are resource sharing states that have completed authentication in the structured data network, the resource demand information includes structured data usage requirements of each group of cloud servers in the Z group of cloud servers in the multiple resource sharing states, the permission statistical information includes differentiated permission labels of the target resource sharing state, the Z group of cloud servers are multiple groups of cloud servers obtained by clustering the X cloud servers according to the current induction strategy, and X, Y and Z are all positive integers not less than 2.

[0017] First, in step 110, the business resource sharing system needs to obtain X groups of cloud service interaction information, resource demand information and authority statistics information. The cloud service interaction information here has a specific structure, wherein the X groups of cloud service interaction information contain Y business behavior descriptions of each cloud server in the X cloud servers. Taking a specific numerical example, for example, X=5, this means that there are 5 groups of cloud service interaction information, and for each cloud server, for example, Y=3, that is, each cloud server has 3 business behavior descriptions. These cloud servers are cloud servers in a round of business resource sharing threads in the target resource sharing state in the application open structured data network. The business behavior description of each cloud server contains the operation behavior coding of each cloud server in multiple resource sharing states, and these resource sharing states are resource sharing states that have completed authentication in the structured data network.

[0018] From the perspective of operation behavior coding, a system based on digital coding can be envisioned. For example, an operation behavior coded as 1 may represent that the cloud server is performing a data reading operation, a coded as 2 represents a data writing operation, a coded as 3 represents a data transmission operation, etc. This coding system helps to accurately describe the behavior of the cloud server under different resource sharing states.

[0019] Resource demand information includes the structured data usage requirements of each group of cloud servers in the Z group of cloud servers under multiple resource sharing states. For example, Z=4, that is, there are 4 groups of cloud servers that are obtained by clustering X cloud servers according to the current induction strategy. The structured data usage requirements can be measured by indicators such as data volume, data type, and data access frequency. For example, a group of cloud servers may need 100GB of structured data under a specific resource sharing state, the data type is customer order data in a relational database, and the access frequency is 10 times per hour.

[0020] Permission statistics include differentiated permission labels for the target resource sharing state. These permission labels can be represented by a permission matrix. For example, a 3×3 permission matrix, where rows represent different cloud server groups and columns represent different operations (such as read, write, delete), can have either 0 or 1, with 0 indicating no permission and 1 indicating permission. This allows for a clear definition of the permissions for each cloud server group in the target resource sharing state.

[0021] Step 120: input the X group of cloud service interaction information, resource demand information and authority statistics information into the debugged target resource sharing processing network to obtain the current coordination weight output by the target resource sharing processing network.

[0022] In an embodiment of the present application, the current collaboration weight is used to characterize the confidence score of the Z group of cloud servers completing business collaboration in the round of business resource sharing threads based on determining the Z group of cloud servers as the groups of cloud servers connected to the round of business resource sharing threads.

[0023] Next, in step 120, the business resource sharing system inputs the obtained X groups of cloud service interaction information, resource demand information and permission statistics information into the debugged target resource sharing processing network. The target resource sharing processing network can be a neural network-based model, such as a multi-layer perceptron (MLP).

[0024] When building this MLP model, the number of neurons in the input layer can be set to correspond to the dimension of the input information. For example, if the cloud service interaction information is encoded and quantized into a 100-dimensional vector, the resource demand information is a 50-dimensional vector, and the permission statistics information is a 30-dimensional vector, then the number of neurons in the input layer can be set to 180. The hidden layer can be set to 2 layers, with 100 neurons in the first hidden layer and 50 neurons in the second hidden layer, using ReLU as the activation function. The output layer has only one neuron, which is used to output the current collaborative weight.

[0025] When the information is input into the target resource sharing processing network, the network will perform a series of complex calculations based on its internal weights and activation functions. For example, in the calculation from the input layer to the first hidden layer, for the output of each neuron, the calculation formula is: h_1=ReLU(sum_{i=1}^{180}w_{i1}x_i+b_1), where w_{i1} is the weight from the input layer to the first hidden layer, x_i is the element of the input vector, and b_1 is the bias term. Through such layer-by-layer calculations, the current collaborative weight of the output of the target resource sharing processing network is finally obtained. This current collaborative weight is used to characterize the confidence score of the Z group of cloud servers in completing business collaboration in a round of business resource sharing threads, based on the determination of the Z group of cloud servers as the groups of cloud servers connected to a round of business resource sharing threads. For example, the current collaborative weight is 0.8, indicating that there is a high degree of confidence that the Z group of cloud servers can complete the task well in business collaboration.

[0026] Step 130 , on the basis that the current collaboration weight meets the set sharing judgment index, the Z groups of cloud servers are determined as the groups of cloud servers connected to the round of business resource sharing thread.

[0027] Finally, in step 130, the business resource sharing system will judge the current collaboration weight. The sharing judgment index can be set to a specific value, such as 0.7. If the current collaboration weight meets this set sharing judgment index, for example, the current collaboration weight is 0.8 greater than 0.7, then the Z group of cloud servers will be determined as the groups of cloud servers connected to a round of business resource sharing threads. This means that these cloud servers meet the system requirements in terms of resource sharing and can participate in business collaboration in this round of business resource sharing threads.

[0028] Overall, the implementation process of this application obtains detailed cloud service-related information, uses a complex target resource sharing processing network to calculate the coordination weight, and determines the appropriate cloud server group to participate in the business resource sharing thread based on the set discrimination index, thus realizing a scientific and rigorous business resource sharing decision-making mechanism. This mechanism fully considers the behavior, resource requirements and permissions of cloud servers in different resource sharing states, and can effectively improve the efficiency and security of business resource sharing.

[0029] In practical applications, this mechanism can be widely used in cloud computing environments of large enterprises or cross-enterprise cloud service cooperation scenarios. For example, in a large multinational enterprise, there are multiple data centers distributed in different regions, which can be regarded as different cloud servers. Through this business resource sharing mechanism, it is possible to reasonably arrange them to participate in specific business resource sharing threads according to the business behavior, resource requirements and authority of each data center, thereby improving the data processing efficiency and resource utilization of the entire enterprise. At the same time, in cross-enterprise cloud service cooperation, cloud servers of different enterprises can make resource sharing decisions according to this mechanism, and achieve mutually beneficial and win-win business collaboration on the premise of ensuring their respective authority and meeting their own resource requirements.

[0030] From the perspective of the depth and breadth of technology, the operational behavior coding system in cloud service interaction information can be continuously expanded and refined. For example, in addition to basic data operation coding, it is also possible to add coding for data encryption and decryption operations, and coding for data backup and recovery operations. This will enable cloud service interaction information to more comprehensively reflect the actual operation of cloud servers. For resource demand information, in addition to the above-mentioned indicators such as data volume, type, and access frequency, it is also possible to consider introducing data timeliness requirements. For example, some businesses have very high demands for real-time data and need to obtain the latest structured data within a few milliseconds, which needs to be reflected in the resource demand information.

[0031] In terms of the target resource sharing processing network, in addition to using the MLP model, other more advanced neural network models can also be explored, such as variants of convolutional neural networks (CNN) or recurrent neural networks (RNN). If CNN is used, the input information can be regarded as a two-dimensional matrix, and features are extracted by sliding the convolution kernel on the matrix. This method may be more suitable for processing cloud service information with spatial structure. If RNN variants are used, such as long short-term memory networks (LSTM), the temporal relationship in cloud service interaction information can be better processed, such as changes in the business behavior of cloud servers at different time points.

[0032] In terms of permission statistics, as the business of an enterprise develops and security requirements increase, permission labels can be further refined. In addition to operation permissions, they can also include scope permissions for data access, such as certain cloud server groups can only access customer data in a specific region, or can only access customer data of a specific age group, etc. This refined permission label will enable permission statistics to more accurately reflect the permission restrictions of cloud server groups in resource sharing.

[0033] During the implementation of the embodiments of the present application, various links are interrelated and influence each other. The accuracy and completeness of cloud service interaction information, resource demand information, and permission statistical information directly affect the reliability of the current collaborative weight output by the target resource sharing processing network. Whether the current collaborative weight meets the set sharing judgment index determines which cloud server groups can access the business resource sharing thread. This close logical relationship constitutes a complete business resource sharing decision-making system, which provides an innovative and effective solution for resource sharing in a cloud service environment.

[0034] In terms of data processing, after obtaining cloud service interaction information, resource demand information and permission statistics, data cleaning and preprocessing are required. For example, there may be some incorrect coding or missing coding for the operation behavior coding in the cloud service interaction information. These errors can be discovered and corrected through data verification algorithms. For indicators such as data volume and type in resource demand information, if there are inconsistent data or inconsistent units, standardization processing is required. For example, convert the data volume into bytes, unify the date format into a standard format, etc. These data processing steps help improve the quality of data input into the target resource sharing processing network, thereby improving the performance of the entire system.

[0035] In the training process of the target resource sharing processing network, a large amount of sample data is required. These sample data can be obtained by simulating various business scenarios in the cloud service environment. For example, simulating the behavior of cloud servers under different business loads, different resource demand distributions, and different permission settings. By labeling these sample data, for example, labeling the ideal value of the collaboration weight according to the actual business collaboration results, and then using the labeled sample data to train the target resource sharing processing network. During the training process, optimization algorithms such as stochastic gradient descent (SGD) can be used to adjust the weights of the network so that the output of the network is as close to the labeled ideal value as possible.

[0036] From a security perspective, the entire business resource sharing decision-making process needs to consider the confidentiality, integrity, and availability of data. When obtaining cloud service interaction information, resource demand information, and permission statistics, secure communication protocols such as SSL / TLS protocols need to be used to prevent data from being stolen or tampered with during transmission. In the target resource sharing processing network, input and output data also need to be encrypted, especially for permission statistics containing sensitive information. In addition, after determining that the cloud server group has access to the business resource sharing thread, a monitoring mechanism needs to be established to monitor the behavior of the cloud server group in the business collaboration process in real time to prevent unauthorized operations or malicious attacks, and ensure the security of business resource sharing.

[0037] In terms of scalability, as the scale of cloud services continues to expand, more cloud servers may be added to the resource sharing system. At this time, the business resource sharing system needs to be able to flexibly adapt to this change. For the acquisition and processing of cloud service interaction information, a distributed collection and aggregation method can be used to improve the efficiency of information acquisition. In terms of the target resource sharing processing network, more input information and more complex business scenarios can be adapted by increasing the number of layers or neurons in the network. For setting sharing discrimination indicators, dynamic adjustments can also be made according to actual conditions to ensure that the appropriate cloud server group can still be accurately screened to participate in the business resource sharing thread when the scale of cloud services expands.

[0038] In terms of resource optimization, this business resource sharing decision-making mechanism can realize the reasonable allocation and optimal utilization of cloud server resources. For example, based on the resource demand information of the cloud server group, resources can be preferentially allocated to those cloud server groups that have more urgent resource needs and have higher coordination weights in business collaboration. At the same time, by analyzing the operation behavior encoding of cloud servers under different resource sharing states, the bottleneck of resource utilization can be found. For example, some cloud servers occupy too much bandwidth resources during data transmission, which can be optimized in a targeted manner, such as using data compression technology or optimizing network topology, so as to improve the resource utilization efficiency of the entire cloud service environment.

[0039] Compared with the prior art, this business resource sharing mechanism has unique advantages. Traditional cloud service resource sharing methods may be based more on static configurations or simple rules to determine the participation of cloud servers. However, the embodiments of the present application comprehensively consider cloud service interaction information, resource demand information, and authority statistics, and use the target resource sharing processing network to perform dynamic collaborative weight calculations, which can more accurately adapt to the complex and changeable cloud service business environment. For example, in the traditional method, it may only be based on the hardware configuration of the cloud server and the pre-set resource allocation ratio to determine its participation in business resource sharing threads, while ignoring factors such as the business behavior of the cloud server during actual operation, changes in resource demand at different times, and dynamic adjustment of permissions. The embodiments of the present application comprehensively consider these factors, thereby being able to provide a more efficient, flexible, and secure business resource sharing solution.

[0040] In summary, this application, which takes the business resource sharing system as the execution body, provides a comprehensive, efficient, secure and scalable solution for business resource sharing in the cloud service environment through rigorous procedures and innovative technical means, which is of great significance in the development and application of cloud service technology.

[0041] In the above technical solution, cloud service interaction information (features), resource demand information (features) and permission statistics information (features) are very critical elements, which are further explained below.

[0042] The cloud service interaction information (features) includes Y business behavior descriptions of each of the X cloud servers, which cover the operation behavior coding of the cloud server in multiple authenticated resource sharing states. From a deeper perspective, this operation behavior coding is a quantitative representation of various operations of the cloud server in a shared environment. For example, in a large-scale cloud computing data center, the cloud server may perform a variety of operations, such as data storage, retrieval, processing, and interaction with other cloud servers or external systems. The operation behavior coding can adopt a multi-level coding system, such as using a three-digit code to represent different operations. For example, a data storage operation may be coded as 101, indicating that this is a write-type data operation and is an operation on a specific type of data storage structure (such as a table in a relational database); while a data retrieval operation may be coded as 202, indicating that this is a query operation and involves a specific index mechanism or query condition.

[0043] In order to more specifically represent the cloud service interaction information (features), an exemplary feature vector can be constructed. For example, there are 3 cloud servers (X=3), each cloud server has 5 business behavior descriptions (Y=5). In a specific resource sharing state, the exemplary feature vector corresponding to the operation behavior encoding of cloud server 1 can be [101, 202, 303, 105, 401], where 101 represents data storage operations, 202 represents data retrieval operations, 303 represents data processing operations (such as performing some mathematical calculations or conversions on data), 105 represents specific data interaction operations with external systems (which may involve specific protocols or interfaces, and 105 is used here to represent the type of operation), and 401 represents some data transmission operations within the system (such as transmitting data between different storage modules). The exemplary feature vector of cloud server 2 can be [203, 102, 301, 205, 501], and the exemplary feature vector of cloud server 3 can be [103, 302, 201, 403, 104]. These exemplary feature vectors can carefully reflect the behavior patterns of cloud servers in the resource sharing process.

[0044] At the same time, cloud service interaction information (characteristics) also reflects the dynamic behavior of cloud servers under different resource sharing states. For example, during peak resource sharing periods, cloud servers may perform more data processing and transmission operations, and the corresponding operational behavior encoding will reflect this trend. During off-peak periods, more operations related to data backup or system maintenance may be performed, and these operations will also be accurately encoded in the cloud service interaction information (characteristics). Moreover, this feature is not isolated, it is interrelated with other elements in the entire cloud service system, such as resource demand information (characteristics), because the operational behavior of cloud servers is often based on their own resource needs or to meet the resource needs of other cloud servers.

[0045] Resource demand information (features) include the structured data usage requirements of each group of cloud servers in group Z of cloud servers under multiple resource sharing states. This feature describes the cloud server's demand for structured data from multiple dimensions. From the perspective of data volume, different cloud server groups may require different amounts of structured data under different resource sharing states. For example, a cloud server group responsible for processing large-scale data analysis tasks may require several TB of structured data, which may include a large amount of user behavior records, transaction information, etc. On the other hand, a cloud server group that mainly performs user identity authentication may only require a relatively small amount (such as a few hundred MB) of structured data such as user registration information.

[0046] From the perspective of data types, resource demand information (features) also has rich connotations. There are many types of structured data, such as table structure data in relational databases (including different data types such as integers, strings, dates, etc.), structured data in XML format, etc. Different cloud server groups may have requirements for specific data types based on their business functions. For example, a cloud server group related to a financial system may mainly require financial statement data in a relational database. These data are stored in a specific table structure, including amount data of digital type, transaction date data of date type, etc.; while a cloud server group related to a content management system may pay more attention to content metadata in XML format, which describes the attributes, classification, and other information of the content.

[0047] In order to represent resource demand information (features), an exemplary feature vector is also constructed. For example, there are 4 groups of cloud servers (Z=4). For cloud server group 1, under a certain resource sharing state, the exemplary feature vector of its resource demand information (features) can be [5000, 1, 10, 0.5], where 5000 represents the data volume demand in MB (for example, 5GB here), 1 represents that the main demand is data in a relational database (1 can be set to represent a relational database, 2 represents XML format data, etc.), 10 represents that the data needs to be accessed 10 times a day (i.e., data access frequency), and 0.5 represents the timeliness requirement for data (0 represents that real-time data is not required, 1 represents that real-time data is required, and 0.5 represents that there are certain requirements for data timeliness but it is not completely real-time). The exemplary feature vector of cloud server group 2 can be [200, 2, 1, 0], the exemplary feature vector of cloud server group 3 can be [10000, 1, 100, 1], and the exemplary feature vector of cloud server group 4 can be [800, 1, 5, 0.3]. These resource demand information (characteristics) in different dimensions together constitute the demand profile of the cloud server in the resource sharing system, and interact with the cloud service interaction information (characteristics) and permission statistics information (characteristics).

[0048] Permission statistics (features) include differentiated permission labels for the target resource sharing status. This feature mainly defines and describes the permissions of cloud servers in resource sharing from the perspective of security and access control. Permission labels can be built based on a complex permission model, such as a role-permission model. In this model, different cloud server groups are assigned different roles, and each role corresponds to a different set of permissions. Taking the cloud service environment within an enterprise as an example, there are cloud server groups corresponding to the R&D department, the marketing department, and the finance department. The cloud server group of the R&D department may be given the permission to create, modify, and delete experimental data (structured data). These permission labels reflect their operational needs during the R&D process; the cloud server group of the marketing department may only have the permission to read and analyze market research data (structured data) for market trend analysis and marketing strategy formulation; while the cloud server group of the finance department has a high degree of permission to finance-related structured data, including the permission to read, write, and audit data, to ensure the security and accuracy of financial data.

[0049] Permission statistics (features) can also be refined in terms of data access scope and operation restrictions. For example, for some sensitive customer data (structured data), only a specific cloud server group (such as the cloud server group corresponding to the customer service department) has access rights when certain conditions are met (such as customer authorization or in a specific business scenario).

[0050] An exemplary feature vector can be used to represent the permission statistics (features). For example, there are also 4 groups of cloud servers (corresponding to the cloud server groups in the resource demand information (features) to reflect the mutual relationship). For cloud server group 1, the exemplary feature vector of its permission statistics (features) can be [1, 1, 0, 0]. Here, for example, the elements in the vector correspond to the permissions of create, read, modify, and delete operations (1 means permission, 0 means no permission). This vector indicates that cloud server group 1 has create and read permissions, but no modify and delete permissions. The exemplary feature vector of cloud server group 2 can be [0, 1, 0, 0], the exemplary feature vector of cloud server group 3 can be [1, 1, 1, 1], and the exemplary feature vector of cloud server group 4 can be [0, 0, 0, 0]. Moreover, the permission statistics (features) are not static. They will change dynamically with changes in business needs, adjustments to security policies, and the evolution of the cloud service environment. For example, when an enterprise launches a new business product line, it may reallocate the permissions of the relevant cloud server groups, and these changes will be reflected in the permission statistics (features) in a timely manner. In addition, there is a constraint relationship between the permission statistics information (features) and the cloud service interaction information (features) and resource demand information (features). The operation behavior of the cloud server (cloud service interaction information (features)) must be within the scope of its permissions (permission statistics information (features)). At the same time, the allocation of permissions will also affect the resource requirements of the cloud server (resource demand information (features)). For example, the restriction of permissions may result in the cloud server only being able to obtain part of the structured data, thereby affecting its demand for data volume and data type.

[0051] In some optional embodiments, the step of inputting the X groups of cloud service interaction information, resource demand information, and authority statistical information into a debugged target resource sharing processing network to obtain a current collaborative weight output by the target resource sharing processing network includes: mapping the X groups of cloud service interaction information into X cloud service interaction attribute knowledge, mapping the resource demand information into resource demand quantification knowledge, and mapping the authority statistical information into authority statistical distribution knowledge, wherein the X cloud service interaction attribute knowledge correspond to the X cloud servers respectively; performing knowledge collision on the X cloud service interaction attribute knowledge and the authority statistical distribution knowledge respectively to obtain X business interaction authority collision features, wherein the X business interaction authority collision features correspond to the X cloud servers respectively; and determining the current collaborative weight based on the X business interaction authority collision features, the resource demand quantification knowledge, and the authority statistical distribution knowledge.

[0052] In the next step, the X cloud service interaction attribute knowledge are respectively subjected to knowledge collision with the authority statistical distribution knowledge to obtain X business interaction authority collision features, including: performing knowledge collision on the u-th cloud service interaction attribute knowledge among the X cloud service interaction attribute knowledge and the authority statistical distribution knowledge in the following manner to obtain the u-th business interaction authority collision feature among the X business interaction authority collision features, wherein u is a positive integer not less than 1 and not greater than X: performing feature fusion on the u-th cloud service interaction attribute knowledge and the authority statistical distribution knowledge to obtain the u-th target fusion knowledge; determining a weighted result of the u-th cloud service interaction attribute knowledge and the u-th target fusion knowledge as the u-th business interaction authority collision feature; or, performing knowledge feature enhancement on the u-th target fusion knowledge using a set u-th feature enhancement coefficient to obtain the u-th target enhanced knowledge, and determining the integration result of the u-th cloud service interaction attribute knowledge and the u-th target enhanced knowledge as the u-th business interaction authority collision feature.

[0053] Furthermore, the determining the current collaborative weight based on the X business interaction permission collision characteristics, the resource demand quantification knowledge and the permission statistical distribution knowledge includes: splicing the X business interaction permission collision characteristics, the resource demand quantification knowledge and the permission statistical distribution knowledge into target splicing knowledge; performing attention processing on the target splicing knowledge to obtain attention splicing knowledge, wherein the target splicing knowledge and the attention splicing knowledge have the same knowledge feature recognition degree; performing full connection processing on the attention splicing knowledge to obtain fully connected attention knowledge, and using a set business collaboration analysis model to perform business collaboration operations on the fully connected attention knowledge to obtain the current collaborative weight.

[0054] In this optional embodiment, the entire technical solution revolves around how to input X groups of cloud service interaction information, resource demand information and authority statistics information into the target resource sharing processing network to obtain the current collaboration weight.

[0055] First, map X groups of cloud service interaction information to X cloud service interaction attribute knowledge. This mapping is a deep processing of cloud service interaction information. Take the operation behavior coding in the cloud service interaction information mentioned above as an example. If X=5, for each group of cloud service interaction information, the operation behavior coding, such as 101, 202, etc., under the three-digit coding system, is converted into cloud service interaction attribute knowledge through a specific mapping algorithm. For example, for the data storage operation represented by the operation behavior coding 101, when it is mapped to cloud service interaction attribute knowledge, an algorithm based on a lookup table is used to set the attribute knowledge corresponding to the storage operation in the lookup table to attribute A=0.8, attribute B=0.1, and attribute C=0.1 (here, attributes A, B, and C are quantitative representations of the characteristics related to the storage operation, such as the complexity of the operation, the importance of the data flow, etc.). In this way, the entire operation behavior coding is converted into cloud service interaction attribute knowledge.

[0056] At the same time, the resource demand information is mapped into quantitative knowledge of resource demand. Taking the numerical feature vector of resource demand characteristics as [5000, 1, 10, 0.5] as an example (here 5000 represents the data volume demand, 1 represents the data type as a relational database, 10 represents 10 data accesses per day, and 0.5 represents the data timeliness requirement), a weighted quantification algorithm is used. Set the data volume demand weight to 0.4, the data type weight to 0.2, the data access frequency weight to 0.3, and the data timeliness weight to 0.1, then the resource demand quantitative knowledge is calculated as:

[0057] 5000*0.4+1*0.2+10*0.3+0.5*0.1=2000+0.2+3+0.05=2003.25.

[0058] And the permission statistics information is mapped to permission statistics distribution knowledge. For example, for the permission statistics feature numerical feature vector [1, 1, 0, 0], a 4×4 permission distribution matrix is ​​constructed (corresponding to the four operation permissions of create, read, modify, and delete). In this matrix, for the cloud server group corresponding to the vector, the values ​​of the create permission position (first row and first column) and the read permission position (second row and second column) are 1, the values ​​of the modify permission position (third row and third column) and the delete permission position (fourth row and fourth column) are 0, and the other elements are 0, so as to convert the permission statistics information into permission statistics distribution knowledge.

[0059] The next step is to collide the X cloud service interaction attribute knowledge with the authority statistical distribution knowledge to obtain the X business interaction authority collision features. The u-th cloud service interaction attribute knowledge and the authority statistical distribution knowledge are collided to obtain the u-th business interaction authority collision feature. The specific method is as follows:

[0060] First, the u-th cloud service interaction attribute knowledge and the authority statistical distribution knowledge are feature fused to obtain the u-th target fusion knowledge. For example, the u-th cloud service interaction attribute knowledge is in the form of a vector [0.8, 0.6, 0.4] (the elements here correspond to different quantized values ​​of operation attributes), and the authority statistical distribution knowledge is the above-mentioned 4×4 matrix. The algorithm for feature fusion can be to multiply each element in the vector with the element related to the corresponding operation permission in the matrix, and then sum them. For example, the first element of the vector 0.8 is multiplied by the value 1 of the corresponding creation permission position in the matrix (for example, the first row and the first column), the second element of the vector 0.6 is multiplied by the value 1 of the corresponding read permission position in the matrix (for example, the second row and the second column), and the third element of the vector 0.4 is multiplied by the value 0 of the corresponding modification permission position in the matrix (for example, the third row and the third column), and then these product results are added: 0.8*1+0.6*1+0.4*0=1.4, and this result is used as an element in the u-th target fusion knowledge. The other elements of the vector and other related elements of the matrix are processed in the same way to obtain the complete fusion knowledge of the u-th target.

[0061] The second is to determine the weighted result of the u-th cloud service interaction attribute knowledge and the u-th target fusion knowledge as the u-th business interaction permission collision feature. For example, the weight of the u-th cloud service interaction attribute knowledge is set to 0.6, and the weight of the u-th target fusion knowledge is set to 0.4. For the u-th cloud service interaction attribute knowledge [0.8, 0.6, 0.4] and the u-th target fusion knowledge (for example, [1.4, 1.2, 0.8] after the above calculation), calculate the weighted result. For the first element:

[0062] 0.8*0.6+1.4*0.4=0.48+0.56=1.04. Calculate other elements in the same way to obtain the u-th business interaction permission collision feature.

[0063] Alternatively, the u-th target fusion knowledge is enhanced with the set u-th feature enhancement coefficient to obtain the u-th target enhancement knowledge, and the integration result of the u-th cloud service interaction attribute knowledge and the u-th target enhancement knowledge is determined as the u-th business interaction authority collision feature. For example, the u-th feature enhancement coefficient is set to 1.2, and for the u-th target fusion knowledge [1.4, 1.2, 0.8], each element is multiplied by 1.2 to obtain the u-th target enhancement knowledge [1.68, 1.44, 0.96]. Then the u-th cloud service interaction attribute knowledge [0.8, 0.6, 0.4] is integrated with the u-th target enhancement knowledge [1.68, 1.44, 0.96]. The integration algorithm can be to add the corresponding elements of the two vectors: [0.8+1.68, 0.6+1.44, 0.4+0.96]=[2.48, 2.04, 1.36], and obtain the u-th business interaction authority collision feature.

[0064] Finally, the current collaborative weight is determined based on X business interaction permission collision characteristics, resource demand quantitative knowledge, and permission statistical distribution knowledge. First, splice the X business interaction permission collision characteristics, resource demand quantitative knowledge, and permission statistical distribution knowledge into the target splicing knowledge. For example, X=3, the three business interaction permission collision characteristics are in vector form: [1.04, 0.8, 0.6], [1.2, 0.9, 0.7], [0.9, 0.7, 0.5], the resource demand quantitative knowledge is 2003.25, and the permission statistical distribution knowledge is the above 4×4 matrix. Splice these contents into a long vector or matrix form (depending on the specific splicing rules) in a certain order to obtain the target splicing knowledge.

[0065] Then, the target splicing knowledge is subjected to attention processing to obtain the attention splicing knowledge. The attention processing can be performed using an attention mechanism algorithm based on a neural network. For example, for each element in the target splicing knowledge, its attention weight is calculated through a small neural network. For example, the target splicing knowledge is a vector [1.04, 0.8, 0.6, 2003.25, 1, 1, 0, 0, ...] (the following elements are the elements in the authority statistical distribution knowledge matrix arranged in order), and the attention weight of each element is obtained through the query (Query), key (Key), and value (Value) calculation in the attention mechanism. For example, for the element 1.04, the calculated attention weight is 0.1. The attention weight of each element is obtained in the same way, and then each element is multiplied by its corresponding attention weight to obtain the attention splicing knowledge.

[0066] Then, the attention splicing knowledge is fully connected to obtain the fully connected attention knowledge. The fully connected processing can adopt the multi-layer perceptron (MLP) algorithm. For example, the attention splicing knowledge is in the form of a vector, and it is input into an MLP with 3 hidden layers. The first hidden layer has 10 neurons, the second hidden layer has 8 neurons, and the third hidden layer has 5 neurons. The activation function of each neuron can adopt the ReLU function. For example, for the element x in the input vector, the output of the first neuron in the first hidden layer is ReLU (w1*x+b1), where w1 is the weight connecting the input layer and the first neuron, and b1 is the bias term. Through such layer-by-layer calculation, the fully connected attention knowledge is obtained.

[0067] Finally, the set business collaboration analysis model is used to perform business collaboration operations on the fully connected attention knowledge to obtain the current collaboration weight. The business collaboration analysis model can be a model built based on a logistic regression algorithm. For example, the fully connected attention knowledge is a vector [0.5, 0.4, 0.3], which is input into the business collaboration analysis model. The formula of the logistic regression model is y=1 / (1+exp(-(w*x+b))), where x is the input vector, w is the weight vector, and b is the bias term. By adjusting w and b, the model outputs a value representing the current collaboration weight, such as 0.8, indicating a higher possibility of collaboration.

[0068] It can be seen that by mapping different types of information, knowledge collision, splicing, attention processing, full connection processing and business collaborative operation, etc., the current collaborative weight is determined. This complex and orderly process can comprehensively and deeply consider the various factors of cloud servers in resource sharing. The mapping operation converts the original information into a quantifiable and computable form of knowledge; knowledge collision accurately integrates cloud service interactions and authority relationships; splicing, attention processing, and full connection processing further mine data features; business collaborative operation derives the collaborative weight, which effectively improves the accuracy, rationality and scientificity of resource sharing decisions, avoids improper resource allocation and authority conflicts, and improves the performance of the entire business resource sharing system.

[0069] In actual application, step 110 can be divided into the following three sub-steps:

[0070] (1) The step of obtaining X sets of cloud service interaction information includes: obtaining resource operation event semantics, security risk detection semantics, and permission authentication update semantics of each cloud server in the X cloud servers to obtain the X sets of cloud service interaction information, wherein the resource operation event semantics of each cloud server includes a shared resource operation record of each cloud server in a previous v rounds of business resource sharing threads, where v is equal to 1 or v is a positive integer not less than 2, the permission authentication update semantics of each cloud server includes a permission authentication tag of each cloud server in each of the multiple resource sharing states within a previous set period, and the security risk detection semantics of each cloud server includes a current security risk detection state of each cloud server in each of the multiple resource sharing states;

[0071] (2) The step of obtaining resource demand information includes: obtaining a global security risk detection status of each group of cloud servers in the Z group of cloud servers in each resource sharing state in the multiple resource sharing states;

[0072] (3) The step of obtaining permission statistical information includes: obtaining a status keyword of the target resource sharing status, or obtaining a status keyword of the target resource sharing status and a service node for applying to open the round of service resource sharing thread.

[0073] In this embodiment, the starting point of the entire technical solution is step 110, which is divided into three sub-steps. Each sub-step and how they are integrated into the entire technical solution are described in detail below.

[0074] The first step is to obtain X sets of cloud service interaction information. This step requires obtaining the resource operation event semantics, security risk detection semantics, and permission authentication update semantics of each of the X cloud servers to obtain the X sets of cloud service interaction information.

[0075] From the perspective of resource operation event semantics, it covers the shared resource operation records of each cloud server in the previous v rounds of business resource sharing threads. For example, when v=3, for a certain cloud server, its shared resource operation records may include 10 data read operations and 5 data write operations in the first round of business resource sharing threads, 8 data reads and 3 data transfer operations in the second round, and 12 data processing operations in the third round. These operation records are stored in a structured manner, which may be an operation log table, in which each row records the timestamp of an operation, the type of operation (such as represented by the three-digit code mentioned above), the amount of data involved in the operation, and other information.

[0076] The permission authentication update semantics of each cloud server include the permission authentication tags of each cloud server in each resource sharing state in multiple resource sharing states within the previous set period. For example, the set period is one month. During this month, for a certain cloud server, the permission authentication tag in resource sharing state 1 is [1, 1, 0, 0] (corresponding to the four operation permissions of create, read, modify, and delete), indicating that there are create and read permissions, but no modify and delete permissions; in resource sharing state 2, the permission authentication tag becomes [1, 0, 1, 0]. This change reflects the dynamic adjustment of permissions under different resource sharing states. The permission authentication tag can be recorded by the permission management system. Every time the permission changes, the system will update the corresponding database table to record information such as the cloud server, resource sharing state, change time, and new permission authentication tag.

[0077] The security risk detection semantics of each cloud server includes the current security risk detection status of each cloud server in each resource sharing state in multiple resource sharing states. The security risk detection status can be represented by a quantitative indicator. For example, a value between 0 and 1 is used, where 0 represents no security risk and 1 represents the highest security risk. In a certain resource sharing state, the security risk detection status of the cloud server is 0.3, which may be due to the detection of some potential network connection anomalies or data access anomalies in the server in this state, but has not yet reached a serious level. Security risk detection can be based on a series of detection algorithms, such as intrusion detection algorithms, to determine the risk value by analyzing the deviation of network traffic patterns and data access patterns from normal patterns. By integrating these resource operation event semantics, security risk detection semantics, and permission authentication update semantics, we can obtain X sets of cloud service interaction information.

[0078] Next is the step of obtaining resource demand information. This step requires obtaining the global security risk detection status of each group of cloud servers in group Z of cloud servers in each resource sharing state in multiple resource sharing states. The global security risk detection status is an assessment of the overall security status of the cloud server group. For example, for a group of cloud servers, in resource sharing state 1, the global security risk detection status is 0.2, which is calculated by comprehensively considering the security risk detection status of each cloud server in the group (such as the risk value between 0 and 1 of a single cloud server mentioned above) and combining some group-level security factors (such as the trust relationship between servers in the group, the encryption degree of data sharing, etc.). The weighted average algorithm can be used to calculate the global security risk detection status. For example, there are 3 cloud servers in the group, and their security risk detection states are 0.1, 0.3, and 0.2 respectively, and the corresponding weights are 0.4, 0.3, and 0.3 respectively (the weights can be determined based on factors such as the importance of the server and the amount of data). Then the global security risk detection status is 0.1*0.4+0.3*0.3+0.2*0.3=0.19.

[0079] Finally, there is the step of obtaining permission statistics, which includes obtaining the status keywords of the target resource sharing state, or obtaining the status keywords of the target resource sharing state and the business node for applying to open a round of business resource sharing threads. The status keywords of the target resource sharing state are a descriptive identifier of the characteristics of the resource sharing state. For example, the status keywords can be "high concurrency", "data intensive", "low latency requirements", etc. These keywords can be obtained by analyzing the resource usage pattern, business requirements, etc. of the resource sharing state. Text mining algorithms can be used to analyze documents describing the resource sharing state (such as business requirements documents, system configuration documents, etc.) to extract words that appear frequently and are related to resource sharing characteristics as status keywords. If the business node for applying to open a round of business resource sharing threads is also obtained, the scope and nature of the permission can be further clarified. The business node can be a specific business department, a link in the business process, etc. For example, if the business node is the sales department, combined with the status keyword "data intensive" of the target resource sharing state, it can be determined that in this case, the permission setting of the cloud server related to the sales department in the resource sharing state should take into account factors such as large data volume and possible need for fast data access.

[0080] After obtaining X groups of cloud service interaction information, resource demand information, and permission statistics information, the subsequent process is consistent with the previous description. That is, map X groups of cloud service interaction information into X cloud service interaction attribute knowledge. Take the shared resource operation record in the resource operation event semantics as an example. If there are many data read operations in the operation record, the attribute weight related to data read may be higher when mapped to cloud service interaction attribute knowledge. Map resource demand information into resource demand quantitative knowledge. Factors such as the global security risk detection status mentioned above can participate in the calculation of quantitative knowledge. Map permission statistics information into permission statistical distribution knowledge. The status keywords of the target resource sharing status can help construct the permission distribution matrix.

[0081] Then, knowledge collision is performed to obtain the business interaction permission collision feature. For example, in the case where the data reading attribute weight in the cloud service interaction attribute knowledge is high, if the corresponding data reading permission in the permission statistical distribution knowledge is high, then the data reading related value in the business interaction permission collision feature obtained in the knowledge collision process will also be high.

[0082] The process of determining the current collaborative weight based on the collision characteristics of business interaction permissions, quantitative knowledge of resource requirements, and statistical distribution knowledge of permissions is similar to the previous description. These knowledge are spliced ​​into target splicing knowledge, such as splicing vectors or matrices containing different attribute weights, quantitative knowledge, and permission distribution in a certain order. When performing attention processing, for example, higher attention weights are given to elements associated with high permissions. The fully connected processing uses a multi-layer perceptron algorithm to calculate the fully connected attention knowledge based on the set number of neurons and activation function. Finally, the business collaborative operation is performed through a business collaborative analysis model (such as a model built based on a logistic regression algorithm) to obtain the current collaborative weight.

[0083] It can be seen that by splitting step 110 into three sub-steps, various types of information related to cloud services can be obtained in a more detailed and comprehensive manner. Resource operations, permission authentication, and security risks are considered when obtaining cloud service interaction information, which can accurately reflect the operation and permission status of the cloud server. Obtaining the global security risk status in resource demand information helps to evaluate resource requirements as a whole. Obtaining status keywords or business nodes in permission statistics makes the permission settings more in line with actual business. A series of subsequent operations determine the collaborative weights based on this comprehensive information, improve the accuracy, security, and adaptability of resource sharing decisions, and optimize the operation of the business resource sharing system.

[0084] In a preferred design idea, the Z group of cloud servers are determined as the groups of cloud servers connected to the round of business resource sharing thread on the basis that the current collaboration weight meets the set sharing judgment index, including:

[0085] On the basis that the comparison result between the current collaboration weight and the set weight coefficient is less than the set comparison value, the Z group of cloud servers are determined as the groups of cloud servers connected to the round of business resource sharing threads;

[0086] Alternatively, based on the fact that the comparison result between the current collaborative weight and the set weight coefficient among the acquired W collaborative weights is the smallest, the Z groups of cloud servers are determined as the groups of cloud servers connected to the round of business resource sharing threads;

[0087] Among them, the W collaborative weights include the collaborative weights determined by the target resource sharing processing network according to the cloud service interaction information, resource demand information and permission statistical information of each cloud server classification cluster in the W cloud server classification clusters, each cloud server classification cluster in the W cloud server classification clusters includes multiple groups of cloud servers obtained by clustering the X cloud servers according to a corresponding one of W different induction strategies, the W different induction strategies include the current induction strategy, the number of groups of cloud servers included in each cloud server classification cluster is the same, and W is a positive integer not less than 2; or on the basis that the comparison result between the current collaborative weight and the set weight coefficient among the W collaborative weights is the smallest, and the comparison result is less than the set comparison value, the Z groups of cloud servers are determined as the groups of cloud servers connected to the round of business resource sharing threads.

[0088] Under this preferred design idea, the entire technical solution revolves around how to determine how to connect Z groups of cloud servers to each group of cloud servers in a round of business resource sharing thread based on the fact that the current collaborative weight meets the set sharing judgment index.

[0089] First, let's look at the first case, that is, on the basis that the comparison result between the current collaborative weight and the set weight coefficient is less than the set comparison value, the Z group of cloud servers are determined as the groups of cloud servers connected to a round of business resource sharing threads. This involves comparing the current collaborative weight with the set weight coefficient, and this comparison result needs to meet the condition of being less than the set comparison value. For example, the set weight coefficient is 0.8, and the set comparison value is 0.1. When the current collaborative weight obtained through the previous series of operations is 0.75, 0.8-0.75=0.05, 0.05 is less than 0.1, and the condition is met. At this time, the Z group of cloud servers can be determined as the groups of cloud servers connected to a round of business resource sharing threads. In this process, the setting of weight coefficients and comparison values ​​is based on the system's comprehensive consideration of multiple factors such as resource sharing security and effectiveness. The weight coefficient can be determined based on factors such as historical experience data of the cloud service system and the importance of business needs. For example, if the business has high requirements for the stability of resource sharing, the weight coefficient may be set relatively high; the comparison value is related to the system's tolerance for errors or risks. If the system has a low tolerance for risks in resource sharing, the comparison value will be set smaller.

[0090] Next, let's look at the second case. On the basis that the comparison result between the current collaborative weight and the set weight coefficient among the W collaborative weights obtained is the smallest, the Z groups of cloud servers are determined as the groups of cloud servers connected to a round of business resource sharing threads. The concept of W collaborative weights is introduced here. The W collaborative weights are collaborative weights determined by the target resource sharing processing network based on the cloud service interaction information, resource demand information and authority statistics of each cloud server classification cluster in the W cloud server classification clusters. Each cloud server classification cluster is a group of cloud servers obtained by clustering X cloud servers according to one of the W different induction strategies, and the number of groups of cloud servers included in each cloud server classification cluster is the same. For example, when W=3, there are three different induction strategies. For the first induction strategy, the cloud server classification cluster obtained by clustering X cloud servers may contain 3 groups of cloud servers. The cloud service interaction information, resource demand information and permission statistics of these 3 groups of cloud servers are input into the target resource sharing processing network to obtain a collaborative weight; similarly, the cloud server classification clusters obtained under the other two induction strategies also obtain a collaborative weight respectively, thus obtaining 3 collaborative weights. For example, the weight coefficient is set to 0.8, and the three collaborative weights are 0.7, 0.75 and 0.85 respectively. For the case of collaborative weight 0.7, 0.8-0.7=0.1, for 0.75, 0.8-0.75=0.05, for 0.85, 0.85-0.8=0.05, and 0.1 is the largest of the three differences, that is, the comparison result between 0.7 and the set weight coefficient 0.8 is the smallest, so in this case, the Z group of cloud servers corresponding to this collaborative weight (here, the Z group of cloud servers is the cloud server group associated with the collaborative weight that produces this minimum comparison result) will be determined as each group of cloud servers in the access round of business resource sharing threads. This determination method based on the minimum comparison results of multiple collaborative weights can select the cloud server group that best meets the system requirements from multiple possible cloud server clustering and collaborative weight calculation results.

[0091] Finally, let's look at the third case, that is, on the basis that the comparison result between the current collaborative weight and the set weight coefficient among the W collaborative weights is the smallest, and this comparison result is less than the set comparison value, the Z group of cloud servers are determined to be the groups of cloud servers connected to a round of business resource sharing threads. This is a combination of the first two cases. For example, when W=3, the set weight coefficient is 0.8, the set comparison value is 0.1, and the three collaborative weights are 0.7, 0.75, and 0.85 respectively. For the case of a collaborative weight of 0.7, 0.8-0.7=0.1, and 0.1 is equal to the set comparison value; for the case of 0.75, 0.8-0.75=0.05, and 0.05 is less than 0.1; for the case of 0.85, 0.85-0.8=0.05, and 0.05 is less than 0.1. Although the difference between 0.7 and 0.8 is equal to the set comparison value, the difference between 0.75 and 0.85 and 0.8 is smaller. Among the two, the difference between 0.75 and 0.8 is relatively smaller, that is, the comparison result between 0.75 and the set weight coefficient is the smallest and less than the set comparison value, so the Z group of cloud servers corresponding to this collaborative weight will be determined as the groups of cloud servers in the access round of business resource sharing threads. This method more strictly selects the cloud server group that is most suitable for accessing the business resource sharing thread, considering both the smallest comparison result with the weight coefficient and the fact that this comparison result is less than the comparison value, and combining multiple factors such as accuracy and risk control.

[0092] With this design, by comparing with the set weight coefficient and comparison value, the cloud server group that meets the system requirements can be accurately screened out to ensure the effectiveness and security of resource sharing. The situation with the smallest comparison result among multiple collaborative weights, or the situation where the comparison result is the smallest and less than the comparison value at the same time, fully considers the various possibilities under different induction strategies, making the selection of cloud server groups more comprehensive and reasonable. This helps to improve the efficiency of business resource sharing, optimize resource allocation, reduce risks in the resource sharing process, and improve the stability and reliability of the entire business resource sharing system.

[0093] Based on the above content, the method further includes a debugging step of the resource sharing processing network, and the debugging step includes:

[0094] (1) Obtain a debugging example set and a collaborative authentication weight set;

[0095] The debugging example set includes H debugging examples, each debugging example includes X groups of cloud service interaction information examples, resource demand information examples and permission statistics information examples, the X groups of cloud service interaction information examples include Y business behavior description examples of each cloud server example in the X cloud server examples, the X cloud server examples are cloud servers that apply to open a round of business resource sharing thread examples under the resource sharing state example in the structured data network, the Y business behavior description examples of each cloud server example include the structured data usage requirements of each cloud server example under the multiple resource sharing states, the resource demand information examples include the structured data usage requirements of each group of cloud server examples in the Z group of cloud server examples under the multiple resource sharing states, the permission statistics information examples include differentiated permission labels of the resource sharing state examples, the Z group of cloud server examples are multiple groups of cloud server examples obtained by clustering the X cloud server examples according to the induction strategy example, the collaborative authentication weight set includes H collaborative authentication weights corresponding to each of the H debugging examples, and H is a positive integer not less than 2;

[0096] (2) debugging the original resource sharing processing network using the debugging example set until the error variable of the target training error corresponding to the resource sharing processing network meets the set training requirements, terminating the debugging, and determining the resource sharing processing network at the time of terminating the debugging as the target resource sharing processing network;

[0097] Wherein, on the basis that the error variable of the target training error corresponding to the resource sharing processing network does not meet the set training requirements, the network weights in the resource sharing processing network are optimized; the error variable of the target training error is a difference value determined based on a collaborative weight example output by the resource sharing processing network and a collaborative authentication weight corresponding to the H collaborative authentication weights, wherein the collaborative weight example is a collaborative weight determined by the resource sharing processing network based on the X groups of cloud service interaction information examples, the resource demand information examples and the permission statistical information examples.

[0098] In this embodiment, the debugging steps of the resource sharing processing network are mainly described, which is a very critical part of the entire technical solution.

[0099] First, in the step of obtaining the debugging example set and the collaborative authentication weight set, many specific elements are included. The debugging example set contains H debugging examples, where H is a positive integer not less than 2. Taking H=3 as an example, each debugging example consists of specific information.

[0100] For each X group of cloud service interaction information examples in each debugging example, it covers Y business behavior description examples of each cloud server example in the X cloud server examples. For example, X=5, Y=4, for each of the five cloud server examples, there are 4 business behavior description examples. These business behavior description examples contain the structured data usage requirements of each cloud server example in multiple resource sharing states. For example, the business behavior description example of a cloud server example may be that in resource sharing state 1, there are 10 data read operations per hour, the amount of data read each time is 5MB, and the main data type read is the user information table in the relational database; in resource sharing state 2, there are 2 data write operations per day, the amount of data written is 20MB, and the data type written is log files, etc. These business behavior description examples record the operations of cloud servers related to structured data in different resource sharing states in a structured way.

[0101] The resource demand information example includes the structured data usage requirements of each group of cloud server examples in the Z group of cloud server examples under multiple resource sharing states. For example, Z=3, for the first group of cloud server examples in resource sharing state 1, a total of 100MB of structured data is required, the main required data type is the configuration file in XML format, and the data access frequency is once every 3 hours; in resource sharing state 2, 200MB of structured data is required, the data type becomes the order table in the relational database, and the access frequency is once every hour, etc.

[0102] The permission statistics example includes differentiated permission labels for resource sharing state examples. For example, the permission label can be expressed as a vector form, such as [1, 0, 1, 0], which means that under this resource sharing state example, the cloud server has permission for certain operations (1 means permission, 0 means no permission), which may mean that it has creation and modification permissions, but no read and delete permissions. These permission labels clearly define the scope of the cloud server's operation permissions in a specific resource sharing state.

[0103] At the same time, the collaborative authentication weight set includes H collaborative authentication weights corresponding to each of the H debugging examples. These H collaborative authentication weights are predetermined standard weights used to measure the accuracy of the output of the resource sharing processing network. For example, the collaborative authentication weight corresponding to the first debugging example is 0.8, the collaborative authentication weight corresponding to the second debugging example is 0.75, the collaborative authentication weight corresponding to the third debugging example is 0.9, and so on.

[0104] Next is the process of debugging the original resource sharing processing network using the debugging example set. The goal of this process is to make the error variable of the target training error corresponding to the resource sharing processing network meet the set training requirements. The error variable of the target training error is the difference value determined based on the collaborative weight example output by the resource sharing processing network and the corresponding collaborative authentication weight in the H collaborative authentication weights.

[0105] Specifically, when the X groups of cloud service interaction information examples, resource demand information examples, and permission statistics information examples in the debugging example set are input into the original resource sharing processing network, the resource sharing processing network will determine the collaborative weight examples based on these inputs. For example, after the first debugging example is input, the collaborative weight example output by the resource sharing processing network is 0.78, and its corresponding collaborative authentication weight is 0.8, then the error variable of the target training error is 0.8-0.78=0.02.

[0106] On the basis that the error variable of the target training error corresponding to the resource sharing processing network does not meet the set training requirements, the network weights in the resource sharing processing network are optimized. A variety of algorithms can be used to optimize the network weights, such as the gradient descent algorithm. For example, the resource sharing processing network is a multi-layer neural network, in which the connection weights between neurons in a certain layer are w1, w2, w3, etc. When calculating the gradient of the target training error to these weights, if the error variable is large, it means that the current weight is not conducive to accurately outputting the collaborative weight example. According to the gradient descent algorithm, the new weight w1'=w1-α*∂E / ∂w1, where α is the learning rate (for example, α=0.01), and ∂E / ∂w1 is the partial derivative of the target training error E with respect to w1. By continuously calculating and updating the weights in this way, the difference between the collaborative weight example output by the resource sharing processing network and the corresponding collaborative authentication weight is gradually reduced.

[0107] This debugging process will continue until the error variable of the target training error corresponding to the resource sharing processing network meets the set training requirements, and the debugging is terminated, and the resource sharing processing network at the time of the termination of the debugging is determined as the target resource sharing processing network. The set training requirement can be that the error variable is less than a certain threshold, such as 0.01. When the error variable meets this requirement by continuously optimizing the network weights, the resource sharing processing network is considered to have been debugged and can be used in the actual business resource sharing processing process.

[0108] Therefore, the debugging step of the resource sharing processing network is of great significance to the entire technical solution. By obtaining a debugging example set containing detailed cloud service-related information and a corresponding collaborative authentication weight set, the resource sharing processing network can be fully trained. The network weights are optimized based on the error variable of the target training error to ensure that the collaborative weights output by the network are as close to the standard collaborative authentication weights as possible. This helps to improve the accuracy and reliability of the resource sharing processing network in practical applications, thereby making the decision of the cloud server access business resource sharing thread determined based on the network more scientific and reasonable, optimizing the performance of the entire business resource sharing system, and improving the effectiveness and security of resource allocation.

[0109] In an extensible technical solution, the debugging of the original resource sharing processing network using the debugging example set includes performing the u-th round of debugging in the following manner, where u is a positive integer not less than 2:

[0110] Inputting the u-th debugging example among the H debugging examples into the resource sharing processing network obtained by the u-1-th round of debugging, wherein the u-th debugging example corresponds to the u-th collaborative authentication weight among the H collaborative authentication weights, and the u-th debugging example includes the Xu group cloud service interaction information example, the u-th resource demand information example and the u-th permission statistical information example, Xu=X;

[0111] In the resource sharing processing network obtained in the u-1th round of debugging, the Xu group of cloud service interaction information is mapped to Xu cloud service interaction attribute knowledge examples, the uth resource demand information example is mapped to the uth resource demand quantification knowledge example, and the uth permission statistical information example is mapped to the uth permission statistical distribution knowledge example, wherein the Xu cloud service interaction attribute knowledge examples correspond to the Xu cloud server examples in the uth debugging example respectively;

[0112] In the resource sharing processing network obtained by the u-1th round of debugging, the Xu cloud service interaction attribute knowledge examples are respectively subjected to knowledge collision with the uth resource demand quantification knowledge example to obtain Xu business interaction authority collision feature examples, wherein the Xu business interaction authority collision feature examples correspond to the Xu cloud server examples in the uth debugging example respectively;

[0113] In the resource sharing processing network obtained by the u-1th round of debugging, based on the Xu business interaction authority collision feature examples, the uth resource demand quantification knowledge example and the uth authority statistical distribution knowledge example, determine the uth collaboration weight;

[0114] Determining a uth error variable of the target training error based on the uth collaborative weight and the uth collaborative authentication weight;

[0115] On the basis that the u-th error variable of the target training error meets the set training requirements, debugging is terminated, and the resource sharing processing network at the time of debugging termination is determined as the target resource sharing processing network; on the basis that the u-th error variable of the target training error does not meet the set training requirements, the network weights in the resource sharing processing network obtained by the u-1-th round of debugging are improved to obtain the resource sharing processing network obtained by the u-th round of debugging.

[0116] In the above technical solution, first, for the u-th round of debugging (u is a positive integer not less than 2), the debugging process has a series of rigorous steps. The u-th debugging example among the H debugging examples is input into the resource sharing processing network obtained by the u-1-th round of debugging. Here, the u-th debugging example corresponds to the u-th collaborative authentication weight among the H collaborative authentication weights, and the u-th debugging example contains Xu groups of cloud service interaction information examples, the u-th resource demand information example, and the u-th permission statistics information example, where Xu=X. For example, when X=5, the 5 groups of cloud service interaction information examples in the u-th debugging example contain rich information about the 5 cloud server examples.

[0117] In the resource sharing processing network obtained by the u-1th round of debugging, a series of mapping operations are performed. Map Xu groups of cloud service interaction information to Xu cloud service interaction attribute knowledge examples. Taking the operation behavior coding in the cloud service interaction information example as an example, if a three-digit coding system is adopted, such as 101 represents data storage operation, 202 represents data retrieval operation, etc. For the operation behavior coding of each cloud server example, it is converted into a cloud service interaction attribute knowledge example through a specific mapping algorithm. For example, for data storage operation 101, it is mapped to the storage-related attribute weight of 0.8, the data flow-related attribute weight of 0.1, the operation complexity-related attribute weight of 0.1, etc. in the cloud service interaction attribute knowledge example by looking up the table. In this way, the operation behavior coding in 5 groups of cloud service interaction information examples is converted into 5 cloud service interaction attribute knowledge examples, and these 5 cloud service interaction attribute knowledge examples correspond to the 5 cloud server examples in the uth debugging example.

[0118] At the same time, the u-th resource demand information example is mapped to the u-th resource demand quantitative knowledge example. For example, the u-th resource demand information example may contain the structured data usage requirements of the cloud server group under different resource sharing states. For example, in a certain resource sharing state, the data volume requirement is 1000MB, the data type is the order table in the relational database (1 can be used to represent the relational database type), the data access frequency is 5 times per hour, and the data timeliness requirement is 0.6 (0 means that real-time data is not required, and 1 means that real-time data is required). Set the data volume requirement weight to 0.4, the data type weight to 0.2, the data access frequency weight to 0.3, and the data timeliness weight to 0.1, then the u-th resource demand quantitative knowledge example is calculated as:

[0119] 1000*0.4+1*0.2+5*0.3+0.6*0.1=400+0.2+1.5+0.06=401.76.

[0120] And the u-th permission statistics information example is mapped to the u-th permission statistics distribution knowledge example. For example, the permission label of the u-th permission statistics information example is [1, 0, 1, 0], which means that there are create and modify permissions, but no read and delete permissions in this resource sharing state. Construct a 4×4 permission distribution matrix, in which the values ​​of the create permission position (first row and first column) and the modify permission position (third row and third column) are 1, the values ​​of the read permission position (second row and second column) and the delete permission position (fourth row and fourth column) are 0, and the other elements are 0, so as to convert the permission statistics information example into the u-th permission statistics distribution knowledge example.

[0121] Next, in the resource sharing processing network obtained by the u-1th round of debugging, the Xu cloud service interaction attribute knowledge examples are respectively subjected to knowledge collision with the u-th resource demand quantification knowledge example to obtain Xu business interaction permission collision feature examples. Take the knowledge collision between the cloud service interaction attribute knowledge example corresponding to the first cloud server example in the u-th debugging example and the u-th resource demand quantification knowledge example as an example. For example, the cloud service interaction attribute knowledge example is in the form of a vector [0.8, 0.6, 0.4] (the elements here correspond to different operation attribute quantification values), and the u-th resource demand quantification knowledge example is 401.76. The algorithm for knowledge collision can be to perform some operation on each element in the vector with the relevant elements in the resource demand quantification knowledge example. For example, for the first element of the vector 0.8, for example, according to the data volume requirement in the business rule and resource demand quantification knowledge example, 0.8*1000 (the 1000 here is simplified from the previous data volume requirement of 1000MB) = 800 is calculated, for the second element 0.6, for example, it is related to the data access frequency, 0.6*5=3 is calculated, for the third element 0.4, for example, it is related to the data timeliness, 0.4*0.6=0.24 is calculated, and these results are combined to obtain a new vector [800, 3, 0.24] as the first business interaction authority collision feature example. In the same way, the cloud service interaction attribute knowledge examples corresponding to other cloud server examples are processed to obtain Xu business interaction authority collision feature examples, and each corresponds to the Xu cloud server examples in the u-th debugging example.

[0122] Then, in the resource sharing processing network obtained in the u-1th round of debugging, the uth collaborative weight is determined based on the Xu business interaction permission collision feature examples, the uth resource demand quantification knowledge example, and the uth permission statistical distribution knowledge example. One possible way to determine is to build a calculation model. For example, the business interaction permission collision feature example is represented by a vector as [800, 3, 0.24] (this is only an example, the actual dimension is related to the number of operation attributes), the uth resource demand quantification knowledge example is 401.76, and the uth permission statistical distribution knowledge example is represented by a matrix as the above-mentioned 4×4 matrix. First, the business interaction permission collision feature example vector is normalized, for example, each element in the vector is divided by the sum of the vector elements to obtain the normalized vector [0.998, 0.004, 0.002]. Then, the normalized vector elements are weighted with the relevant elements in the resource demand quantification knowledge example and the permission statistical distribution knowledge example. For example, the weight of the business interaction permission collision feature example is set to 0.5, the weight of the resource demand quantification knowledge example is set to 0.3, and the weight of the permission statistical distribution knowledge example is set to 0.2. For example, if the value of a key element (such as the element corresponding to the core operation permission) is selected from the permission statistical distribution knowledge example and is 0.8, then the u-th collaborative weight is calculated as:

[0123] 0.998*0.5*401.76+0.004*0.5*0.8+0.002*0.2*0.8=200.27+0.0016+0.00032=200.27192.

[0124] Based on the u-th collaborative weight and the u-th collaborative authentication weight, determine the u-th error variable of the target training error. For example, if the u-th collaborative authentication weight is 0.8 and the u-th collaborative weight is 0.75, then the u-th error variable is 0.8-0.75=0.05.

[0125] On the basis that the u-th error variable of the target training error meets the set training requirements, the debugging is terminated, and the resource sharing processing network at the time of the termination of the debugging is determined as the target resource sharing processing network. The set training requirement can be that the error variable is less than a certain threshold, such as 0.01. If the u-th error variable is 0.005, which is less than 0.01, the requirement is met, and the debugging is terminated at this time, and the current resource sharing processing network is determined as the target resource sharing processing network. On the basis that the u-th error variable of the target training error does not meet the set training requirements, the network weights in the resource sharing processing network obtained by the u-1 round of debugging are improved to obtain the resource sharing processing network obtained by the u-1 round of debugging. The gradient descent algorithm can be used to improve the network weights. For example, the resource sharing processing network is a multi-layer neural network, and the connection weights between neurons in a certain layer are w1, w2, w3, etc. When calculating the gradient of the target training error to these weights, if the error variable is large, it means that the current weights are not conducive to accurately outputting the collaborative weights. According to the gradient descent algorithm, the new weight w1'=w1-α*∂E / ∂w1, where α is the learning rate (for example, α=0.01), and ∂E / ∂w1 is the partial derivative of the target training error E with respect to w1. By continuously calculating and updating the weights in this way, the resource sharing processing network obtained in the uth round of debugging is obtained, so as to carry out the next round of debugging until the set training requirements are met.

[0126] In this way, through multiple rounds of iterative debugging of the resource sharing processing network, detailed information mapping, knowledge collision, collaborative weight calculation and error variable evaluation are performed in each round. This step-by-step optimization method can accurately adjust the network weights so that the collaborative weights output by the resource sharing processing network are constantly close to the collaborative authentication weights. Through a rigorous mapping and calculation process, the impact of multiple factors such as cloud service interaction, resource demand and authority statistics on the collaborative weights is fully considered, thereby improving the accuracy and reliability of the resource sharing processing network, ensuring that scientific and reasonable decisions can be made in actual business resource sharing, and optimizing the effects of resource allocation and sharing.

[0127] In another scalable technical solution, the step of inputting the X group of cloud service interaction information, resource demand information, and authority statistics information into a debugged target resource sharing processing network to obtain a current collaborative weight output by the target resource sharing processing network includes:

[0128] Inputting the X group of cloud service interaction information, resource demand information, and permission statistics information into the debugged target resource sharing processing network to obtain a first collaboration weight output by the target resource sharing processing network, wherein the current collaboration weight includes the first collaboration weight, and the first collaboration weight is used to characterize the confidence score of a group of cloud servers set in the Z group of cloud servers completing business collaboration in the round of business resource sharing threads on the basis of determining the Z group of cloud servers as the groups of cloud servers connected to the round of business resource sharing threads;

[0129] Alternatively, the X group of cloud service interaction information, resource demand information and permission statistical information are input into the debugged target resource sharing processing network to obtain a second collaboration weight output by the target resource sharing processing network, wherein the current collaboration weight includes the second collaboration weight, and the second collaboration weight is used to characterize the confidence score of each group of cloud servers in the Z group of cloud servers completing business collaboration in the round of business resource sharing threads on the basis of determining the Z group of cloud servers as each group of cloud servers connected to the round of business resource sharing threads.

[0130] In this scalable technical solution, the core is to explain how to input X groups of cloud service interaction information, resource demand information and permission statistics information into the debugged target resource sharing processing network, so as to obtain the current coordination weight output by the target resource sharing processing network.

[0131] First, let's look at the situation where this information is input into the target resource sharing processing network to obtain the first collaborative weight. When the X group of cloud service interaction information, resource demand information, and permission statistics information are input into the debugged target resource sharing processing network, the target resource sharing processing network will calculate and output the first collaborative weight based on its internal structure and algorithm. The first collaborative weight here has a specific meaning. It is used to characterize the confidence score of a group of cloud servers set in the Z group of cloud servers completing business collaboration in a round of business resource sharing threads, based on the determination of the Z group of cloud servers as the groups of cloud servers connected to a round of business resource sharing threads.

[0132] For example, X=5, which means there are 5 groups of cloud service interaction information. For cloud service interaction information, it contains information such as the operation behavior coding of each cloud server in multiple resource sharing states. For example, the operation behavior coding of a cloud server is 101 (indicating data storage operation), 202 (indicating data retrieval operation), etc. At the same time, the resource demand information contains the structured data usage requirements of each group of cloud servers in the Z group of cloud servers in multiple resource sharing states. For example, a group of cloud servers needs 1000MB of structured data in a certain resource sharing state, the data type is the order table in the relational database, the data access frequency is 5 times per hour, and the data timeliness requirement is 0.6 (0 means no real-time data is required, 1 means real-time data is required). The permission statistics information contains the differentiated permission labels of the target resource sharing state. For example, the permission label is [1, 0, 1, 0], which means that there are create and modify permissions in this resource sharing state, but no read and delete permissions.

[0133] After this information is input into the target resource sharing processing network, the target resource sharing processing network may be a neural network-based model, such as a multi-layer perceptron (MLP). In this network, the number of neurons in the input layer is set according to the dimension of the input information. For example, the cloud service interaction information is encoded and quantized into a 100-dimensional vector, the resource demand information is a 50-dimensional vector, and the permission statistics information is a 30-dimensional vector, then the number of neurons in the input layer can be set to 180. The hidden layer can be set to 2 layers, with 100 neurons in the first hidden layer and 50 neurons in the second hidden layer, and ReLU is used as the activation function.

[0134] When calculating the first collaborative weight, the network will perform a series of calculations according to its internal weights and activation functions. For example, in the calculation from the input layer to the first hidden layer, the calculation formula for the output of each neuron is:

[0135] h_1=ReLU(sum_{i=1}^{180}w_{i1}x_i+b_1), where w_{i1} is the weight from the input layer to the first hidden layer, x_i is the element of the input vector, and b_1 is the bias term. Through such layer-by-layer calculations, a value representing the first collaborative weight is finally obtained. For example, if this value is 0.8, it means that on the basis of determining the Z group of cloud servers as the groups of cloud servers connected to a round of business resource sharing threads, the set group of cloud servers has a higher confidence score for completing business collaboration in a round of business resource sharing threads.

[0136] Next, we can see the second collaborative weight. Similarly, the X-group cloud service interaction information, resource demand information, and permission statistics information are input into the debugged target resource sharing processing network to obtain the second collaborative weight output by the target resource sharing processing network. The second collaborative weight is used to characterize the confidence score of each group of cloud servers in the Z group of cloud servers completing business collaboration in a round of business resource sharing threads, based on the determination of each group of cloud servers in the Z group of cloud servers as connected to a round of business resource sharing threads.

[0137] For example, still taking X=5 as an example, the specific contents of the input cloud service interaction information, resource demand information and permission statistical information are similar to the previous ones. When the target resource sharing processing network calculates the second collaborative weight, its calculation process is similar to that of obtaining the first collaborative weight, but may differ in certain parameters or calculation focuses. For example, when the target resource sharing processing network calculates the second collaborative weight, the weight distribution of the operation behavior coding in the cloud service interaction information is adjusted. For the data storage operation of the operation behavior coding 101, the weight given when calculating the first collaborative weight is 0.3, and the weight given when calculating the second collaborative weight is 0.4. This weight adjustment may be based on different business considerations or different evaluation methods of resource sharing situations.

[0138] For example, when calculating the second collaboration weight, the value obtained by the target resource sharing processing network is 0.75. This value indicates the confidence score of each group of cloud servers in the Z group of cloud servers completing business collaboration in a round of business resource sharing threads, based on the determination of the Z group of cloud servers as the groups of cloud servers connected to a round of business resource sharing threads. The 0.75 here, compared to the previous 0.8, may mean that when considering all the Z group of cloud servers, the overall confidence in completing business collaboration is relatively low, which may be due to some potential risks or uncertainties in the collaboration between different cloud server groups.

[0139] In the entire technical solution, both the first collaborative weight and the second collaborative weight become components of the current collaborative weight. This way of distinguishing different collaborative weights can evaluate the collaboration of cloud servers in the business resource sharing thread from different angles. By considering a set group of cloud servers (first collaborative weight) and all Z groups of cloud servers (second collaborative weight) separately, the collaborative potential and risk of cloud servers in business resource sharing can be more comprehensively and meticulously grasped.

[0140] Based on the above, by distinguishing the first collaboration weight and the second collaboration weight, the collaboration of cloud servers in the business resource sharing thread can be more comprehensively evaluated. When calculating the collaboration weight, multiple factors such as cloud service interaction information, resource demand information, and permission statistics are taken into account, and specific algorithm examples are used to show how to perform calculations in the target resource sharing processing network. This helps to improve the accuracy of the evaluation of cloud server collaboration, thereby making better resource sharing decisions, optimizing the efficiency and security of business resource sharing, and also provides a more flexible mechanism for responding to different business needs and risk assessments.

[0141] In summary, the embodiments of the present application have many advantages by obtaining various information and performing specific processing. First, the cloud service interaction information, resource demand information and permission statistics of group X are integrated to cover the behavior, demand and permission of cloud servers under various resource sharing states, and comprehensively and meticulously describe the cloud service status. This information is input into the debugged target resource sharing processing network to obtain the current collaboration weight, which can quantify the confidence score of the cloud server group in business collaboration and scientifically evaluate the collaboration potential of the cloud server. Finally, the cloud server group to be connected is determined based on the set sharing discrimination index to ensure that only qualified cloud servers participate in the business resource sharing thread, optimize resource allocation, improve the rationality, security and effectiveness of resource sharing, and ensure smooth and efficient business collaboration.

[0142] Furthermore, a readable storage medium is provided, on which a program is stored, and the program implements the above method when executed by a processor.

[0143] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described in the above exemplary embodiment can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here. In the several embodiments provided in the embodiments of the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

Claims

1. A business resource sharing method based on cloud computing, characterized in that: The method is applied to a business resource sharing system, and the method comprises: Obtaining X groups of cloud service interaction information, resource demand information, and authority statistics information, wherein the X groups of cloud service interaction information include Y business behavior descriptions of each cloud server in the X cloud servers, the X cloud servers are cloud servers that apply for a round of business resource sharing threads in a target resource sharing state in an open structured data network, the Y business behavior descriptions of each cloud server include operation behavior codes of each cloud server in multiple resource sharing states, the multiple resource sharing states are resource sharing states that have completed authentication in the structured data network, the resource demand information includes structured data usage requirements of each group of cloud servers in the Z group of cloud servers in the multiple resource sharing states, the authority statistics information includes differentiated authority labels of the target resource sharing state, the Z group of cloud servers are multiple groups of cloud servers obtained by clustering the X cloud servers according to a current induction strategy, and X, Y, and Z are all positive integers not less than 2; Inputting the X group of cloud service interaction information, resource demand information and authority statistics information into the debugged target resource sharing processing network to obtain the current collaboration weight output by the target resource sharing processing network, wherein the current collaboration weight is used to represent the confidence score of the Z group of cloud servers completing the business collaboration in the round of business resource sharing threads based on determining the Z group of cloud servers as the groups of cloud servers connected to the round of business resource sharing threads; On the basis that the current collaboration weight meets the set sharing judgment index, the Z groups of cloud servers are determined as groups of cloud servers connected to the round of business resource sharing thread.

2. The method according to claim 1, characterized in that The step of inputting the X group of cloud service interaction information, resource demand information and authority statistics information into the debugged target resource sharing processing network to obtain the current coordination weight output by the target resource sharing processing network includes: Mapping the X groups of cloud service interaction information into X cloud service interaction attribute knowledge, mapping the resource demand information into resource demand quantification knowledge, and mapping the authority statistical information into authority statistical distribution knowledge, wherein the X cloud service interaction attribute knowledge corresponds to each of the X cloud servers; Performing knowledge collisions on the X cloud service interaction attribute knowledge and the authority statistical distribution knowledge respectively to obtain X business interaction authority collision features, wherein the X business interaction authority collision features correspond to the X cloud servers respectively; The current collaboration weight is determined based on the X business interaction authority collision characteristics, the resource demand quantification knowledge, and the authority statistical distribution knowledge.

3. The method according to claim 2, characterized in that The performing knowledge collisions on the X cloud service interaction attribute knowledge and the authority statistical distribution knowledge respectively to obtain X business interaction authority collision features includes: The u-th cloud service interaction attribute knowledge in the X cloud service interaction attribute knowledge is collided with the authority statistical distribution knowledge in the following manner to obtain the u-th business interaction authority collision feature in the X business interaction authority collision features, wherein u is a positive integer not less than 1 and not greater than X: Performing feature fusion on the u-th cloud service interaction attribute knowledge and the authority statistical distribution knowledge to obtain the u-th target fusion knowledge; The weighted result of the u-th cloud service interaction attribute knowledge and the u-th target fusion knowledge is determined as the u-th business interaction permission collision feature; or, the knowledge feature of the u-th target fusion knowledge is enhanced using the set u-th feature enhancement coefficient to obtain the u-th target enhanced knowledge, and the integration result of the u-th cloud service interaction attribute knowledge and the u-th target enhanced knowledge is determined as the u-th business interaction permission collision feature.

4. The method according to claim 2, characterized in that: The determining the current collaboration weight based on the X business interaction authority collision characteristics, the resource demand quantification knowledge, and the authority statistical distribution knowledge includes: splicing the X business interaction authority collision features, the resource demand quantification knowledge and the authority statistical distribution knowledge into target splicing knowledge; Performing attention processing on the target splicing knowledge to obtain attention splicing knowledge, wherein the target splicing knowledge and the attention splicing knowledge have the same knowledge feature recognition degree; The attention splicing knowledge is fully connected to obtain fully connected attention knowledge, and a set business collaboration analysis model is used to perform business collaboration operations on the fully connected attention knowledge to obtain the current collaboration weight.

5. The method according to claim 1, characterized in that: The steps of obtaining X sets of cloud service interaction information include: The resource operation event semantics, security risk detection semantics, and authority authentication update semantics of each cloud server among the X cloud servers are obtained to obtain the X groups of cloud service interaction information, wherein the resource operation event semantics of each cloud server includes a shared resource operation record of each cloud server in a previous v-round business resource sharing thread, where v is equal to 1 or v is a positive integer not less than 2, the authority authentication update semantics of each cloud server includes an authority authentication tag of each cloud server in each of the multiple resource sharing states within a previous set period, and the security risk detection semantics of each cloud server includes a current security risk detection state of each cloud server in each of the multiple resource sharing states.

6. The method according to claim 1, characterized in that The step of obtaining resource demand information includes: obtaining a global security risk detection status of each group of cloud servers in the Z group of cloud servers in each resource sharing state in the multiple resource sharing states.

7. The method according to claim 1, characterized in that The steps to obtain permission statistics include: The state keyword of the target resource sharing state is obtained, or the state keyword of the target resource sharing state and the service node for applying to open the round of service resource sharing thread are obtained.

8. The method according to claim 1, characterized in that The determining the Z groups of cloud servers as the groups of cloud servers connected to the round of business resource sharing threads on the basis that the current collaboration weight meets the set sharing judgment index comprises: On the basis that the comparison result between the current collaborative weight and the set weight coefficient is less than the set comparison value, the Z group of cloud servers are determined to be the groups of cloud servers connected to the round of business resource sharing threads; or on the basis that the comparison result between the current collaborative weight and the set weight coefficient among the acquired W collaborative weights is the smallest, the Z group of cloud servers are determined to be the groups of cloud servers connected to the round of business resource sharing threads, wherein the W collaborative weights include the collaborative weights determined by the target resource sharing processing network according to the cloud service interaction information, resource demand information and authority statistics of each cloud server classification cluster in the W cloud server classification clusters. with the same weight, each of the W cloud server classification clusters includes multiple groups of cloud servers obtained by clustering the X cloud servers according to a corresponding induction strategy among W different induction strategies, the W different induction strategies include the current induction strategy, the multiple groups of cloud servers included in each cloud server classification cluster have the same number of groups, and W is a positive integer not less than 2; or on the basis that the comparison result between the current collaborative weight and the set weight coefficient among the W collaborative weights is the smallest and the comparison result is less than the set comparison value, the Z groups of cloud servers are determined as the groups of cloud servers connected to the round of business resource sharing threads.

9. The method according to any one of claims 1 to 8, characterized in that The method further comprises: Obtaining a debugging example set and a collaborative authentication weight set, wherein the debugging example set includes H debugging examples, each debugging example includes X groups of cloud service interaction information examples, resource demand information examples, and permission statistical information examples, the X groups of cloud service interaction information examples include Y business behavior description examples of each cloud server example in the X cloud server examples, the X cloud server examples are cloud servers that apply to open a round of business resource sharing thread examples under the resource sharing state example in the structured data network, the Y business behavior description examples of each cloud server example include the structured data usage requirements of each cloud server example under the multiple resource sharing states, the resource demand information examples include the structured data usage requirements of each group of cloud server examples in the Z group of cloud server examples under the multiple resource sharing states, the permission statistical information examples include differentiated permission labels of the resource sharing state examples, the Z group of cloud server examples are multiple groups of cloud server examples obtained by clustering the X cloud server examples according to the induction strategy example, and the collaborative authentication weight set includes H collaborative authentication weights corresponding to each of the H debugging examples, and H is a positive integer not less than 2; The original resource sharing processing network is debugged using the debugging example set until an error variable of a target training error corresponding to the resource sharing processing network meets a set training requirement, debugging is terminated, and the resource sharing processing network at the time of terminating debugging is determined as the target resource sharing processing network, wherein, on the basis that the error variable of the target training error corresponding to the resource sharing processing network does not meet the set training requirement, the network weights in the resource sharing processing network are optimized; the error variable of the target training error is a difference value determined based on a collaborative weight example output by the resource sharing processing network and a collaborative authentication weight corresponding to one of the H collaborative authentication weights, wherein the collaborative weight example is a collaborative weight determined by the resource sharing processing network based on the X groups of cloud service interaction information examples, the resource demand information examples, and the authority statistical information examples.

10. A business resource sharing system, characterized in that: It comprises a processor and a memory; the processor and the memory are communicatively connected, and the processor is used to read a computer program from the memory and execute it to implement the method described in any one of claims 1 to 9.

Images