Natural language-based ssl certificate full life cycle intelligent management method

Abstract

This invention discloses a natural language-based intelligent management method for the entire lifecycle of SSL certificates, relating to the field of automated operation and maintenance. The method performs semantic parsing of user-input natural language commands to identify certificate operation intentions and generate a structured set of operations; it constructs a certificate operation dependency graph based on the operation set and evaluates the reversibility of operation nodes; it generates and executes an operation execution sequence through topological sorting, incorporating backup and pre-verification mechanisms for operation control during execution; in the event of anomalies, it calculates the propagation factor based on the anomaly propagation range and selects the corresponding rollback strategy, achieving hierarchical recovery through reverse topological sorting; after operation completion, it performs a security scan of the target system and performs structured processing and fusion evaluation of multi-source detection results to generate security diagnostic results and remediation suggestions; this invention enables automated orchestration and fine-grained anomaly control of certificate operations, improving the efficiency and security of SSL certificate management.

Description

Technical Field

[0001] This invention relates to the field of automated operation and maintenance, and in particular to a method for intelligent management of the entire lifecycle of SSL certificates based on natural language. Background Technology

[0002] With the widespread deployment of internet services, SSL certificates play a fundamental role in server authentication, encrypted data transmission, and communication security. Currently, enterprises in multi-business systems and multi-server environments typically need to manage a large number of SSL certificates simultaneously, involving multiple stages such as certificate application, deployment, renewal, revocation, and security testing, exhibiting typical full lifecycle management characteristics.

[0003] In existing technologies, SSL certificate management typically relies on manual operation or script-based automation tools, which mainly suffers from the following problems: First, at the operation execution level, most existing solutions orchestrate certificate operations in a single-task or linear process, lacking the ability to model the system dependencies between multiple operations. In practical applications, different certificate operations often have complex dependencies, such as deployment depending on successful certificate application, and renewal depending on certificate validity status. Existing technologies struggle to structurally express and uniformly schedule these dependencies, easily leading to incorrect execution order or operational conflicts. Second, regarding exception handling and rollback control, existing solutions typically employ simple full rollbacks or manual intervention. When operations fail, there is a lack of precise assessment mechanisms for the scope of the exception's impact, making it impossible to select appropriate compensation strategies based on the degree of exception propagation. This can easily lead to unnecessary expansion of the system rollback scope, affecting system stability and service continuity. Third, in terms of multi-tool security testing, existing technologies typically use single scanning tools or simple result aggregation methods, lacking a unified structured expression model for results and a conflict resolution mechanism between multiple tool results. Different tools may give inconsistent conclusions on the same security issue, resulting in unstable or unreliable security assessment results. Furthermore, in complex operational scenarios, existing technologies typically lack the ability to uniformly manage the contextual information of the entire execution process, making it difficult to achieve cross-step data association, execution status tracking, and complete preservation of audit records, thus affecting the traceability and reliability of the system. Finally, with the development of natural language interaction technology, although some systems have attempted to trigger certificate operations via natural language, most remain at the level of simple command parsing, lacking systematic processing capabilities for multi-intent combinations, parameter completion, and implicit operation inference, making it difficult to meet the actual needs of complex operation and maintenance scenarios. Therefore, this paper proposes a natural language-based intelligent lifecycle management method for SSL certificates. Summary of the Invention

[0004] The main objective of this invention is to provide a natural language-based intelligent management method for the entire lifecycle of SSL certificates, which can effectively solve the problems in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A natural language-based intelligent lifecycle management method for SSL certificates, which uses a unified context manager to persistently manage and associate context information generated in each step during execution, includes the following steps:

[0007] S1: Receive natural language commands input by the user, and based on a classification system that includes seven types of certificate operation intentions such as application, renewal, revocation, deployment, query, scanning and configuration, perform intent recognition on the natural language commands through a pre-trained language model, and extract operation parameters according to the parameter patterns corresponding to the identified intent types; when the operation parameters are incomplete, complete them through multiple rounds of interaction with the user via a dialogue state machine.

[0008] S2: Based on one or more intents identified in S1, dynamically construct a certificate operation dependency graph. This graph is a directed acyclic graph (DAG), where nodes represent certificate operations. Directed edges include strong and weak dependencies. Strong dependency means that if a preceding operation fails, subsequent operations cannot be executed; weak dependency means that if a preceding operation fails, subsequent operations are not blocked. Each operation node is labeled with a reversibility level, which is based on the ratio of the number of recoverable state elements after the operation to the total number of state elements affected by the operation. The system performs quantitative classification, including three levels: fully reversible, partially reversible, and irreversible; it generates an ordered operation plan through topological sorting and identifies operation groups that can be executed in parallel.

[0009] S3: After presenting the operation plan to the user for confirmation, the operation plan is automatically executed according to the topological order. The execution process adopts a four-stage control mechanism of backup-pre-verification-execution-recovery: before executing the target operation, the current state is backed up and the feasibility of the compensation operation is pre-verified. When an anomaly occurs, the minimum rollback range is determined based on the reverse topological sorting of the certificate operation dependency graph and the state recovery is executed. Among them, when the anomaly occurs in a batch operation scenario involving multiple certificates, a full rollback strategy, a selective compensation strategy, or a minimum impact compensation strategy is selected according to the scope of the anomaly's impact.

[0010] S4: After the operation is completed, the security scanning tool is invoked through a standardized description model that includes four dimensions: functional semantics, input mode, output mode and preconditions. The scanning results returned by the security scanning tool are then structured and mapped into five security domains: protocol layer, cryptographic layer, certificate layer, vulnerability layer and compliance layer.

[0011] S5: Input the five-layer domain structured scanning results into the large language model to perform a security assessment from four dimensions: protocol security, cryptographic suite strength, certificate configuration compliance, and known vulnerabilities. When multiple scanning tools produce different conclusions on the same security item, a confidence-weighted fusion model is used for adjudication. The confidence level is calculated based on three factors: tool reliability factor, method strength factor, and time decay factor. Finally, a diagnostic report containing executable repair commands for different server types is generated.

[0012] Furthermore, in S1, multi-round interactive completion with the user is performed through a dialogue state machine, specifically including: based on the current user's certificate operation history, inferring the possible values ​​of missing parameters through context association, and verifying with the user in the form of a confirmation question; during the multi-round dialogue, maintaining the slot filling state based on a finite state machine, the states of which include an initial state, an intent identified state, a parameter filling state, and an intent completed state, and triggering an intent completed event when all necessary slots are filled; performing implicit intent recognition on user commands, and automatically incorporating the pre-operation into the operation plan when a pre-operation that is not explicitly stated but logically necessary is detected in the user command.

[0013] Furthermore, the selection of the rollback strategy in S3 specifically includes:

[0014] Calculate the propagation factor of anomalous operations ,in This is the set of downstream dependent nodes for abnormal operations. Let be the set of all nodes in the certificate operation dependency graph; when When the value exceeds a preset first threshold, a full rollback strategy is executed, and compensation operations are performed on all executed operations in the certificate operation dependency graph in reverse topological order; when When the value is between the second threshold and the first threshold, a selective compensation strategy is implemented, performing compensation only for abnormal operations and their downstream dependent operations; when... When the value is less than the second threshold, a minimum impact compensation strategy is executed, which only performs compensation operations on the abnormal operation itself, and its downstream operations are marked as pending verification.

[0015] Furthermore, the unified context manager specifically includes: constructing an operation audit chain using a Merkle chain structure, wherein each operation record contains operation content, timestamp, operator identifier, and hash value of the preceding record, and achieving the immutability of operation history through chained hash verification; storing the intermediate data and execution results generated by each step from S1 to S5 in the context storage in the form of key-value pairs, and establishing cross-step context associations through operation identifiers, so that subsequent steps can access the complete execution context of the preceding steps.

[0016] Furthermore, S5 further includes an automatic repair verification loop: based on the security issues in the diagnostic report, an executable repair command is generated using a large language model for the specific server type and operating system; the repair command is automatically executed after user confirmation; after the repair is completed, a new security scan of the target server is triggered, and the new scan results are compared and analyzed with the results before the repair to verify whether the security issues have been eliminated; the repair and verification steps are repeated until the security score reaches a preset qualified threshold or the maximum number of repair rounds is reached.

[0017] A natural language-based intelligent management system for the entire lifecycle of SSL certificates, including:

[0018] The natural language understanding module is used to receive natural language commands input by the user. Based on a classification system that includes seven types of certificate operation intentions such as application, renewal, revocation, deployment, query, scanning and configuration, it performs intent recognition and parameter extraction through a pre-trained language model. When the parameters are incomplete, it completes them through multiple rounds of interaction with the user through a dialogue state machine.

[0019] The operation orchestration module is used to dynamically construct a certificate operation dependency graph based on one or more intentions identified by the natural language understanding module. The certificate operation dependency graph is a directed acyclic graph, in which directed edges include two types: strong dependency and weak dependency. Each operation node is labeled with a reversibility level including fully reversible, partially reversible, and irreversible. An ordered operation plan is generated through topological sorting.

[0020] The reversible execution engine module is used to execute the operation plan in topological order. It adopts a four-stage control mechanism of backup-pre-verification-execution-recovery. When an anomaly occurs, it determines the minimum rollback range based on the reverse topological sorting of the certificate operation dependency graph and performs state recovery.

[0021] The security scanning integration module is used to call security scanning tools through a standardized description model that includes four dimensions: functional semantics, input mode, output mode, and preconditions, and to perform structured domain mapping of the scan results into five security domains: protocol layer, cryptographic layer, certificate layer, vulnerability layer, and compliance layer.

[0022] The intelligent diagnostic module is used to input the five-layer domain structured scanning results into a large language model for security assessment. When multiple scanning tools produce different conclusions on the same security item, a confidence-weighted fusion model is used to make a decision and generate a diagnostic report containing executable repair commands.

[0023] The unified context management module is used to persistently manage and associate the context information generated by each module.

[0024] Furthermore, the unified context management module includes a Merkle chain audit submodule, which records the immutable audit trail of all operations in a chained hash structure, wherein each audit record contains the operation content, timestamp, operator identifier and hash value of the preceding record.

[0025] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a natural language-based intelligent lifecycle management method for SSL certificates.

[0026] An electronic device includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement a natural language-based intelligent lifecycle management method for SSL certificates.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. This invention constructs a certificate operation dependency graph, which uniformly models multiple types of certificate operations in the form of a directed acyclic graph, distinguishes between strong and weak dependencies, generates an ordered execution sequence through topological sorting, and identifies operation groups that can be executed in parallel, thereby achieving structured orchestration and efficient execution in multiple operation scenarios.

[0029] 2. This invention establishes a reversibility assessment model, quantifies the operational recovery capability by using the ratio between recoverable state elements and affected state elements, and combines it with the recovery path for corrective assessment, thereby providing accurate basis for anomaly handling and improving the rationality of system recovery decisions.

[0030] 3. This invention constructs anomaly propagation paths and calculates propagation factors to quantify the scope of anomaly impact. It also combines reverse topology sorting to perform compensation operations, achieving hierarchical control of full rollback, selective compensation, and minimum impact compensation. This significantly reduces the rollback scope while ensuring system consistency.

[0031] 4. This invention constructs a unified structured description model to perform domain mapping on the detection results of multiple security scanning tools, and performs fusion adjudication on conflicting results based on a confidence weighting mechanism, thereby improving the consistency and reliability of security assessment results and generating actionable remediation suggestions.

[0032] 5. This invention uses a unified context manager to associate, store, and track data during execution, and combines an automatic repair and verification loop mechanism to achieve closed-loop processing from problem identification, repair execution to result verification, thereby improving the traceability and overall security of the system. Attached Figure Description

[0033] Figure 1 This is a system module diagram of the present invention;

[0034] Figure 2 This is a flowchart of the method of the present invention;

[0035] Figure 3 This is a diagram showing the dependencies and rollback mechanism of this invention;

[0036] Figure 4 This is a diagram illustrating the safety detection fusion and diagnosis of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0038] Example 1

[0039] like Figure 1-4 As shown, the natural language-based intelligent lifecycle management method for SSL certificates, by adopting the above technical solution, can be deployed in a multi-server enterprise environment to automate operations such as certificate application, deployment, renewal, and security testing. Compared with traditional methods that rely on manual intervention or scripts, this invention drives operations through natural language and combines operation dependency modeling and exception control mechanisms to enable multi-step certificate operations to be executed automatically in a reasonable order.

[0040] In practice, users can input operation commands via natural language, such as "renew the certificate for example.com and deploy it to server A, while performing a security check." Upon receiving this command, the system first performs semantic parsing, identifying multiple operation intentions through a pre-defined certificate operation intent classification system and extracting corresponding parameters. If parameters are incomplete, such as without explicitly specifying a deployment server, the system can complete them through dialogue, prompting the user to select a target server, thus obtaining the complete set of operations. Furthermore, during multiple rounds of interaction, historical operation records can be used to infer parameters, improving processing efficiency.

[0041] After obtaining the structured set of operations, the system further constructs the dependencies between certificate operations. Specifically, each operation is treated as a node, and directed connections are established between different operations based on their execution relationships. A strong dependency is defined when a subsequent operation must depend on the result of a preceding operation, while a weak dependency is defined as one that only affects the execution effect but does not block execution. For example, a certificate deployment operation depends on a successful certificate application or renewal, which can be considered a strong dependency, while a security scan operation can typically be executed after deployment or independently under abnormal circumstances, and can be considered a weak dependency. During the construction process, the system also checks for dependency loops. When circular dependencies are found, weak dependencies are adjusted or removed first, thus ensuring that the entire relational structure is a directed acyclic graph.

[0042] Once dependencies are determined, the system assesses the reversibility of each operation. Specifically, it quantifies the recoverability of an operation by calculating the ratio between the number of recoverable state elements after the operation is executed and the total number of state elements affected by the operation. This is further refined by considering factors such as the existence of a recovery path, the availability of resources required for recovery, and historical recovery success rates, thus classifying operations as fully reversible, partially reversible, or irreversible. For example, certificate deployment can usually be rolled back by restoring the original configuration, while certificate revocation is typically irreversible.

[0043] Subsequently, the system performs topological sorting based on the aforementioned dependency graph to obtain the operation execution order and identifies operations that can be executed in parallel. For example, when performing deployment operations on multiple servers, if there are no resource conflicts, they can be executed as parallel tasks, thereby improving overall execution efficiency.

[0044] During the execution phase, the system performs each operation in a predetermined order, backing up the current state before execution. The backup includes certificate files, private keys, configuration files, and service status. Simultaneously, compensation operations are verified before execution, such as checking for recovery permissions, the integrity of the backup data, and whether the service can start normally after recovery. When an operation fails, the system determines its impact scope based on dependencies and constructs an anomaly propagation path. The propagation factor is obtained by calculating the ratio between the number of affected nodes and the total number of nodes, and adjusting this ratio based on the propagation path depth.

[0045] Based on the propagation factor, the system selects an appropriate rollback strategy. For example, when the propagation factor is large, a full rollback can be performed to ensure system consistency; when the propagation factor is in the intermediate range, compensation is only performed on the affected operations and their downstream nodes; when the propagation factor is small, only the current operation is repaired, and subsequent operations are marked as pending verification. During the rollback process, the recovery order is determined by performing a reverse topological sort on the dependency graph, thereby ensuring the correctness of the recovery process.

[0046] After completing the certificate operations described above, the system further invokes security scanning tools to detect the target environment. Since different scanning tools have different output formats, this invention organizes the scan results using a unified structured description method and categorizes them according to the protocol layer, cryptographic layer, certificate layer, vulnerability layer, and compliance layer, thus forming a multi-dimensional security view. When multiple tools give different conclusions on the same detection item, a confidence-weighted mechanism is introduced for fusion, where the confidence level is calculated based on tool reliability, detection method strength, and time factors, thereby obtaining a unified evaluation result.

[0047] Based on this, the system can generate repair suggestions for specific environments and further generate executable commands. After user confirmation, the repair operation is executed, and a second test is performed after the repair is completed. By comparing the results before and after the repair, it is determined whether the problem has been resolved, thus forming a closed loop of automatic repair and verification.

[0048] Furthermore, throughout the execution process, the system manages the data generated at each step through a unified context management mechanism and establishes data associations between different steps using operation identifiers. Simultaneously, a chained structure can be used to record operation logs to ensure the traceability of the execution process and the integrity of the data.

[0049] Through the above implementation methods, the present invention can realize automatic orchestration of multiple operations, anomaly control, and secure closed-loop processing in complex certificate management scenarios, thereby significantly improving the automation level and system stability of SSL certificate management.

[0050] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A natural language-based intelligent lifecycle management method for SSL certificates, characterized in that: The method, during execution, uses a unified context manager to persistently manage and associate the context information generated in each step, including the following steps: S1: Receive natural language commands input by the user, and based on a classification system that includes seven types of certificate operation intentions such as application, renewal, revocation, deployment, query, scanning and configuration, perform intent recognition on the natural language commands through a pre-trained language model, and extract operation parameters according to the parameter patterns corresponding to the identified intent types; when the operation parameters are incomplete, complete them through multiple rounds of interaction with the user via a dialogue state machine. S2: Based on one or more intents identified in S1, dynamically construct a certificate operation dependency graph. This graph is a directed acyclic graph (DAG), where nodes represent certificate operations. Directed edges include strong and weak dependencies. Strong dependency means that if a preceding operation fails, subsequent operations cannot be executed; weak dependency means that if a preceding operation fails, subsequent operations are not blocked. Each operation node is labeled with a reversibility level, which is based on the ratio of the number of recoverable state elements after the operation to the total number of state elements affected by the operation. The system performs quantitative classification, including three levels: fully reversible, partially reversible, and irreversible; it generates an ordered operation plan through topological sorting and identifies operation groups that can be executed in parallel. S3: After presenting the operation plan to the user for confirmation, the operation plan is automatically executed according to the topological order. The execution process adopts a four-stage control mechanism of backup-pre-verification-execution-recovery: before executing the target operation, the current state is backed up and the feasibility of the compensation operation is pre-verified. When an anomaly occurs, the minimum rollback range is determined based on the reverse topological sorting of the certificate operation dependency graph and the state recovery is executed. Among them, when the anomaly occurs in a batch operation scenario involving multiple certificates, a full rollback strategy, a selective compensation strategy, or a minimum impact compensation strategy is selected according to the scope of the anomaly's impact. S4: After the operation is completed, the security scanning tool is invoked through a standardized description model that includes four dimensions: functional semantics, input mode, output mode and preconditions. The scanning results returned by the security scanning tool are then structured and mapped into five security domains: protocol layer, cryptographic layer, certificate layer, vulnerability layer and compliance layer. S5: Input the five-layer domain structured scan results into the large language model to perform a security assessment from four dimensions: protocol security, cryptographic suite strength, certificate configuration compliance, and known vulnerabilities. When multiple scanning tools produce different conclusions on the same security item, a confidence-weighted fusion model is used for adjudication. The confidence level is calculated based on three factors: tool reliability factor, method strength factor, and time decay factor. Finally, a diagnostic report containing executable remediation commands for different server types is generated.

2. The method for intelligent management of the entire lifecycle of SSL certificates based on natural language according to claim 1, characterized in that, In step S1, multi-round interactive completion is performed with the user through a dialogue state machine. Specifically, this includes: inferring the possible values ​​of missing parameters based on the current user's certificate operation history through contextual association, and verifying with the user in the form of a confirmation question; maintaining the slot filling state based on a finite state machine during the multi-round dialogue, wherein the states of the finite state machine include an initial state, an intent-recognized state, a parameter-filling state, and an intent-completed state, and triggering an intent-completed event when all necessary slots are filled; and recognizing implicit intents in user commands, automatically incorporating the implicit intent into the operation plan when a user command contains a pre-operation that is not explicitly stated but logically necessary.

3. The method for intelligent management of SSL certificates throughout their entire lifecycle based on natural language as described in claim 1, characterized in that, The selection of the rollback strategy in S3 specifically includes: calculating the propagation factor of the abnormal operation. ,in This is the set of downstream dependent nodes for abnormal operations. Let be the set of all nodes in the certificate operation dependency graph; when When the value exceeds a preset first threshold, a full rollback strategy is executed, and compensation operations are performed on all executed operations in the certificate operation dependency graph in reverse topological order; when When the value is between the second threshold and the first threshold, a selective compensation strategy is implemented, performing compensation only for abnormal operations and their downstream dependent operations; when... When the value is less than the second threshold, a minimum impact compensation strategy is executed, which only performs compensation operations on the abnormal operation itself, and its downstream operations are marked as pending verification.

4. The method for intelligent management of the entire lifecycle of SSL certificates based on natural language according to claim 1, characterized in that, The unified context manager specifically includes: constructing an operation audit chain using a Merkle chain structure, where each operation record contains operation content, timestamp, operator identifier, and hash value of the preceding record, and achieving the immutability of operation history through chained hash verification; storing the intermediate data and execution results generated by each step from S1 to S5 in the context storage in the form of key-value pairs, and establishing cross-step context associations through operation identifiers, so that subsequent steps can access the complete execution context of the preceding steps.

5. The method for intelligent management of the entire lifecycle of SSL certificates based on natural language according to claim 1, characterized in that, The S5 further includes an automatic repair verification loop: based on the security issues in the diagnostic report, an executable repair command is generated using a large language model for the specific server type and operating system; the repair command is automatically executed after user confirmation; After the repair is completed, a new security scan of the target server is triggered. The new scan results are compared and analyzed with the results before the repair to verify whether the security issues have been eliminated. The repair and verification steps are repeated until the security score reaches the preset qualified threshold or the maximum number of repair rounds is reached.

6. A natural language-based intelligent management system for the entire lifecycle of SSL certificates, characterized in that: include: The natural language understanding module is used to receive natural language commands input by the user. Based on a classification system that includes seven types of certificate operation intentions such as application, renewal, revocation, deployment, query, scanning and configuration, it performs intent recognition and parameter extraction through a pre-trained language model. When the parameters are incomplete, it completes them through multiple rounds of interaction with the user through a dialogue state machine. The operation orchestration module is used to dynamically construct a certificate operation dependency graph based on one or more intentions identified by the natural language understanding module. The certificate operation dependency graph is a directed acyclic graph, in which directed edges include two types: strong dependency and weak dependency. Each operation node is labeled with a reversibility level including fully reversible, partially reversible, and irreversible. An ordered operation plan is generated through topological sorting. The reversible execution engine module is used to execute the operation plan in topological order. It adopts a four-stage control mechanism of backup-pre-verification-execution-recovery. When an anomaly occurs, it determines the minimum rollback range based on the reverse topological sorting of the certificate operation dependency graph and performs state recovery. The security scanning integration module is used to call security scanning tools through a standardized description model that includes four dimensions: functional semantics, input mode, output mode, and preconditions, and to perform structured domain mapping of the scan results into five security domains: protocol layer, cryptographic layer, certificate layer, vulnerability layer, and compliance layer. The intelligent diagnostic module is used to input the five-layer domain structured scanning results into a large language model for security assessment. When multiple scanning tools produce different conclusions on the same security item, a confidence-weighted fusion model is used to make a decision and generate a diagnostic report containing executable repair commands. The unified context management module is used to persistently manage and associate the context information generated by each module.

7. The intelligent management system for the entire lifecycle of SSL certificates based on natural language as described in claim 6, characterized in that, The unified context management module includes a Merkle chain audit submodule, which records the immutable audit trail of all operations in a chain-like hash structure, wherein each audit record contains the operation content, timestamp, operator identifier and hash value of the preceding record.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the natural language-based intelligent lifecycle management method for SSL certificates as described in any one of claims 1 to 5.

9. An electronic device comprising a processor and a memory, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the natural language-based intelligent lifecycle management method for SSL certificates as described in any one of claims 1 to 5.

Images