Computers, systems, and methods for assisting in the design of network systems

Abstract

This system provides a computer that can assist in the design of network systems based on design specifications written in natural language. [Solution] The processor inputs a design document into a natural language processing program to obtain a set of descriptions directly or indirectly related to each entity, stores the obtained set of descriptions in a memory hierarchy in association with each entity, inputs the description related to the first entity into the natural language processing program to obtain the first configuration parameter of the first entity, and inputs the description related to the second entity and the first configuration parameter into the natural language processing program to obtain the second configuration parameter of the second entity.

Description

Technical Field

[0001] The present invention generally relates to a technology for assisting in the design of a network system.

Background Art

[0002] Since the design of a network system including a plurality of devices is complex, a design support technology is desired. Patent Document 1 is a document that discloses such a design support technology.

[0003] Patent Document 1 discloses, for example, "In the network design processing device 1, the network diagram creation processing unit 10 creates / edits a physical device network diagram composed of actually existing devices from an input using a computer screen, and represents one or a plurality of actually existing devices having the same use or function as virtual devices, creates / edits a virtual device network diagram composed of the virtual devices, and stores the configuration data of the physical device network diagram, the configuration data of the virtual device network diagram, and the relationship information of each of these configuration data in the network configuration data storage unit 15. The network configuration data output processing unit 16 switches and displays the physical device network diagram, the virtual device network diagram, and a network diagram obtained by overlapping them according to a designation from the outside."

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] <> In the technology described in Patent Document 1, assistance is provided with a network diagram as an input. Therefore, in the technology described in Patent Document 1, there is room for improvement in performing the design of a network system based on the text of a design document in which the design of the network system is described in natural language.

[0006] This invention was made in consideration of the above points, and aims to propose a computer and the like that can support the design of network systems based on design documents written in natural language. [Means for solving the problem]

[0007] To solve the above problems, the present invention provides a computer that assists in the design of a network system comprising a memory hierarchy and a processor, wherein the entities are devices or components within devices that are physically or virtually included in the computer, the memory hierarchy stores a design document in which the design contents of the network system are described in a plurality of descriptions expressed in natural language, and the processor: inputs the design document into a natural language processing program to obtain a set of descriptions that are directly or indirectly related to each entity, stores the obtained set of descriptions in the memory hierarchy in association with each entity, inputs the descriptions related to a first entity into the natural language processing program to obtain a first configuration parameter of the first entity, and inputs the descriptions related to a second entity and the first configuration parameter into the natural language processing program to obtain a second configuration parameter of the second entity, wherein the second entity is the same as or different from the first entity.

[0008] When determining the configuration parameters of a network system, some parameters can be directly determined from the information described in the design document (called independent parameters), while others cannot be determined without first determining the independent parameters (called first-order dependent parameters). In some cases, there are also parameters that cannot be determined without first determining the first-order dependent parameters (called second-order dependent parameters). According to the inventor's analysis, when a natural language processing program is asked to generate configuration parameters, it was found that instructing the program to generate dependent parameters without first determining the independent parameters tends to result in inaccurate results or no results being generated at all. Furthermore, it was found that this tendency becomes more pronounced as the order of the dependent parameters increases.

[0009] With the above configuration, while taking into account the above trends of natural language processing programs, it is possible to accurately generate a set of configuration parameters, including dependent parameters, from a design document written in natural language. [Effects of the Invention]

[0010] According to the present invention, it is possible to realize a computer that can assist in the design of a network system based on design documents written in natural language. Other problems, configurations, and effects will be clarified by the following description of embodiments. [Brief explanation of the drawing]

[0011] [Figure 1] This figure shows an example of a network detailed design support device according to the first embodiment. [Figure 2] This figure shows an example of a basic design document according to the first embodiment. [Figure 3] This figure shows an example of a parameter sheet according to the first embodiment. [Figure 4] This figure shows an example of a related information storage unit according to the first embodiment. [Figure 5] This figure shows an example of a related information storage unit according to the first embodiment. [Figure 6] This figure shows an example of a related information storage unit according to the first embodiment. [Figure 7] It is a diagram showing an example of a flowchart according to the first embodiment. [Figure 8A] It is a diagram showing an example of a flowchart according to the first embodiment. [Figure 8B] It is a diagram showing an example of a flowchart according to the first embodiment. [Figure 9] It is a diagram showing an example of a flowchart according to the first embodiment. [Figure 10] It is a diagram showing an example of a flowchart according to the first embodiment. [Figure 11] It is a diagram showing an example of a flowchart according to the first embodiment. [Figure 12] It is a diagram showing an example of a flowchart according to the first embodiment. [Figure 13] It is a diagram showing an example of a sequence diagram according to the first embodiment. [Figure 14] It is a diagram showing an example of a sequence diagram according to the second embodiment.

Mode for Carrying Out the Invention

[0012] (I) First Embodiment Hereinafter, an embodiment of the present invention will be described in detail. However, the present invention is not limited to the embodiment. With the shortage of IT (Information Technology) human resources in recent years, the shortage of network engineers has also become prominent. In addition, it has become difficult to have sufficient knowledge in the design processes of requirement definition, basic design, detailed design, and implementation construction for each network system design project. Therefore, there are concerns that the man-hours from the previous stage to the next stage increase, or that defects are created due to insufficient re-examination at each step, affecting the operation process after design.

[0013] The design of a network system, similar to the design of software, etc., proceeds in steps of requirement definition, basic design, detailed design, and implementation construction. After that, testing is conducted, and it enters the stage of operation. In requirement definition, the requirements to be satisfied are determined. In basic design, the operation of the entire network system is determined. In detailed design, the parameters to be set for each network switch are determined. Implementation construction is carried out based on the determined parameters. The design involves two steps: basic design and detailed design. Basic design is divided into physical design and logical design. In the logical design that shows the operation of the entire network system, the designer describes it by omitting physical network switches. On the other hand, in detailed design, the designer must determine the parameters to be set for each network switch. Therefore, when the designer progresses from basic design to detailed design, the designer must extract and summarize the relevant information regarding the omitted network switches.

[0014] The design of a network system is a field that requires knowledge and experience regarding networks. Amidst the shortage of IT human resources, various efforts are being made to handle projects. In this regard, devices and programs for automating the construction of network systems are known based on the information of physical design and logical design defined in basic design. However, in order to automate the construction of a network system, there are restrictions such as inputting physical design and logical design in a GUI or specifying the setting contents of network switches that constitute the output destination network system. It is still difficult to flexibly respond to a wide range of network system construction projects. Therefore, in this embodiment, a configuration is realized that deciphers the physical design and logical design texts described in natural language in basic design and automates the output of parameters for each network switch in detailed design.

[0015] In recent years, natural language processing programs known as generative AI (Artificial Intelligence), which are becoming increasingly common, use vast amounts of document data as training data and employ Large Language Models (LLMs) that have learned the relationships between hundreds of billions of word tokens to interpret natural language text input and produce output according to instructions. In this embodiment, for each category of network switch parameters to be compiled in the detailed design, the relevant information that needs to be deciphered from the basic design document and the method for compiling the parameters from that relevant information are specified, and the generation AI is utilized at each step. With this configuration, the text of the basic design document is interpreted and the network switch parameters for the detailed design are output.

[0016] In determining the parameters for the detailed design, some parameters are directly derived from the content described in the basic design document, while others are derived by deciphering one or more dependencies. LLM incorporates an attention mechanism that focuses not only on adjacent word tokens but also on highly relevant word tokens located at a distance, enabling accurate learning and inference of text. This allows learning and inference to be performed with minimal degradation of accuracy, even for texts of a certain length. However, attention is calculated based on computational complexity of the order of the square of the word token queue. Therefore, determining dependencies between chapters and sections at a distance in long texts such as basic design documents is difficult both in terms of computational complexity and accuracy. This tendency becomes even more pronounced as the degree of parameter dependencies increases. In this regard, the computer according to this embodiment, while taking into account the trends of LLM, sequentially deciphers dependencies from the basic design document written in natural language and accurately generates a set of parameters for the detailed design.

[0017] Next, embodiments of the present invention will be described with reference to the drawings. The following description and drawings are illustrative examples for illustrating the present invention, and have been omitted and simplified as appropriate for clarity of explanation. The present invention can also be carried out in various other forms. Unless otherwise specified, each component may be singular or plural. In the following description, identical elements in the drawings will be numbered the same, and their descriptions will be omitted as appropriate. The designations "First," "Second," "Third," etc., used in this specification are for identifying components and do not necessarily limit their number or order. Furthermore, the numbers used to identify components are used on a context-by-context basis, and a number used in one context does not necessarily indicate the same component in another context. Moreover, this does not prevent a component identified by one number from also performing the function of a component identified by another number.

[0018] In Figure 1, 100 represents a network detailed design support device (an example of a computer) according to the first embodiment. Figure 1 shows an example of the configuration of the network detailed design support device 100. The main input to the network detailed design support device 100 is the text of the basic design document 101. The main output to the network detailed design support device 100 is the parameter sheet 102 for each network switch in the detailed design. The configuration items for a network switch are parameters related to four categories: switch ports, virtual interfaces, routing, and others (hereinafter referred to as the four categories). These parameters constitute the main configuration. The output parameter sheet 102 is output for each of the four categories. The switch port parameters are responsible for configuring each physical port on the network switch. The virtual interface parameters are responsible for configuring VLAN (Virtual Local Area Network) interfaces and other interfaces that are virtually declared within the network switch. The routing settings are responsible for configuring packet transfer for the entire network switch, such as static routing and dynamic routing. Other settings include security settings and management settings, but these will not be explained in this embodiment.

[0019] The main functional configuration of the network detailed design support device 100 consists of an entity 110 that receives input from the basic design document 101 and an entity 120 that performs parameter design. Entity 110 extracts descriptions related to switch ports, virtual interfaces, routing, and other parameters from the basic design document 101, which is entered via the design document input unit 130, and aggregates them in the related information storage unit 140 for each of the four categories. Entity 120 summarizes parameter designs by referring to the related information storage unit 140 for each of the four categories, and aggregates them as parameter sheets 102 for each network switch. The parameter sheets 102 are output via the parameter aggregation unit 150. At this time, Entity 120 can also output the information from the related information storage unit 140 for each of the four categories as a parameter related information extraction list 103, and provide detailed feedback, the rationale for parameter design, etc., to users 104 such as designers and reviewers.

[0020] The related information storage unit 140 includes a switch port related information storage unit 141, a virtual interface related information storage unit 142, a routing related information storage unit 143, and a other configuration item related information storage unit 144. Furthermore, entity 110 may accept input for information extraction prompt patch 105 every four categories as an update to the operation of entity 110. Entity 120 may accept input for parameter design prompt patch 106 every four categories as an update to the operation of entity 120.

[0021] More specifically, entity 110 is a basic design document input receiving mechanism and includes a switch port-related description extraction unit 111, a virtual interface-related description extraction unit 112, a routing-related description extraction unit 113, and a other configuration item-related description extraction unit 114. The switch port-related description extraction unit 111 comprises a text interpretation unit 111A, an LLM inference mechanism 111B, and a switch port-related description extraction prompt list 111C. The virtual interface-related description extraction unit 112 comprises a text interpretation unit 112A, an LLM inference mechanism 112B, and a virtual interface-related description extraction prompt list 112C. The routing-related description extraction unit 113 comprises a text interpretation unit 113A, an LLM inference mechanism 113B, and a routing-related description extraction prompt list 113C. The other configuration item-related description extraction unit 114 comprises a text interpretation unit 114A, an LLM inference mechanism 114B, and a other configuration item-related description extraction prompt list 114C.

[0022] Entity 120 is a parameter design mechanism and consists of a switch port parameter design unit 121, a virtual interface parameter design unit 122, a routing parameter design unit 123, and an other configuration item parameter design unit 124. The switch port parameter design unit 121 includes a switch port information acquisition agent 121A, a parameter design unit 121B, an LLM inference mechanism 121C, and a parameter design prompt list 121D. The virtual interface parameter design unit 122 includes a virtual interface information acquisition agent 122A, a parameter design unit 122B, an LLM inference mechanism 122C, and a parameter design prompt list 122D. The routing parameter design unit 123 includes a routing information acquisition agent 123A, a parameter design unit 123B, an LLM inference mechanism 123C, and a parameter design prompt list 123D. The other setting item parameter design unit 124 includes an other setting item information acquisition agent 124A, a parameter design unit 124B, an LLM inference mechanism 124C, and a parameter design prompt list 124D.

[0023] The network detailed design support device 100 includes a CPU (Central Processing Unit) 161, a MEM (Memory) 162, a storage 163, and an IO (Input / Output) 164. The CPU 161 is a device that performs arithmetic processing. The CPU 161 may be a processor such as an MPU (Micro Processing Unit), GPU (Graphics Processing Unit), or AI chip. The MEM 162 is a device that stores programs, data, etc. The MEM 162 may be a main memory such as a ROM (Read Only Memory) or RAM (Random Access Memory). The storage 163 is an auxiliary storage device such as a hard disk drive, flash memory, SSD (Solid State Drive), or optical memory device. Programs, data, etc. stored in the storage 163 are read into the MEM 162 as needed. The IO 164 is an input device, output device, communication device, etc. The input device is a user interface that receives information from the user. The output device is a user interface that outputs various types of information (display output, audio output, print output, etc.). The communication device is a communication interface that communicates with other devices via a communication medium. The communication device can also function as an input device that receives information from other devices that are connected to it for communication. Furthermore, the communication device can also function as an output device that transmits information to other devices that are connected to it in a communication-enabled manner.

[0024] The functions of the network detailed design support device 100 (entities 110, 120, design document input unit 130, related information storage unit 140, parameter aggregation unit 150, etc.) may be implemented, for example, by the CPU 161 reading a program stored in the storage 163 into the MEM 162 and executing it (software), or by hardware such as a dedicated circuit, or by a combination of software and hardware. One function of the network detailed design support device 100 may be divided into multiple functions, or multiple functions may be combined into one function. Furthermore, some functions of the network detailed design support device 100 may be provided as separate functions or included in other functions. Also, some functions of the network detailed design support device 100 may be implemented by another computer that can communicate with the network detailed design support device 100. In addition, each hardware component of the network detailed design support device 100 may be one or multiple.

[0025] Figure 2 shows an example of a basic design document 101 for a network system. In this embodiment, we will explain using as an example a network connection in which multiple server systems within a site are connected to a WAN (Wide Area Network) using a redundant configuration with two L3 switches and two L2 switches.

[0026] Basic Design Document 101 is divided into a physical design, which shows the combination and connection of network switches, and a logical design, which shows the operation of the entire network system, and each is described separately. Based on the guidelines for the basic design in this embodiment, Basic Design Document 101 is divided into a total of six sections for both the physical design and the logical design, consisting of configuration, resource allocation, and high availability design.

[0027] Physical Design Configuration 201 lists the network switches. Physical Design Resource Allocation 202 describes the allocation policy for the use of each switch port of the network switches. Physical Design High Availability Design 203 describes the physical redundancy configuration and the allocation of Active or Standby. Logical design configuration 204 lists the VLAN interfaces, which are defined virtual interfaces. Logical design resource allocation 205 describes the allocation policy for IP addresses (Internet Protocol Addresses) and VLAN numbers for each VLAN interface. Logical design high availability design 206 describes the allocation policy for VLAN interface redundancy and route redundancy through dynamic routing.

[0028] In the basic design shown in this embodiment, considering scalability such as when the number of server systems increases, and operational ease such as aligning the VLAN numbers of the VLAN interfaces of the same function to the hundreds digit, the value may not necessarily be a specific value but may be specified by an allocation policy. Regarding the allocation policy for lower and upper numbered entries, the general rule is to allocate them in the order of WAN connection port, L3 switch, L2 switch, and server system, starting with the upper numbered entries, and then in the order of A-series and B-series, also starting with the upper numbered entries. Therefore, depending on the item, the allocation order may differ from lower to upper numbered entries. In this embodiment, the description is presented in natural language, but some explanations may include diagrams. In that case, a generative AI equipped with a multimodal function, which accepts not only natural language but also diagrams, would be used.

[0029] In this embodiment shown in Figure 2, for simplicity, the virtual interface is defined as a VLAN interface, and the routing is defined as dynamic routing using OSPF (Open Shortest Path First). In practice, it is expected that virtual interfaces will include GRE (Generic Routing Encapsulation), VPN (Virtual Private Network) tunnel interfaces, and loopback interfaces used for return traffic. Furthermore, in terms of routing, in addition to static routing, routing with other ASs (Autonomous Systems) such as BGP (Border Gateway Protocol) will also be handled depending on the requirements. However, in all cases, the procedure for translating the physical and logical designs in Basic Design Document 101 into detailed design parameters remains the same.

[0030] Figure 3 shows an example of a parameter sheet 102 for the detailed design of the network system in this embodiment. Parameter sheet 102 summarizes the parameter values ​​for four categories (switch ports, virtual interfaces, routing, and other configuration items) corresponding to the number of network switches. Switch port parameter 301 is a parameter related to the switch port. Virtual interface parameter 302 is a parameter related to the virtual interface. Routing parameter 303 is a parameter related to routing. Other configuration item parameter 304 is a parameter related to other configuration items. In this embodiment, the network detailed design support device 100 interprets the assembly of the devices shown in the physical design and the operation of the entire network system shown in the logical design of the basic design document 101 as shown in Figure 2, and designs the parameters of each network switch shown in Figure 3.

[0031] Figure 4 shows an example of the switch port-related information storage unit 141 in this embodiment. Entity 110 extracts and stores the necessary information from the basic design document 101 in Figure 2, associating it with the output of the required parameters in Figure 3. In this embodiment, it is illustrated in the form of a graph database, but the storage method is not limited. As shown in Figure 3, Entity 110 needs to consolidate the following information for all switch ports of the network switch in the switch port parameter 301: whether the port is enabled, whether it is a LAG (Link Aggregation Group), the mode, and the VLAN number to be associated with the port.

[0032] First, Entity 110 enumerates the port numbers of the switch ports of each network switch from the physical design configuration 201 of the basic design document 101. This can be determined from the specifications of the network switch. For each enumerated port number, Entity 110 associates the extracted usage results from the resource allocation 202 of the physical design of the basic design document 101. For the extracted usage, Entity 110 extracts the mode from the resource allocation 202 of the physical design, the presence or absence of LAG from the high availability design 203 of the physical design, the VLAN list from the logical design configuration 204, and the VLAN number from the resource allocation 205 of the logical design. If, for operational considerations, the allocation policy associates the VLAN number assigned to the VLAN interface with an IP address, etc., this information is also extracted and stored as Nth-order related information.

[0033] Figure 5 shows an example of the virtual interface-related information storage unit 142 in this embodiment. The storage method is the same as in Figure 4. In this embodiment, the virtual interface is limited to VLAN interfaces, and the L2 switch is described as not having a defined VLAN interface. As shown in Figure 3, in the virtual interface parameter 302, entity 110 needs to consolidate the IP address and segment for each VLAN number, the VRRP (Virtual Router Redundancy Protocol) number, the VRRP virtual IP address, and the VRRP priority for each network switch.

[0034] First, entity 110 lists the VLAN interfaces from the logical design configuration 204, and then lists the VLAN interfaces that match either system A or system B for each L3 switch. Entity 110 extracts information related to the VLAN number, IP address, and segment from the resource allocation 205 of the logical design for each enumerated VLAN interface. As described above, if the allocation policy associates the VLAN number, IP address, and segment, Entity 110 extracts and stores this information together as Nth-order related information. Furthermore, entity 110 extracts VRRP number and priority information from the logical design's high availability design 206. If necessary, entity 110 extracts virtual IP address information from the logical design's resource allocation 205, and Active or Standby system information from the physical design's high availability design 203. The Active and Standby information is used to determine the VRRP priority.

[0035] Figure 6 shows an example of the routing-related information storage unit 143 in this embodiment. The storage method is the same as in Figure 5. In this embodiment, one OSPF routing is defined for each L3 switch. As shown in Figure 3, for each network switch, entity 110 needs to consolidate the routing-related parameters 303, including the IP address to advertise, the VLAN number of the associated VLAN interface, whether it is Passive or not, and the cost.

[0036] In this embodiment, entity 110 first extracts information on VLAN interfaces and IP addresses associated with OSPF routing from the logical design's high availability design 206. Next, entity 110 extracts the IP address segment of the corresponding VLAN interface from the logical design's resource allocation 205. In parallel, entity 110 extracts the VLAN number of the corresponding VLAN interface from the logical design's resource allocation 205. Entity 110 also extracts passive information and cost information for each VLAN interface from the logical design's high availability design 206.

[0037] In this embodiment, the related information storage unit 140 in Figures 4 to 6 stores an excerpt of the text from the basic design document 101, but it may also be stored after being processed in a format that is easy to process or easy to review visually.

[0038] Figure 7 shows an example of a flowchart relating to the document interpretation unit 111A. As shown in this flowchart, the document interpretation unit 111A issues prompts to the LLM inference mechanism 111B to specify information extraction from the input basic design document 101, and stores the extracted relevant information in the switch port related information storage unit 141. The document interpretation unit 111A repeats this operation until all the information shown in the switch port related information storage unit 141 is available, or until no more relevant information can be confirmed. The prompts to be issued are stored in the switch port related description extraction prompt list 111C. Therefore, the operation of the flowchart can also be changed by the user 104 updating the switch port related description extraction prompt list 111C. The same applies to Figures 8A, 8B, and 9. In this embodiment, the flowchart shows the prompt operations from the basic design document 101 input in Figure 2 to the extraction of information from the switch port-related information storage unit 141 shown in Figure 4. The document interpretation unit 111A issues the information extraction command prompt as shown in Figure 4 to the LLM inference mechanism 111B, and executes the subsequent loops and prompts according to the result (steps S701 to S714).

[0039] In step S701, the document interpretation unit 111A receives the input of the basic design document 101 via the design document input unit 130 and inputs the received basic design document 101 to the LLM inference mechanism 111B. In step S702, the text interpretation unit 111A issues the prompt "List the network switches from chapter 201 of the physical design configuration" to the LLM inference unit 111B. In step S703, the text interpretation unit 111A performs the processing from steps S704 to S713 for each of the listed network switches.

[0040] In step S704, the text interpretation unit 111A issues the prompt "List the switch ports of the network switch from chapter 201 of the physical design configuration" to the LLM inference unit 111B. In step S705, the text interpretation unit 111A performs the processing from steps S706 to S712 for each of the listed switch port numbers.

[0041] In step S706, the text interpretation unit 111A issues the prompt "Extract the purpose of the relevant port number from chapter 202 of the physical design resource allocation" to the LLM inference unit 111B. In step S707, the text interpretation unit 111A replaces [Purpose] in the prompt with the answer obtained in step S706. The text interpretation unit 111A issues the prompt "Extract the description of the port mode for [Purpose] from chapter 202 of the physical design resource allocation" to the LLM inference mechanism 111B. In step S708, the text interpretation unit 111A issues the prompt "Extract the description related to the LAG of the port for [Application] from chapter 203 of the high availability design of the physical design" to the LLM inference unit 111B. In step S709, the text interpretation unit 111A issues the prompt "List the VLAN interfaces for [Purpose] from chapter 204 of the Logical Design Configuration" to the LLM inference unit 111B. In step S710, the text interpretation unit 111A replaces the [VLAN interface list] in the prompt with the answer obtained in step S709. The text interpretation unit 111A issues the prompt "Extract the description related to the VLAN number in the [VLAN interface list] from chapter 205 of the logical design resource allocation" to the LLM inference mechanism 111B.

[0042] In step S711, the text interpretation unit 111A determines whether or not additional information is needed to determine the VLAN number. At this time, the text interpretation unit 111A may issue a prompt for this determination to the LLM inference mechanism 111B. If the text interpretation unit 111A determines that additional information is needed, it proceeds to step S712; otherwise, it proceeds to step S713. In step S712, the text interpretation unit 111A issues the prompt "Extract the information for determining the VLAN number from the Logical Design chapter (Configuration 204, Resource Allocation 205, and High Availability Design 206)" to the LLM inference mechanism 111B. For example, the text interpretation unit 111A obtains information related to the VLAN number, IP address, and segment as information for determining the VLAN number.

[0043] In step S713, if the text interpretation unit 111A has finished processing each port number of the listed switch ports, it moves to step S714; otherwise, it moves to step S706. In step S714, if the text interpretation unit 111A has finished processing each of the listed network switches, it moves to step S715; otherwise, it moves to step S704. In step S715, the text interpretation unit 111A stores the information obtained through the processing in steps S702 to S714 in the switch port-related information storage unit 141 and terminates the process.

[0044] In this way, the document interpretation unit 111A extracts Nth-order related information from the basic design document 101 by considering the dependencies of the switch port parameters, extracting primary related information, and then extracting secondary related information belonging to that primary related information. For example, the document interpretation unit 111A extracts information step by step in the order of network switch (an example of primary related information), switch port number (an example of secondary related information), and purpose (an example of tertiary related information). Then, the document interpretation unit 112A extracts, for example, descriptions related to LAG, mode, and VLAN interface belonging to the purpose as quaternary related information, and extracts information for determining the VLAN number of the VLAN interface as quintal related information.

[0045] Figures 8A and 8B show examples of flowcharts related to the text interpretation unit 112A. The text interpretation unit 112A issues prompts for information extraction commands as shown in Figure 5 to the LLM inference mechanism 112B, and executes the subsequent loops and prompts according to the results (steps S801 to S819).

[0046] In step S801, the document interpretation unit 112A receives the input of the basic design document 101 via the design document input unit 130 and inputs the received basic design document 101 to the LLM inference mechanism 112B. In step S802, the text interpretation unit 112A issues the prompt "List the network switches from chapter 201 of the physical design configuration" to the LLM inference unit 112B. In step S803, the text interpretation unit 112A performs the processing from steps S804 to S817 for each of the listed network switches.

[0047] In step S804, the text interpretation unit 112A issues the prompt "List the VLAN interfaces from chapter 204 of the logical design configuration" to the LLM inference unit 112B. In step S805, the text interpretation unit 112A performs the processing from steps S806 to S816 for each of the listed VLAN interfaces.

[0048] In step S806, the text interpretation unit 112A issues the prompt "List the uses of the resources allocated to the relevant network switch from chapter 202 of the physical design resource allocation" to the LLM inference unit 112B. In step S807, the text interpretation unit 112A performs the processing from steps S808 to S815 for each of the listed uses.

[0049] In step S808, the text interpretation unit 112A issues the prompt "Extract the description of the IP address pool for the relevant purpose from chapter 205 of the Logical Design Resource Allocation" to the LLM inference unit 112B. In step S809, the text interpretation unit 112A issues the prompt "Extract the description of the IP address for the relevant use from chapter 205 of the logical design resource allocation" to the LLM inference unit 112B. In step S810, the text interpretation unit 112A issues the prompt "Extract the description of the VLAN number for the relevant use from chapter 205 of the logical design resource allocation" to the LLM inference unit 112B. In step S811, the text interpretation unit 112A determines whether any additional information is needed to determine the VLAN number. At this time, the text interpretation unit 112A may issue a prompt for this determination to the LLM inference mechanism 112B. If the text interpretation unit 112A determines that additional information is needed, it proceeds to step S812; otherwise, it proceeds to step S813.

[0050] In step S812, the text interpretation unit 112A issues the prompt "Extract the information for determining the VLAN number from the Logical Design chapter (Configuration 204, Resource Allocation 205, and High Availability Design 206)" to the LLM inference unit 112B. In step S813, the text interpretation unit 112A issues the prompt "Extract the description of the VRRP design (VRRP number, etc.) for the relevant application from chapter 206 of the Logical Design High Availability Design" to the LLM inference unit 112B. In step S814, the text interpretation unit 112A determines whether additional information is needed regarding whether the network switch in question is Active or Standby. At this time, the text interpretation unit 112A may issue a prompt for this determination to the LLM inference mechanism 112B. If the text interpretation unit 112A determines that it is needed, it proceeds to step S815; if it determines that it is not needed, it proceeds to step S816. In step S815, the text interpretation unit 112A issues the prompt "Extract the relevant information regarding whether the network switch in question is Active or Standby from chapter 203 of the physical design's high availability design" to the LLM inference unit 112B.

[0051] In step S816, if the text interpretation unit 112A has finished processing for each of the listed uses, it moves to step S817; otherwise, it moves to step S808. In step S817, if the document interpretation unit 112A has finished processing each of the listed VLAN interfaces, it moves to step S818; otherwise, it moves to step S806. In step S818, if the text interpretation unit 112A has finished processing each of the listed network switches, it moves to step S819; otherwise, it moves to step S804. In step S819, the text interpretation unit 112A stores the information obtained through the processing in steps S802 to S818 in the virtual interface-related information storage unit 142 and terminates the processing.

[0052] In this way, the document interpretation unit 112A extracts Nth-order related information from the basic design document 101 by considering the dependencies of the virtual interface parameters, extracting primary related information, and then extracting secondary related information belonging to that primary related information. For example, the document interpretation unit 112A extracts information step by step in the order of network switch (an example of primary related information), VLAN interface (an example of secondary related information), and purpose (an example of tertiary related information). Then, for example, the document interpretation unit 112A extracts information such as the description of the IP address belonging to the purpose and the description of the VLAN number as fourth-order related information, extracts information for determining the VLAN number as fifth-order related information, and extracts information for determining the priority of VRRP as sixth-order related information.

[0053] Figure 9 shows an example of a flowchart relating to the text interpretation unit 113A. The text interpretation unit 113A issues the information extraction command prompt as shown in Figure 6 to the LLM inference mechanism 113B, and executes the subsequent loops and prompts according to the result (steps S901 to S916).

[0054] In step S901, the text interpretation unit 113A receives the input of the basic design document 101 via the design document input unit 130 and inputs the received basic design document 101 to the LLM inference mechanism 113B. In step S902, the text interpretation unit 113A issues the prompt "List the network switches from chapter 201 of the physical design configuration" to the LLM inference unit 113B. In step S903, the text interpretation unit 113A performs the processing from steps S904 to S914 for each of the listed network switches.

[0055] In step S904, the text interpretation unit 113A issues the prompt "Extract information on whether OSPF routing is declared on the relevant network switch from chapter 206 of the Logical Design's High Availability Design" to the LLM inference unit 113B. In step S905, the text interpretation unit 113A determines whether or not an OSPF design exists. At this time, the text interpretation unit 113A may issue a prompt for this determination to the LLM inference mechanism 113B. If the text interpretation unit 113A determines that an OSPF design exists, it moves the process to step S906; if it determines that there is no OSPF design, it moves the process to step S915.

[0056] In step S906, the text interpretation unit 113A issues the prompt "List the uses of advertised IP addresses from chapter 206 of the Logical Design High Availability Design" to the LLM inference unit 113B. In step S907, the text interpretation unit 113A performs the processing from steps S908 to S913 depending on the purpose of the IP address to be advertised. In step S908, the text interpretation unit 113A issues the prompt "Extract the information on the IP address pool for the relevant purpose (description of the IP address pool and description of the IP address) from chapter 205 of the logical design resource allocation" to the LLM inference unit 113B. In step S909, the text interpretation unit 113A issues the prompt "Extract the description of the VLAN number for the relevant use from chapter 205 of the logical design resource allocation" to the LLM inference unit 113B. In step S910, the text interpretation unit 113A determines whether or not additional information is needed to determine the VLAN number. At this time, the text interpretation unit 113A may issue a prompt for this determination to the LLM inference mechanism 113B. If the text interpretation unit 113A determines that additional information is needed, it proceeds to step S911; otherwise, it proceeds to step S912.

[0057] In step S911, the text interpretation unit 113A issues the prompt "Extract the information for determining the VLAN number from the Logical Design chapter (Configuration 204, Resource Allocation 205, and High Availability Design 206)" to the LLM inference unit 113B. In step S912, the text interpretation unit 113A issues the prompt "Extract the description of the VRRP design (VRRP number, etc.) and cost for the relevant application from chapter 206 of the Logical Design High Availability Design" to the LLM inference unit 113B. In step S913, the text interpretation unit 113A issues the prompt "Extract the description of whether the VLAN interface for the relevant application is Passive from chapter 206 of the Logical Design's High Availability Design" to the LLM inference unit 113B.

[0058] In step S914, if the text interpretation unit 113A has finished processing each IP address to be advertised according to its purpose, it moves to step S915; otherwise, it moves to step S908. In step S915, if the text interpretation unit 113A has finished processing each of the listed network switches, it moves to step S916; otherwise, it moves to step S904. In step S916, the text interpretation unit 113A stores the information obtained from the processing in steps S902 to S915 in the routing-related information storage unit 143 and terminates the process.

[0059] In this way, the document interpretation unit 113A extracts Nth-order related information from the basic design document 101 by considering the dependencies of routing parameters, extracting primary related information, and then extracting secondary related information belonging to that primary related information. For example, the document interpretation unit 113A extracts information step by step in the order of network switch (an example of primary related information) and the purpose of the advertised IP address (an example of secondary related information). Then, for example, the document interpretation unit 113A extracts information to determine the IP address and VLAN number belonging to the purpose as tertiary related information, and extracts information to determine the presence or absence of Passive and the cost for each VLAN number as quaternary related information.

[0060] Figure 10 shows an example of a flowchart related to the parameter design unit 121B. The parameter design unit 121B issues a command prompt to the LLM inference mechanism 121C to acquire information on the switch port parameter 301, and executes the subsequent loops and prompts according to the result (steps S1001 to S1016).

[0061] In step S1001, the parameter design unit 121B receives an instruction to start parameter design. In step S1002, the parameter design unit 121B obtains a list of network switches via the switch port information acquisition agent 121A (from the information of the switch port-related information storage unit 141 acquired by the switch port information acquisition agent 121A). In step S1003, the parameter design unit 121B performs the processing from steps S1004 to S1014 for each network switch.

[0062] In step S1004, the parameter design unit 121B obtains the port numbers of the switch port list via the switch port information acquisition agent 121A. In step S1005, the parameter design unit 121B performs the processing from steps S1006 to S1013 for each port number of the switch port.

[0063] In step S1006, the parameter design unit 121B acquires application information via the switch port information acquisition agent 121A. In step S1007, the parameter design unit 121B replaces the [Purpose] in the prompt with the answer obtained in step S1006. The parameter design unit 121B issues the prompt "Determine whether to enable this switch port based on the following information. If enabling, write no shutdown; if not enabling, write shutdown. [Purpose]" to the LLM inference mechanism 121C. For example, although not shown in Figure 4, the purpose of a spare switch port may include information indicating "spare". This information allows for the determination of "shutdown" (whether or not to close the switch port). In step S1008, the parameter design unit 121B obtains VLAN number-related information, including Nth-order dependencies, via the switch port information acquisition agent 121A. In step S1009, the parameter design unit 121B replaces the [VLAN number related information] in the prompt with the answer obtained in step S1008. The parameter design unit 121B issues the prompt "List the VLAN numbers to associate with this switch port based on the following information: [VLAN number related information]" to the LLM inference mechanism 121C.

[0064] In step S1010, the parameter design unit 121B acquires mode-related information, including Nth-order dependencies, via the switch port information acquisition agent 121A. In step S1011, the parameter design unit 121B replaces the [Mode-related information] in the prompt with the answer obtained in step S1010. The parameter design unit 121B issues the prompt "Indicate whether the mode of this switch port is Access or Trunk based on the following information. [Mode-related information]" to the LLM inference unit 121C. In step S1012, the parameter design unit 121B acquires application and LAG-related information, including Nth-order dependencies, via the switch port information acquisition agent 121A. In step S1013, the parameter design unit 121B replaces the [Application and LAG related information] in the prompt with the answer obtained in step S1012. The parameter design unit 121B issues the prompt "Indicate whether to apply LAG to this switch port based on the following information. If LAG is applied, indicate all port numbers to include. [Application and LAG related information]" to the LLM inference unit 121C.

[0065] In step S1014, if the parameter design unit 121B has finished processing each port number of the listed switch ports, it moves to step S1015; otherwise, it moves to step S1006. In step S1015, if the parameter design unit 121B has finished processing each of the listed network switches, it moves to step S1016; otherwise, it moves to step S1004. In step S1016, the parameter design unit 121B outputs the information obtained from the processing in steps S1002 to S1015 to the parameter aggregation unit 150.

[0066] In this way, the parameter design unit 121B acquires Nth-order parameters based on the information stored in the switch port-related information storage unit 141, taking into account the dependencies of the switch port parameters, acquiring primary parameters, and then acquiring secondary parameters belonging to those primary parameters. For example, the parameter design unit 121B acquires information step by step in the order of network switch, switch port number (an example of a primary parameter), and purpose. Then, for example, the parameter design unit 121B acquires secondary parameters such as whether the port belonging to the purpose is enabled or disabled, whether LAG is present or disabled, and the mode, and acquires tertiary parameters such as the port number to be included in LAG and the VLAN number of the VLAN interface.

[0067] Figure 11 shows an example of a flowchart related to the parameter design unit 122B. The parameter design unit 122B issues a prompt to the LLM inference mechanism 122C for an instruction to acquire information on the virtual interface parameter 302, and executes the subsequent loops and prompts according to the result (steps S1101 to S1115).

[0068] In step S1101, the parameter design unit 122B receives an instruction to start parameter design. In step S1102, the parameter design unit 122B obtains a list of network switches via the virtual interface information acquisition agent 122A (from the information of the virtual interface-related information storage unit 142 acquired by the virtual interface information acquisition agent 122A). In step S1103, the parameter design unit 122B performs the processing from steps S1104 to S1113 for each network switch.

[0069] In step S1104, the parameter design unit 122B obtains a list of VLAN interfaces via the virtual interface information acquisition agent 122A. In step S1105, the parameter design unit 122B performs the processing from steps S1106 to S1112 for each VLAN interface.

[0070] In step S1106, the parameter design unit 122B obtains VLAN number-related information, including Nth-order dependencies, via the virtual interface information acquisition agent 122A. In step S1107, the parameter design unit 122B issues the prompt "Determine the VLAN number based on the following information [VLAN number related information]" to the LLM inference unit 122C. In step S1108, the parameter design unit 122B obtains IP address-related information, including Nth-order dependencies, via the virtual interface information acquisition agent 122A. In step S1109, the parameter design unit 122B issues the prompt "Based on the following information, determine the IP address in CIDR (Classless Inter-Domain Routing) notation [IP address related information]" to the LLM inference unit 122C.

[0071] In step S1110, the parameter design unit 122B acquires VRRP-related information, including Nth-order dependencies, via the virtual interface information acquisition agent 122A. In step S1111, the parameter design unit 122B issues the prompt "Determine whether VRRP design exists based on the following information. If VRRP design exists, determine the VRRP virtual IP address and VRRP priority [VRRP related information]" to the LLM inference unit 122C. In step S1112, the parameter design unit 122B sets the VRRP number to the same value as the VLAN number if VRRP is present.

[0072] In step S1113, if the parameter design unit 122B has finished processing each of the listed VLAN interfaces, it moves to step S1114; otherwise, it moves to step S1106. In step S1114, if the parameter design unit 122B has finished processing each of the listed network switches, it moves to step S1115; otherwise, it moves to step S1104. In step S1115, the parameter design unit 122B outputs the information obtained from the processing in steps S1102 to S1114 to the parameter aggregation unit 150.

[0073] In this way, the parameter design unit 122B acquires Nth-order parameters based on the information stored in the virtual interface-related information storage unit 142, taking into account the dependencies of the virtual interface parameters, acquiring primary parameters, and then acquiring secondary parameters belonging to those primary parameters. For example, the parameter design unit 122B acquires information step by step in the order of network switch and VLAN interface. Then, for example, the parameter design unit 122B acquires the VLAN number as a primary parameter, the IP address and segment for each VLAN number as secondary parameters, and if there is a VRRP design, it acquires the VRRP number, the VRRP virtual IP address, and the VRRP priority as tertiary parameters.

[0074] Figure 12 shows an example of a flowchart related to the parameter design unit 123B. The parameter design unit 123B issues a prompt to the LLM inference mechanism 123C for an instruction to acquire information on the routing parameter 303, and executes the subsequent loops and prompts according to the result (steps S1201 to S1213).

[0075] In step S1201, the parameter design unit 123B receives an instruction to start parameter design. In step S1202, the parameter design unit 123B obtains a list of network switches via the routing information acquisition agent 123A (from the information in the routing-related information storage unit 143 acquired by the routing information acquisition agent 123A). In step S1203, the parameter design unit 123B performs the processing from steps S1204 to S1211 for each network switch.

[0076] In step S1204, the parameter design unit 123B obtains whether or not routing settings are configured via the routing information acquisition agent 123A. In step S1205, the parameter design unit 123B determines whether or not routing settings exist. If the parameter design unit 123B determines that routing settings exist, it moves the process to step S1206; if it determines that there are no routing settings, it moves the process to step S1212. In step S1206, the parameter design unit 123B performs the processing from steps S1207 to S1210 for each routing setting.

[0077] In step S1207, the parameter design unit 123B obtains information related to the IP address to be advertised, including Nth-order dependencies, via the routing information acquisition agent 123A. In step S1208, the parameter design unit 123B issues the prompt "Determine the advertised IP address based on the following information [IP address related information]" to the LLM inference unit 123C. In step S1209, the parameter design unit 123B obtains information related to VLAN numbers for routing, including Nth-order dependencies, via the routing information acquisition agent 123A. In step S1210, the parameter design unit 123B issues the prompt "Based on the following information, determine a list of VLAN numbers related to routing, whether they are passive or not, and the cost [VLAN number related information]" to the LLM inference unit 123C.

[0078] In step S1211, if the parameter design unit 123B has finished processing each of the listed routing settings, it moves to step S1212; otherwise, it moves to step S1207. In step S1212, if the parameter design unit 123B has finished processing each of the listed network switches, it moves to step S1213; otherwise, it moves to step S1204. In step S1213, the parameter design unit 123B outputs the information obtained from the processing in steps S1202 to S1212 to the parameter aggregation unit 150. In this way, the parameter design unit 123B obtains Nth-order parameters based on the information in the routing-related information storage unit 143, taking into account the dependencies of routing parameters, obtaining primary parameters, and then obtaining secondary parameters belonging to those primary parameters. For example, the parameter design unit 123B obtains network switch and routing configuration information. Then, for example, the parameter design unit 123B obtains the advertised IP address and VLAN number as primary parameters, and the presence or absence of Passive status and cost for each VLAN number as secondary parameters.

[0079] The parameter aggregation unit 150 aggregates the parameters acquired by the parameter design units (121B to 124B) and outputs them as a parameter sheet 102 for each network switch. If the LLM inference mechanism determines that the switch port-related information storage unit 141 does not have the necessary information, it outputs "—" or the like to indicate NotAvailable for the corresponding item. As a supplement, the LLM inference mechanism (111B~114B, 121C~124C) may store the LLM model of the generated AI within the network detailed design support device 100, or it may issue an API (Application Programming Interface) for an LLM inference service provided by a public cloud or the like. If there are limitations on the number of input words due to the specifications of the LLM inference service, etc., preprocessing such as extracting relevant chapters from the basic design document 101 via prompts may be performed.

[0080] In this embodiment, the related information storage unit 140 stores excerpts of text from the basic design document 101, and the parameter design unit refers to the stored information and issues prompts to the LLM inference mechanism to consolidate the parameter values. If the information is stored in a structured state in the related information storage unit 140 using the functions or prompts of the document interpretation unit (111A~114A) of entity 110, then it is not necessarily required for entity 120 to issue prompts to the LLM inference mechanism. Furthermore, in this embodiment, the output of another entity may be referenced. For example, the text interpretation unit may reference N-th order related information from another related information storage unit 140. Also, for example, the parameter design unit may reference N-th order related information from another related information storage unit 140, or it may use parameters obtained by another parameter design unit.

[0081] Figure 13 shows an example sequence diagram illustrating the operation of the entire network system, including entities 110 and 120.

[0082] In entity operation 1310, as shown in the flowcharts in Figures 7 to 9, a user 104 such as a designer inputs the basic design document 101 (1311), and entity 110 extracts relevant information for each of the four categories (1312) and stores it in the relevant information storage unit 140 (1313). In entity operation 1320, as shown in the flowcharts in Figures 10 to 12, users 104 such as designers and reviewers start parameter design (1321), and entity 120 creates and outputs parameters based on the information stored in the related information storage unit 140 (1322, 1323, 1324). Furthermore, user 104 obtains the contents of the related information storage unit 140 in a form that can be interpreted in natural language (for example, the parameter-related information extraction list 103) as more detailed supporting information (1325). In this way, the black box created by the LLM inference mechanism is mitigated, and by confirming how entities 110 and 120 interpreted the data and recognized the parameters, it becomes possible to make concrete improvements, such as whether to revise the basic design document 101 or modify the prompts of the LLM inference mechanism, if the parameter output differs from what was expected.

[0083] According to this embodiment, the man-hours required to translate the basic design into detailed design can be reduced. Furthermore, by repeating the cycle of verifying the detailed design against the written basic design document, it is possible to avoid omissions or errors in the content of the basic design document and discrepancies with the detailed design.

[0084] (II) Second Embodiment Figure 14 is an example of a sequence diagram for updating the prompt list of entity 110 and the prompt list of entity 120. In the first embodiment, we described the parameter design for an example of configuration items when setting up a network system in a redundant configuration. However, in reality, a wider variety of settings may be required.

[0085] In the category of switch ports, the rules for counting port numbers may change if you are configuring a stack with multiple L3 switches. The virtual interface category is not limited to VLAN interfaces; definitions for tunnel interfaces, loopback interfaces, etc., are also made, and redundancy for multiple interfaces using vendor-specific protocols is applied. In the routing category, vendor-specific protocols may be used, or connections with other ASes may be established using BGP (Border Gateway Protocol). In the other configuration item category, items may be modified by vendor-specific settings.

[0086] Furthermore, while the above-described embodiments assume a physical network switch, when deploying a virtual network switch on a public cloud, IP address allocation depends on the allocation of subnet resources in the public cloud. Therefore, the resource allocation policies for both physical and logical resources in the basic design may differ, and the applicable methods for availability design may also differ.

[0087] As described above, the items to be described in the basic design document 101 and the parameters in the detailed design are determined by the design requirements of the network system. Therefore, users 104 such as administrators, designers, and reviewers respond to the prompts of entity 110 by applying the corresponding information extraction prompt patch 105 (1410). In addition, users 104 respond to the prompts of entity 120 by applying the corresponding parameter design prompt patch 106 (1420).

[0088] According to this embodiment, it is possible to accurately generate detailed design parameters by sequentially deciphering dependencies from the basic design document, in response to a variety of settings.

[0089] (III) Addendum In the above-described embodiment, the present invention was described in the case where it is applied to a network detailed design support device, but the present invention is not limited to this and can be broadly applied to various other systems, devices, methods, and programs.

[0090] In the embodiments described above, part or all of the program may be installed from the program source onto a device such as a computer that implements the network detailed design support device. The program source may be, for example, a program distribution server connected via a network or a recording medium readable by a computer (e.g., a non-temporary recording medium). Also, in the above description, two or more programs may be implemented as one program, or one program may be implemented as two or more programs. In the above-described embodiment, the output of information is not limited to display on a screen. The output of information may be audio output by a speaker, output to a file, printed on paper media by a printing device, projected onto a screen by a projector, or in any other form. In the above description, information such as programs, tables, and files that implement each function can be stored in memory, storage devices such as hard disks and SSDs (Solid State Drives), or recording media such as IC cards, SD cards, and DVDs.

[0091] The above-described embodiment has, for example, the following characteristic configuration. (1) A computer (e.g., network detailed design support device 100) that supports the design of a network system, having a memory hierarchy (MEM162, storage 163, etc.) and a processor (e.g., CPU161), and including multiple entities (entity 110, entity 120, switch port related description extraction unit 111, virtual interface related description extraction unit 112, routing related description extraction unit 113, other setting item related description extraction unit 114, etc.), wherein the entities are devices or components within the computer, physically or virtually included in the computer, the memory hierarchy stores a design document (e.g., basic design document 101) in which the design contents of the network system are described in multiple descriptions expressed in natural language, and the processor: inputs the design document into a natural language processing program (e.g., LLM inference mechanism) to directly or indirectly extract the descriptions related to each entity. A set of (switch port-related information, virtual interface-related information, routing-related information, other configuration item-related information, assignment policy, etc.) is obtained, and the obtained set of descriptions is associated with each entity and stored in the memory hierarchy. The description related to the first entity (for example, switch port-related description extraction unit 111) is input to a natural language processing program to obtain the first configuration parameter of the first entity (for example, switch port parameter 301). The description related to the second entity (switch port-related description extraction unit 111, virtual interface-related description extraction unit 112, etc.) and the first configuration parameter are input to a natural language processing program to obtain the second configuration parameter of the second entity (switch port parameter 301, virtual interface parameter 302, etc.), where the second entity is the same as or different from the first entity. With the above configuration, while taking into account the above trends of natural language processing programs, it is possible to accurately generate a set of configuration parameters, including dependent parameters, from a design document written in natural language.

[0092] (2) The above-mentioned computer, in order to obtain the set of above-mentioned descriptions, the processor: generates a first prompt for obtaining the first description of the design document relating to a first type of information (e.g., switch port) of the first entity, and inputs the first prompt to a natural language processing program to obtain the first description; generates a second prompt for obtaining the second description of the design document relating to a second type of information (e.g., application) relating to the information of the first description, and inputs the second prompt to a natural language processing program to obtain the second description. With the above configuration, by gradually narrowing down the information input to the natural language processing program, it becomes easier to obtain the expected description, and accuracy can be improved.

[0093] (3) In the above-mentioned computer, as for obtaining the first configuration parameter, the processor: generates a prompt to obtain a configuration parameter related to the first description (e.g., VLAN number), inputs the prompt to a natural language processing program to obtain the configuration parameter, generates a prompt to obtain a configuration parameter related to the second description (IP address and segment, VRRP number, VRRP virtual IP address, VRRP priority, etc.), inputs the prompt to a natural language processing program to obtain the configuration parameter. With the above configuration, by gradually narrowing down the information input to the natural language processing program, it becomes easier to obtain the desired parameters, thereby improving accuracy.

[0094] (4) The above-mentioned computer, wherein the processor: accepts prompt change information (e.g., information extraction prompt patch 105) for changing the prompt for obtaining the description of the design document related to each entity, and accepts prompt change information (e.g., parameter design prompt patch 106) for changing the prompt for obtaining the configuration parameters related to the description of the design document. With the above configuration, the prompt can be changed in accordance with the network system settings, so it is possible to obtain the design document description and generate configuration parameters in accordance with a variety of settings.

[0095] (5) The above-mentioned computer, wherein the processor outputs: the first configuration parameter and the second configuration parameter, and the above-mentioned description relating to the first entity and the above-mentioned description relating to the second entity (for example, the parameter-related information extraction list 103). With the above configuration, for example, if the output of the configuration parameters differs from what was expected, the user can implement specific improvements such as revising the design document or modifying the prompts of the natural language processing program.

[0096] (6) The above-mentioned computer includes a switch port entity relating to the configuration of the switch port of the network system (e.g., a switch port-related description extraction unit 111), a virtual interface entity relating to the configuration of the virtual interface of the network system (e.g., a virtual interface-related description extraction unit 112), and a routing entity relating to the configuration of the routing of the network system (e.g., a routing-related description extraction unit 113), wherein the processor: inputs the description of the design document relating to the switch port entity into a natural language processing program to obtain the configuration parameters of the switch port (e.g., switch port parameters 301), inputs the description of the design document relating to the virtual interface entity into a natural language processing program to obtain the configuration parameters of the virtual interface (e.g., virtual interface parameters 302), inputs the description of the design document relating to the routing entity into a natural language processing program to obtain the configuration parameters of the routing (e.g., routing parameters 303), and outputs parameter information (e.g., parameter sheet 102) including the configuration parameters of the switch port, the configuration parameters of the virtual interface, and the configuration parameters of the routing. With the above configuration, the configuration parameters for the switch port, the virtual interface, and the routing are output, allowing users to easily utilize the network switch parameters.

[0097] Furthermore, the above-described configuration may be modified, rearranged, combined, or omitted as appropriate, as long as it does not exceed the essence of the present invention. [Explanation of symbols]

[0098] 100...Network detailed design support device, 110...Entity, 120...Entity.

Claims

1. Memory hierarchy and Processor and A computer that has and assists in the design of a network system including multiple entities, The aforementioned entity is a device or component within the computer, either physically or virtually included in the computer. The aforementioned memory hierarchy stores design documents that describe the design details of the network system in multiple descriptions expressed in natural language. The aforementioned processor is: By inputting the design document into a natural language processing program, the set of descriptions directly or indirectly related to each entity is obtained. The acquired set of descriptions is associated with each entity and stored in the memory hierarchy. By inputting the description relating to the first entity into a natural language processing program, the first configuration parameter of the first entity is obtained. By inputting the description relating to the second entity and the first configuration parameter into a natural language processing program, the second configuration parameter of the second entity is obtained, where the second entity is the same as or different from the first entity. calculator.

2. A computer according to claim 1, As for obtaining the set of the above description, the processor is: A first prompt is generated to obtain the first description of the design document relating to the first type of information of the first entity, By inputting the first prompt into the natural language processing program, the first description is obtained, A second prompt is generated to obtain the second description of the design document relating to the second type of information relating to the information of the first description, By inputting the second prompt into the natural language processing program, the second description is obtained. calculator.

3. A computer according to claim 2, To obtain the first configuration parameter, the processor: A prompt is generated to obtain the configuration parameters related to the first description above. By inputting the aforementioned prompt into the natural language processing program, the configuration parameters are obtained, A prompt is generated to obtain the configuration parameters related to the second description above. By inputting the aforementioned prompt into the natural language processing program, the configuration parameters are obtained. calculator.

4. A computer according to claim 3, The aforementioned processor is: It accepts prompt change information to change the prompt for obtaining the description of the design document related to each entity, It accepts prompt change information to change the prompt for obtaining configuration parameters related to the description in the aforementioned design document. calculator.

5. A computer according to claim 1, The aforementioned processor is: Output the first configuration parameter and the second configuration parameter, and the description related to the first entity and the description related to the second entity. calculator.

6. A computer according to claim 1, This includes a switch port entity relating to the configuration of the switch ports of the network system, a virtual interface entity relating to the configuration of the virtual interfaces of the network system, and a routing entity relating to the configuration of the routing of the network system, The aforementioned processor is: By inputting the description of the design document related to the switch port entity into a natural language processing program, the configuration parameters of the switch port are obtained. By inputting the description of the design document related to the virtual interface entity into a natural language processing program, the configuration parameters of the virtual interface are obtained. By inputting the description of the design document related to the routing entity into a natural language processing program, the configuration parameters of the routing are obtained. The system outputs parameter information including the configuration parameters of the switch port, the configuration parameters of the virtual interface, and the configuration parameters of the routing. calculator.

7. A system including the computer described in claims 1 to 6.

8. A method for assisting in the design of a network system, Multiple entities are devices, whether physically or virtually, or components within such devices. The memory hierarchy stores design documents that describe the design details of the network system in multiple descriptions expressed in natural language. The processor inputs the design document into a natural language processing program to obtain a set of descriptions that are directly or indirectly related to each entity, The processor stores the acquired set of descriptions in the memory hierarchy, associating each entity with it. The processor obtains the first configuration parameter of the first entity by inputting the description relating to the first entity into a natural language processing program, The processor inputs the description relating to the second entity and the first configuration parameter to a natural language processing program, thereby obtaining the second configuration parameter of the second entity. The aforementioned second entity is either identical to or different from the aforementioned first entity. method.

9. A program that causes a computer to execute the method described in claim 8.

Images