Information processing device, system, processing method, and program

Abstract

This information processing device comprises: an execution processing unit that transmits, to a language model, an inquiry pertaining to a target network from an operator, and executes a network function designated in an instruction generated by the language model on the basis of the inquiry; and a response processing unit that acquires a response to the operator on the basis of a result obtained by the execution of the network function.

Description

Information processing device, system, processing method, and program

[0001] The present invention relates to the technical field of network management.

[0002] With the recent evolution of information technology, the use of the Internet and corporate networks has rapidly increased. As the scale of network systems expands, the complexity of network operation and management has also increased, making it important to improve their efficiency and sophistication.

[0003] In conventional network management, operators must manually select and operate multiple network functions, which always carries the risk of operational errors and delays. Network functions refer to individual software and systems used for network monitoring, configuration, troubleshooting, etc., and come in a wide variety of types and functions.

[0004] In network management, for example, when a failure occurs, it is necessary to identify the cause and quickly implement appropriate measures, but this is made difficult by compatibility issues and operational complexity between network functions provided by different vendors and platforms.In addition, to quickly respond to new security threats and technology updates, cooperation and automation of network functions is essential, but these network functions are often operated individually, making integrated management difficult to achieve.

[0005] OpenAI. "GPT-4 Technical Report." arXiv preprint arXiv:2303.08774 (2023).

[0006] As described above, the conventional technology requires an operator to manually select and operate multiple network functions, which is inefficient. Furthermore, the conventional technology executes network functions according to predefined procedures, which means it lacks the flexibility to respond to changes in network conditions.

[0007] The present invention has been made in view of the above points, and has an object to provide a technique for making network management more efficient and improving the flexibility of network management.

[0008] According to the disclosed technology, an information processing device is provided that includes an execution processing unit that sends an inquiry from an operator regarding a target network to a language model and executes a network function specified in an instruction generated by the language model based on the inquiry, and a response processing unit that obtains a response to the operator based on a result obtained by executing the network function.

[0009] The disclosed technology provides a technology for improving the efficiency and flexibility of network management.

[0010] It is a block diagram of a network function collaboration system. It is a flowchart showing the operation of the network function collaboration system. It is a block diagram of a network function collaboration device (information processing device). It is a diagram showing an example of the hardware configuration of the device.

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The embodiment described below is merely an example, and the embodiment to which the present invention is applied is not limited to the following embodiment.

[0012] The technology according to the present embodiment is a technology that uses a language model to automate the selection and invocation of network functions. Below, we will first explain in more detail the conventional technology and its problems, and then explain the technology according to the present embodiment.

[0013] (Regarding the Prior Art) A wide variety of technologies have traditionally been used for network operation and management. These technologies cover functions such as network monitoring, fault detection and response, and configuration changes. Large-scale networks, in particular, require efficient management of various network devices and software provided by different vendors.

[0014] Traditional network functions function as independent software and are specialized for specific tasks. Traditional network management requires manual coordination between these individual network functions, making integrated operation difficult. For example, if a network function that detects a fault and a network function that analyzes the cause of the fault are operated separately, operators must manually transfer information between these network functions and determine appropriate countermeasures. This manual coordination process reduces the efficiency of network management and can lead to errors.

[0015] Furthermore, conventional technologies have introduced a scenario-based automation approach. This scenario-based method automatically executes multiple network functions in a specific predefined order. However, because network conditions change dynamically, fixed scenarios are inflexible and difficult to respond to when undefined problems arise. Furthermore, scenarios must be redesigned every time a new network function is introduced or updated, making their management and maintenance time-consuming and costly.

[0016] That is, the prior art lacks the ability to flexibly and efficiently respond to the complexity and dynamic demands of networks.

[0017] (Problems) The main problems that the technology according to this embodiment aims to solve are the inefficiency and lack of flexibility in conventional network management. Specifically, the technology according to this embodiment focuses on the following three problems (1) to (3).

[0018] (1) Inefficiency of Manual System Interactions As described above, conventional network functions operate independently, and data and commands are transferred manually between different network functions. This manual process is not only time-consuming, but also carries the risk of problems caused by human error. In this embodiment, multiple network functions are automatically coordinated to improve the speed and accuracy of operations.

[0019] (2) Limited Response Capability Due to Fixed Scenarios Conventional scenario-based automation approaches execute network functions according to predefined procedures, but lack the flexibility to respond to changes in network conditions. This embodiment provides a function that analyzes network conditions in real time and dynamically selects and executes the most appropriate network functions as needed. This allows for quick and effective response to unexpected problems and changes in conditions.

[0020] (3) Difficulty in Updating the System When introducing a new network function or updating an existing network function, conventional technology requires manual redesign of many parts, including the scenario. This is a major factor that reduces operational efficiency. By using the technology according to this embodiment, system updates and expansions become easier, enabling sustainable operation.

[0021] By solving the above-mentioned problems, it is expected that the efficiency of network management will be improved, flexibility will be improved, and operational costs will be reduced. Note that it is not essential that all of the above-mentioned problems be solved. It is also possible that only some of the above-mentioned problems will be solved.

[0022] (Overview of Technology Relating to the Embodiments) Attempts to apply artificial intelligence technology to network management are increasing. The development of large language models (LLMs) is particularly noteworthy (Non-Patent Document 1). LLMs have the ability to learn from large amounts of text data and generate appropriate answers to questions posed in natural language. In the field of network management, the use of LLMs makes it possible to build a system that automatically recommends the appropriate selection and operation of network functions for complex situations and problems faced by operators. This system enables operators to centrally manage multiple network functions and respond quickly and accurately.

[0023] In this embodiment, the LLM is used to automatically select and invoke network functions, reducing the risks associated with traditional manual operations and improving the efficiency of network management. By communicating instructions in natural language to the LLM, the LLM automatically proposes the optimal combination and execution order of network functions, and the device according to this embodiment executes the network functions according to the instructions from the LLM. The technology according to this embodiment significantly improves the flexibility and accuracy of network operations, enabling the optimization of resource management across an organization.

[0024] In other words, in this embodiment, the LLM is used to automate the selection and operation of network functions, thereby improving the efficiency and sophistication of network management. This technology makes it possible to select and execute the optimal function based on the operator's natural language inquiry, thereby reducing operational errors and delays caused by manual operation.

[0025] In this embodiment, the LLM is used, but it is also possible to use a language model that is not called an LLM instead of the LLM. The LLM is an example of a language model. The configuration and operation of the system and device according to this embodiment will be described in detail below.

[0026] (System Configuration Example) Fig. 1 shows an example of the configuration of a network function linkage system according to this embodiment. Of the elements shown in Fig. 1, operator 60 is not included in the network function linkage system. For convenience of description, the network function linkage system may be simply referred to as the system (or this system).

[0027] As shown in Figure 1, the network function linkage system has a target network 10, network functions 21-23, databases 31-32, a network function linkage apparatus 40, and an LLM server 50. An overview of each component is as follows. The network functions 21-23 are collectively referred to as network function 20. The databases 31-32 are collectively referred to as database 30.

[0028] Although three network functions 20 and two databases 30 are shown in FIG. 1, the number of network functions 20 and the number of databases 30 are not limited to three and two, respectively.

[0029] The target network 10 is a network that is to be managed and monitored. The target network 10 may be, for example, an internal network of a company, a data center, or a cloud-based service. The status of the target network 10 is monitored and evaluated in real time through the network function 20.

[0030] The network functions 20 are components that perform specific functions or services in the target network 10, such as fault detection, traffic analysis, security monitoring, etc. The network functions 20 are individual software modules or systems that are used to manage different aspects of the target network 10.

[0031] As an example, the functions are divided so that network function 21 performs fault detection, network function 22 performs traffic analysis, and network function 23 performs security monitoring. However, this division of functions is an example, and one network function 20 may have multiple functions.

[0032] The database 30 is a component for storing information used by the network functions 20 and information used by the network function linkage device 40. Various data necessary for network operation, such as user data, configuration information, history data, and failure logs, is stored in the database 30.

[0033] The network function collaboration device 40 has the function of registering the type of network function 20, the type of information provided by the database 30, and the conditions for appropriately utilizing the network function 20 and the database 30 in the LLM server 50 (LLM).

[0034] The above conditions include the "network function 20 or database 30" that should be accessed in a specific situation, the "parameters or triggers" that should be used, and so on.

[0035] Through this process, the LLM server 50 can obtain the information necessary to select the most appropriate "network function 20 or database 30" in response to an inquiry received from the operator 60 and to determine an action based on that selection. When the network function linking device 40 receives specific instructions from the LLM server 50, it accesses the selected function 20 or database 30 according to the instructions and executes the specified task. The results of the task execution are sent to the LLM server 50, and the LLM server 50 uses the execution results to generate an integrated response to the operator 60.

[0036] The LLM server 50 is equipped with an LLM (Large Scale Language Model), analyzes the request of the operator 60 through natural language processing, and instructs the network function linking device 40 on the action to be taken. The LLM server 50 plays a central role in integrating the obtained information and creating a response to the operator 60. Note that, since the operation of the LLM server 50 is mainly the operation of the LLM, the LLM server 50 may be replaced with "LLM."

[0037] In this embodiment, the LLM may use any language, and the instructions given to the LLM and the responses received from the LLM may be in the same language, or the instructions given to the LLM and the responses received from the LLM may be in different languages.

[0038] (Processing Flow) A flowchart of the operation of the network function linking system is shown in Fig. 2. An overview of the operation of the network function linking system will be explained with reference to Fig. 2.

[0039] In S101, as a preliminary step, the network function linkage device 40 registers in the LLM server 50 information indicating the situation in which the network functions 20 and the information held in the database 30 should be utilized.

[0040] In S102, the operator 60 transmits an inquiry (such as a problem, request, or instruction) regarding the target network 10 to the network function linking device 40. The network function linking device 40 receives the inquiry from the operator 60. The operator 60 is, for example, a business operator that provides the target network 10. More specifically, the operator 60 is, for example, a terminal held by a person in charge of the business operator.

[0041] In S103, the network function linkage device 40 transmits the contents of the inquiry received in S102 to the LLM server 50.

[0042] In S104, the LLM server 50 analyzes the content of the inquiry received in S103 and returns specific instructions to the network function linkage device 40, such as activating an appropriate network function 20 or querying an appropriate database 30.

[0043] In S105, the network function linkage device 40 executes the specified network function 20 in accordance with the instruction from the LLM server 50, and queries the database 30 as necessary.

[0044] In S106, the network function collaboration device 40 collects the execution results of the executed network functions 20 and the query results from the database 30, and transmits these to the LLM server 50.

[0045] In S107, the LLM server 50 generates a final response to be provided to the operator 60 based on the result received in S106, and provides the generated response to the operator 60 via the network function linkage device 40.

[0046] In the above example, the response to the operator 60 is generated by the LLM server 50, but this is not limiting. For example, the network function linkage device 40 may collect the execution results of the executed network functions 20 and the query results from the database 30, and based on these, create (obtain) a response without using the LLM server 50, and send the created response to the operator 60.

[0047] (Detailed Processing) The processing content of each step from S101 to S107 will now be described in more detail.

[0048] <S101: Advance Preparation (Registration of Network Functions and Registration of Database Information)> In S101, the network function linkage device 40 registers, in the LLM server 50, information on the various network functions 20 and information on the various databases 20 required for network operation.

[0049] Specifically, the network function linkage device 40 registers information including the "type, functions provided, parameter settings, and execution status" for each of the multiple network functions 20 in the LLM server 50. The database 30 also includes user information, configuration data, and past operation history, and defines how this information will be used in conjunction with the network functions 20. The defined information is then registered in the LLM server 50.

[0050] The information registered in the LLM server 50 in S101 serves as the basis for the LLM in which the information is registered to effectively analyze inquiries from operators 60, select appropriate network functions 20, and instruct specific actions.

[0051] <S102: Receiving an inquiry from an operator> In S102, the operator 60 transmits an inquiry about the target network 10 or an instruction about the target network 10 to the network function collaboration device 40. The inquiry from the operator 60 comes in various forms, such as a failure report, a request for a setting change, or a performance monitoring data request.

[0052] <S103: Sending Inquiry Content to LLM Server> In S103, the network function linkage device 40 adds a system instruction to the inquiry content received from the operator 60, and sends the inquiry content with the system instruction added to the LLM server 50. The system instruction includes information such as the purpose for which the LLM should operate as a service and instructions for processing the inquiry. Note that there may be cases where the system instruction is not added.

[0053] An example of a system instruction is shown below. ---System Instruction Example--- [System Instruction] You are a system that combines multiple network functions to respond to operator requests. You must check the definitions of network functions and the information in the database. You must perform the appropriate function activation and information reading from the database to meet the operator's request. ---System Instruction Example Ends--- The Japanese translation of the above example is as follows: ---System Instruction Example in Japanese--- [System Instruction] You are a system that combines multiple network functions to respond to operator requests. You must check the definitions of network functions and the information in the database. You must activate the appropriate function and read information from the database to meet the operator's request. ---System Instruction Example in Japanese Ends--- <S104: Analysis and Instruction Generation by LLM Server> In S104, the LLM server 50 analyzes the inquiry content and system instruction sent from the network function linkage device 40. Through this analysis, the LLM server 50 formulates specific countermeasures, such as activating appropriate network functions 20, accessing necessary databases 30, or other action instructions. The instructions generated by the LLM server 50 are sent to the network function collaboration device 40.

[0054] The process of S104 may be realized, for example, by a function called Function Calling (Reference: OpenAI. "Function Calling", https: / / platform.openai.com / docs / guides / function-calling). Function Calling is a function that dynamically selects the most appropriate function based on an inquiry from the operator 60 and a system instruction. Note that using Function Calling is just one example. The process of S104 may also be realized without using Function Calling.

[0055] <S105: Execution of Network Functions 20 and Databases 30> In S105, the network function linkage device 40 activates the specified network functions 20 and executes queries to the specified databases 30 based on specific instructions received from the LLM server 50. The instructions, based on the analysis results by the LLM server 50, clearly indicate when and how to activate which network functions 20, and which databases 30 should be queried for which information.

[0056] The network function linkage device 40 performs the following operations (1) and (2), for example, in accordance with these instructions.

[0057] (1) Invoking a Network Function 20 Invoking a specific network function 20 to perform a specific network management task, such as detecting a fault, monitoring performance, and changing settings.

[0058] (2) Executing a Database Query A specific database 30 is accessed to collect necessary data, such as obtaining user information, analyzing log data, and updating setting information.

[0059] <S106: Collecting and Sending Results> In S106, the network function collaboration device 40 collects the results obtained from the executed network functions 20 and the results obtained from the database query. The information collected in S106 includes various information such as fault detection results, performance monitoring data, and configuration change confirmation results. The network function collaboration device 40 sends the collected data to the LLM server 50. The LLM server 50 uses this data to generate a consolidated response to provide to the operator 60.

[0060] <S107: Generating and Providing a Response to the Operator> In the final step, S107, the LLM server 50 generates a response to the operator 60 based on the result data received from the network function linkage device 40, and transmits the generated response to the operator 60 via the network function linkage device 40. In this process, the LLM server 50 comprehensively evaluates all of the information collected up to that point, and provides the operator 60 with a clear answer to the problem of the target network 10 inquired about by the operator 60, as well as the results of the requested action, via the network function linkage device 40. The creation of the response content and transmission of the response are described more specifically below.

[0061] Response Content Creation: The LLM server 50 generates a specific answer (or solution) based on both the results of the executed network functions 20 and information from the database. The answer may include a detailed cause analysis if a fault is detected, confirmation of configuration changes that have been made, or the results of an analysis of performance data.

[0062] Sending a response: The LLM server 50 organizes the generated response into a format that can be easily understood by the operator 60, and sends it to the network function collaboration device 40. The network function collaboration device 40 sends the organized response directly to the operator 60. The response that the operator 60 receives from the network function collaboration device 40 serves as a basis for the operator 60 to understand the current state of the target network 10 and take further action as necessary.

[0063] (Example of functional configuration of network function collaboration device 40) Fig. 3 shows an example of functional configuration of the network function collaboration device 40. As shown in Fig. 3, the network function collaboration device 40 includes a registration unit 41, an execution processing unit 42, and a response processing unit 43. The network function collaboration device 40 may also be called an information processing device.

[0064] The registration unit 41 executes the registration process of S101. The execution processing unit 42 executes S102 to S105. The response processing unit 43 executes S106 to S107.

[0065] For example, the execution processing unit 42 transmits an inquiry from an operator regarding the target network to the language model, and executes a network function specified in an instruction generated by the language model based on the inquiry. The response processing unit 43 obtains a response to the operator based on the result obtained by executing the network function.

[0066] (Hardware Configuration Example) Any of the devices described in this embodiment (network function linkage device, information processing device, LLM server, etc.) can be realized by, for example, having a computer execute a program. This computer may be a physical computer or a virtual machine on the cloud.

[0067] That is, the device can be realized by executing a program corresponding to the processing performed by the device using hardware resources such as a CPU and memory built into a computer. The program can be recorded on a computer-readable recording medium (such as a portable memory) and stored or distributed. The program can also be provided via a network such as the Internet or email.

[0068] Fig. 4 is a diagram showing an example of the hardware configuration of the computer. The computer in Fig. 4 includes a drive device 1000, an auxiliary storage device 1002, a memory device 1003, a CPU 1004, an interface device 1005, a display device 1006, an input device 1007, an output device 1008, and the like, all of which are interconnected via a bus B. The computer may further include a GPU.

[0069] The program that realizes the processing on the computer is provided by a recording medium 1001, such as a CD-ROM or a memory card. When the recording medium 1001 storing the program is set in the drive device 1000, the program is installed from the recording medium 1001 to the auxiliary storage device 1002 via the drive device 1000. However, the program does not necessarily have to be installed from the recording medium 1001, but may be downloaded from another computer via a network. The auxiliary storage device 1002 stores the installed program as well as necessary files, data, etc.

[0070] The memory device 1003 reads and stores a program from the auxiliary storage device 1002 when an instruction to start the program is received. The CPU 1004 realizes functions related to the device in accordance with the program stored in the memory device 1003. The interface device 1005 is used as an interface for connecting to a network, etc. The display device 1006 displays a GUI (Graphical User Interface) or the like according to the program. The input device 1007 is composed of a keyboard, mouse, buttons, a touch panel, etc., and is used to input various operation instructions. The output device 1008 outputs the results of calculations.

[0071] (Summary of Technology According to the Embodiment) As described above, the technology according to the embodiment makes it possible to realize automation of a network function collaboration system (herein referred to as the present system) that utilizes a large-scale language model (LLM). Specific points of the present system are as follows:

[0072] High degree of automation: LLM automates various network management processes, minimizing human intervention, allowing for faster and more accurate responses and improving the overall efficiency of the system.

[0073] Dynamic response capability: This system enables real-time data analysis and problem resolution. The optimal network function 20 is selected depending on the status of the target network 10, and necessary measures are immediately implemented, allowing for a response before problems escalate.

[0074] Scalability and adaptability: Even when a new network function 20 or a new database 30 is added, it can be properly integrated. This allows for flexible response to future technological advances.

[0075] Simultaneous execution of multiple actions: This system has the ability to simultaneously execute multiple network actions within a single transaction, enabling rapid processing even when complex operations are required.

[0076] Remote access and command execution: In addition to commands that can be executed directly from the network function linkage device 40, it is also possible to execute specific commands by remotely accessing another network device, thereby improving the range and flexibility of the system.

[0077] Saving and reusing conversation content and execution results: Past conversation content between the operator 60 and this system, and execution results of the network functions 20 are all saved in a storage device in the network function linkage device 40, and the operator 60 can provide feedback in natural language if the execution results differ from expectations. Based on this feedback and past data, the LLM server 50 can select the optimal network function 20 to be executed again.

[0078] (Advantages of the Technique According to the Embodiment) The advantages of the technique according to the present embodiment are as follows.

[0079] Rapid problem resolution: Because the LLM server 50 automates the execution of analysis and the generation of instructions, it is possible to quickly respond to problems or failures in the target network 10. The network function collaboration system in this embodiment processes inquiries in real time and can immediately activate the appropriate network functions 20, thereby significantly reducing downtime.

[0080] Improved accuracy: For example, Function Calling selects the most appropriate network function 20 or the most appropriate database 30 to access, thereby reducing human error and improving overall operational accuracy.

[0081] Reduction of operator burden: Many of the tasks that the operator 60 should perform are automated, thereby reducing the workload of the operator 60. This allows the operator 60 to focus on more strategic tasks, improving overall business efficiency.

[0082] Improved flexibility and scalability: Even when new network functions 20 or new databases 30 are added to the network function collaboration system, the LLM server 50 can automatically recognize and incorporate them. This improves the flexibility and scalability of the system and allows for future expansion.

[0083] Improved cost efficiency: Efficient network management and faster problem resolution reduce operational costs, while automation reduces errors, lowering long-term maintenance costs.

[0084] The following additional notes are provided regarding the above-described embodiments.

[0085] <Additional Notes> (Additional Item 1) An information processing device comprising: an execution processing unit that sends a query about a target network from an operator to a language model, and executes a network function specified in an instruction generated by the language model based on the query; and a response processing unit that obtains a response to the operator based on a result obtained by executing the network function. (Additional Item 2) The information processing device according to Additional Item 1, wherein the execution processing unit transmits the query and a system instruction including an instruction for processing the query to the language model. (Additional Item 3) The information processing device according to Additional Item 1, wherein the response processing unit transmits the result to the language model, obtains the response from the language model, and transmits the response to the operator. (Additional Item 4) The information processing device according to Additional Item 1, wherein the execution processing unit executes a query to a database specified in the instruction generated by the language model. (Supplementary Item 5) A system comprising a language model and an information processing device, wherein the information processing device comprises: an execution processing unit that sends an inquiry about a target network from an operator to the language model, and executes a network function specified in an instruction generated by the language model based on the inquiry, and a response processing unit that obtains a response to the operator based on a result obtained by executing the network function. (Supplementary Item 6) A processing method executed by an information processing device, comprising: a step of sending an inquiry about a target network from an operator to a language model, and executing a network function specified in an instruction generated by the language model based on the inquiry, and a step of obtaining a response to the operator based on a result obtained by executing the network function. (Supplementary Item 7) A non-transitory storage medium that stores a program for causing a computer to function as each unit in the information processing device described in any one of Supplementary Items 1 to 4.

[0086] Although the present embodiment has been described above, the present invention is not limited to such a specific embodiment, and various modifications and changes are possible within the scope of the gist of the present invention described in the claims.

[0087] 10 Target network 21 to 23 Network functions 31 to 32 Database 40 Network function cooperation device 41 Registration unit 42 Execution processing unit 43 Response processing unit 50 LLM server 60 Operator 1000 Drive device 1001 Recording medium 1002 Auxiliary storage device 1003 Memory device 1004 CPU 1005 Interface device 1006 Display device 1007 Input device 1008 Output device

Claims

1. An information processing device comprising: an execution processing unit that sends an inquiry from an operator regarding a target network to a language model, and executes a network function specified in an instruction generated by the language model based on the inquiry; and a response processing unit that obtains a response to the operator based on the result obtained by executing the network function.

2. The information processing device according to claim 1, wherein the execution processing unit transmits the query and a system instruction including an instruction for processing the query to the language model.

3. The information processing device according to claim 1, wherein the response processing unit transmits the result to the language model, obtains the response from the language model, and transmits the response to the operator.

4. The information processing device according to claim 1, wherein the execution processing unit executes a query to a database specified in an instruction generated by the language model.

5. A system comprising a language model and an information processing device, wherein the information processing device comprises: an execution processing unit that sends an inquiry from an operator regarding a target network to the language model and executes a network function specified in an instruction generated by the language model based on the inquiry; and a response processing unit that obtains a response to the operator based on a result obtained by executing the network function.

6. A processing method executed by an information processing device, comprising the steps of: sending an inquiry from an operator regarding a target network to a language model, and executing a network function specified in an instruction generated by the language model based on the inquiry; and obtaining a response to the operator based on a result obtained by executing the network function.

7. A program for causing a computer to function as each unit in the information processing device according to any one of claims 1 to 4.

Images