Programs, equipment, and systems to support supply chain risk assessment.

Abstract

This disclosure aims to provide programs, devices, and systems to support supply chain risk assessment. [Solution] This disclosure provides a program for supporting supply chain risk assessment, which causes one or more computers to perform the steps of: acquiring one or more risk keywords or risk-related questions or requests entered by a user; generating prompts to be entered into a large-scale language model based on the risk keywords or risk-related questions or requests; entering the prompts into the large-scale language model to output one or more risk scenarios; and presenting the output one or more scenarios to the user. A device and system having similar functions are also provided.

Description

Technical Field

[0001] The present disclosure relates to a program, an apparatus, and a system for assisting in the risk assessment of a supply chain. More specifically, the present disclosure relates to a program, an apparatus, and a system that input a prompt created based on a risk keyword, a question sentence, or a request sentence related to a risk into a large language model to generate one or more risk scenarios.

Background Art

[0002] A supply chain generally refers to a series of processes and activities until a product or service reaches a consumer. This process includes procurement of raw materials, production, logistics, distribution, sales, and final delivery to customers. A supply chain is a network in which multiple companies and businesses cooperate to function, and its purpose is to provide goods and services efficiently and effectively. A supply chain is an important factor that affects a company's competitiveness and customer satisfaction. Through efficient supply chain management, a company can achieve cost reduction, lead time shortening, risk management enhancement, and establish a competitive advantage in the market.

[0003] Risks that can cause disruptions in the supply chain include, for example, the risk of natural disasters such as earthquakes, typhoons, and floods interrupting the supply chain, the risk of political changes and economic crises affecting the supply chain, the risks affected by supplier bankruptcies, quality problems, and delivery delays, risks due to disruptions in the logistics network, shortages of transportation means, infrastructure failures, etc., and the risk of disruptions due to cyberattacks on the supply chain management system. By comprehensively implementing these risk assessments and countermeasures, it is necessary to enhance the resilience of the supply chain and prepare for unexpected disruptions.

[0004] Japanese Patent Publication No. 2019-86955 (Patent Document 1) discloses a supply chain management system that can quantitatively grasp the impact of an anomaly on a business when an anomaly occurs in the supply chain. More specifically, this supply chain management system is characterized by comprising: order storage means for storing orders related to the supply chain; anomaly information acquisition means for acquiring supply chain information including anomaly information that describes the content of anomalies that have occurred in the nodes and links constituting the supply chain; affected order information that describes the affected orders and their impact levels as described; and response information that describes the status of responding to the anomaly; anomaly information extraction means for extracting the anomaly information from the supply chain information; affected order information extraction means for extracting the affected order information from the supply chain information; related order information extraction means for extracting related order information of related orders related to the affected orders; impact level determination means for determining whether the impact level suffered by the affected orders and related orders exceeds a predetermined threshold; risk order extraction means for extracting the affected orders and related orders whose impact levels exceed the threshold as risk orders; and risk order display means for displaying the risk orders. This system, with the configuration described above, extracts orders affected by anomalies in the supply chain and determines the degree of impact of those anomalies on each order. Orders determined to be significantly affected are stored as risk orders, and related information about those risk orders is displayed on the display device.

[0005] Japanese Patent Publication No. 2017-211990 (Patent Document 2) discloses a modeling solution that includes a visual model of a supply chain network, including risks (uncertainties) associated with the supply chain network. The visual model disclosed in this document represents physical sites called logical stations as nodes and the physical transport of materials between physical sites called logical transits as links or paths. Furthermore, the modeling solution disclosed in this document obtains extended supply chain-related data, including supply and production data collected from each physical site, as well as location-related data (e.g., environmental, geopolitical, economic, and technical conditions of the physical site) obtained from external sources (e.g., external aggregators). The modeling solution then obtains this supply chain-related data to identify and assess the risks associated with each indicated physical site and the risks associated with the flow of materials on each path of the represented supply chain network. The modeling solution disclosed in this document presents identified risks at each logical station and logical transit that constitute the visual model, providing visibility into risks across the entire supply chain network (as well as specific physical sites and specific physical transports between physical sites that contribute to the risks). [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2019-86955 [Patent Document 2] Japanese Patent Publication No. 2017-211990 [Overview of the Initiative] [Problems that the invention aims to solve]

[0007] When assessing risks in a supply chain, it is necessary to assume hypothetical situations, making it difficult to reach a common understanding among stakeholders. Therefore, it would be beneficial to provide a simulation program, device, or system that quantitatively visualizes the scope and degree of impact, and supports risk assessment and countermeasures. [Means for solving the problem]

[0008] This disclosure provides programs, devices, and systems to support supply chain risk assessment. More specifically, this disclosure provides the following: [Embodiment 1] A program to support supply chain risk assessment, which is installed on one or more computers, A step of obtaining one or more risk keywords or risk-related questions or requests entered by the user, A step of generating prompts to input into a large-scale language model based on risk keywords or risk-related questions or requests, The steps include inputting a prompt into a large-scale language model to output one or more risk scenarios, A step to present one or more of the output scenarios to the user. A program that executes something. [Embodiment 2] The aforementioned one or more computers further include: A step that accepts the user's selection of one or more of the above scenarios and / or modification of their content. The program described in Embodiment 1, which causes the execution of the program. [Embodiment 3] The aforementioned one or more computers further include: A step to retrain the machine learning model based on the user's selection of one or more scenarios and / or modification of their content. The program described in Embodiment 2, which causes the execution of the program. [Embodiment 4] The aforementioned one or more computers further include: Based on the above scenario, the step of presenting one or more components that can be used in supply chain simulation. A program according to any one of Embodiments 1 to 3 that causes the following to be executed. [Embodiment 5] The aforementioned one or more computers further include: A step of accepting the user's selection and / or rearrangement of one or more of the above-mentioned parts. The program described in Embodiment 4, which causes the execution of the program. [Embodiment 6] The aforementioned one or more computers further include: A step in which a machine learning model is retrained based on the user's selection and / or rearrangement of one or more of the aforementioned components. The program described in Embodiment 5, which causes the execution of the program. [Embodiment 7] The aforementioned one or more computers further include: A step of presenting the user with one or more parameters that can be used in the simulation. A program according to any one of embodiments 1 to 6 that causes the following to be executed. [Embodiment 8] The aforementioned one or more computers further include: A step that accepts the user's selection and / or modification of one or more of the above parameters. The program described in Embodiment 7, which causes the execution of the program. [Embodiment 9] The aforementioned one or more computers further include: A step of retraining the machine learning model based on the user's selection and / or modification of one or more parameters. The program described in Embodiment 8, which causes the following to be executed. [Embodiment 10] The aforementioned one or more computers further include: Steps to perform a supply chain simulation A program according to any one of Embodiments 1 to 9 that causes the following to be executed. [Embodiment 11] The one or more computers are further caused to present the results of the simulation to the user The program according to Embodiment 10, which causes the above to be executed. [Embodiment 12] The program according to Embodiment 11, wherein the results of the simulation include at least one of the expected values corresponding to production volume, production volume per unit time, product inventory volume, transportation volume per unit time, transportation time, transportation distance, transportation cost, and transportation route, consumption volume, consumption volume per unit time, inventory disposal volume, inventory disposal rate, inventory disposal cost, and occurrence frequency (probability). [Embodiment 13] The one or more computers are further caused to save and / or read scenarios The program according to any one of Embodiments 1 to 12, which causes the above to be executed. [Embodiment 14] The one or more computers are further caused to select a large language model (LLM) The program according to any one of Embodiments 1 to 13, which causes the above to be executed. [Embodiment 15] The program according to any one of Embodiments 1 to 14, wherein RAG (Retrieval-Augmented Generation) is used for prompt generation. [Embodiment 16] The one or more computers are further caused to update the database (DB) used for RAG The program according to Embodiment 15, which causes the above to be executed. [Embodiment 17] An apparatus for assisting in risk assessment of a supply chain, which has a query input unit that obtains one or more risk keywords, a question sentence, or a request sentence related to risk input by a user, A prompt generation unit that generates prompts to be input into a large-scale language model based on risk keywords or risk-related questions or requests, A risk scenario generation unit that inputs a prompt into a large-scale language model and outputs one or more risk scenarios, Scenario presentation unit that presents one or more outputted scenarios to the user. A device including a device. [Embodiment 18] The aforementioned device further, A scenario operation unit that accepts the user's selection of one or more scenarios and / or modification of their contents. The apparatus according to Embodiment 17, including the apparatus described above. [Embodiment 19] The aforementioned device further, Scenario retraining unit that retrains the machine learning model based on the user's selection of one or more scenarios and / or modification of their content. The apparatus according to Embodiment 18, including the apparatus described above. [Embodiment 20] The aforementioned device further, A component presentation unit presents one or more components that can be used in supply chain simulations based on the aforementioned scenario. The apparatus according to any one of embodiments 17 to 19, including the apparatus described in any one of embodiments 17 to 19. [Embodiment 21] The aforementioned device further, A component operation unit that accepts the user's selection and / or rearrangement of one or more of the aforementioned components. The apparatus according to Embodiment 20, including the apparatus described above. [Embodiment 22] The aforementioned device further, A component retraining unit that retrains a machine learning model based on the user's selection and / or rearrangement of one or more components. The apparatus according to Embodiment 21, including the apparatus described above. [Embodiment 23] The aforementioned device further, A parameter presentation unit that presents the user with one or more parameters that can be used in the simulation. The apparatus according to any one of embodiments 17 to 22, including the apparatus described in any one of embodiments 17 to 22. [Embodiment 24] The aforementioned device further, A parameter operation unit that accepts the selection and / or modification of one or more parameters by the user. The apparatus according to Embodiment 23, including the apparatus described above. [Embodiment 25] The aforementioned device further, A parameter relearning unit that retrains the machine learning model based on the user's selection and / or modification of one or more parameters. The apparatus according to Embodiment 24, including the apparatus described above. [Embodiment 26] The aforementioned device further, Simulation execution unit that performs supply chain simulations. The apparatus according to any one of embodiments 17 to 25, including the apparatus described in any one of embodiments 17 to 25. [Embodiment 27] The aforementioned device further, Simulation result presentation unit that presents the results of the aforementioned simulation to the user. The apparatus according to embodiment 26, including the apparatus described above. [Embodiment 28] The results of the above simulation correspond to the expected values ​​of production volume, production volume per unit time, product inventory volume, transport volume per unit time, transport time, transport distance, transport cost, and transport route, consumption volume, consumption volume per unit time, inventory waste volume, inventory waste rate, inventory waste cost, and frequency (probability) of occurrence. The apparatus according to embodiment 27, comprising at least one of the following. [Embodiment 29] The aforementioned device further, Scenario input / output unit for saving and / or loading scenarios. The apparatus according to any one of embodiments 17 to 28, including the apparatus described in any one of embodiments 17 to 28. [Embodiment 30] The aforementioned device further, LLM selection unit that selects a Large-Scale Language Model (LLM) The apparatus according to any one of embodiments 17 to 29, including the apparatus described in any one of embodiments 17 to 29. [Embodiment 31] The apparatus according to any one of embodiments 17 to 30, wherein RAG (Retrieval-Augmented Generation) is used to generate prompts. [Embodiment 32] The aforementioned device further, Database (DB) update unit that updates the database used by RAG The apparatus according to Embodiment 31, including the apparatus described above. [Embodiment 33] A system for supporting supply chain risk assessment, comprising one or more computers, The one or more computers include memory and a processor connected to the memory. The aforementioned processor, A step of obtaining one or more risk keywords or risk-related questions or requests entered by the user, A step of generating prompts to input into a large-scale language model based on risk keywords or risk-related questions or requests, The steps include inputting a prompt into a large-scale language model to output one or more risk scenarios, A step to present one or more of the output scenarios to the user. A system that executes this. [Embodiment 34] The aforementioned processor further, A step that accepts the user's selection of one or more of the above scenarios and / or modification of their content. The system described in embodiment 33, which performs the following actions. [Embodiment 35] The aforementioned processor further, A step to retrain the machine learning model based on the user's selection of one or more scenarios and / or modification of their content. The system described in embodiment 34, which performs the following actions. [Embodiment 36] The aforementioned processor further, Based on the above scenario, the step of presenting one or more components that can be used in supply chain simulation. A system according to any one of embodiments 33 to 35 that performs the following: [Embodiment 37] The aforementioned processor further, A step of accepting the user's selection and / or rearrangement of one or more of the above-mentioned parts. The system according to embodiment 36, which performs the following actions. [Embodiment 38] The aforementioned processor further, A step in which a machine learning model is retrained based on the user's selection and / or rearrangement of one or more of the aforementioned components. The system according to embodiment 37, which performs the following. [Embodiment 39] The aforementioned processor further, A step of presenting the user with one or more parameters that can be used in the simulation. A system according to any one of embodiments 33 to 38 that performs the following: [Embodiment 40] The aforementioned processor further, A step that accepts the user's selection and / or modification of one or more of the above parameters. The system according to embodiment 39, which performs the following actions. [Embodiment 41] The aforementioned processor further, A step of retraining the machine learning model based on the user's selection and / or modification of one or more parameters. The system described in embodiment 40, which performs the following actions. [Embodiment 42] The aforementioned processor further, Steps to perform a supply chain simulation A system according to any one of embodiments 33 to 41, which performs the following actions. [Embodiment 43] The aforementioned processor further, Steps to present the results of the simulation to the user. The system according to embodiment 42, which performs the following actions. [Embodiment 44] The system according to Embodiment 43, wherein the results of the simulation include at least one of the expected values ​​corresponding to production volume, production volume per unit time, product inventory volume, transport volume per unit time, transport time, transport distance, transport cost, and transport route, consumption volume, consumption volume per unit time, inventory waste volume, inventory waste rate, inventory waste cost, and frequency (probability). [Embodiment 45] The aforementioned processor further, Steps to save and / or load a scenario A system according to any one of embodiments 33 to 44, which performs the following actions. [Embodiment 46] The aforementioned processor further, Steps to select a Large-Scale Language Model (LLM) A system according to any one of embodiments 33 to 45 that performs the following: [Embodiment 47] A system according to any one of embodiments 33 to 46, wherein RAG (Retrieval-Augmented Generation) is used to generate prompts. [Embodiment 48] The aforementioned processor further, Steps to update the database (DB) used by RAG The system described in embodiment 47, which performs the following actions. [Embodiment 49] A non-temporary computer-readable recording medium storing the program described in any one of Embodiments 1 to 16. [Brief explanation of the drawing]

[0009] [Figure 1] Figure 1 shows an example of scenario generation using RAG that may be used in the program of this disclosure. [Figure 2] Figure 2 shows an example of scenario generation based on a query. [Figure 3] Figure 3 shows an example of scenario generation in a chat format. [Figure 4] Figure 4 shows an exemplary risk scenario screen that may be used in the program of this disclosure. [Figure 5] Figure 5 shows an exemplary flowchart of the scenario generation process in the program of this disclosure. [Figure 6] Figure 6 shows an exemplary creation palette and a model creation panel consisting of parts boxes that may be used in the program of this disclosure. [Figure 7] Figure 7 shows an exemplary simulation execution screen that may be used in the program of this disclosure. [Figure 8] Figure 8 shows an exemplary simulation results screen that may be used in the program of this disclosure. [Figure 9] Figure 9 shows an exemplary program logic structure of the program of this disclosure. [Figure 10] Figure 10 shows an exemplary flowchart of the program of this disclosure. [Figure 11] Figure 11 shows a schematic configuration of an exemplary computer that may be used in an implementation of the present disclosure. The exemplary computer (100) includes a control unit (101), a storage unit (102), a peripheral device interface unit (103), an input unit (104), a display unit (105), a communication unit (106), and a bus (110). The computer (100) may connect to an external server (130) and a database (140) via a network (120). [Modes for carrying out the invention]

[0010] Programs to support supply chain risk assessment This disclosure relates, in one aspect, to a program for supporting supply chain risk assessment.

[0011] In some embodiments, the programs of this disclosure can help assess the risk of an incident occurring in a supply chain from a manufacturer of a substance (e.g., a pharmaceutical product) through intermediaries and retailers (including healthcare facilities) to the end consumer, in terms of the magnitude (represented by the scope and monetary value) and frequency (probability) of the threat of a shortage of the product in the supply chain.

[0012] As used herein, "supply chain" refers to the entire process from the procurement of raw materials to the manufacturing, distribution, and final sale to the consumer. Within the supply chain, the aim is to improve the quality of life for citizens through cost reduction, shorter lead times, increased customer satisfaction, and stable social infrastructure, by enhancing cooperation and efficiency among companies. In the context of this specification, a supply chain could, for example, be a pharmaceutical supply chain.

[0013] The main elements of a supply chain are procurement, manufacturing, distribution, and retail. Procurement is the process of purchasing raw materials and components from suppliers in the required quality and quantity. Manufacturing is the process of producing products using the procured materials. Distribution is the process of transporting finished products from warehouses and distribution centers to customers and retailers. Retail is the stage where products are sold to the end consumer. Each of these stages is closely linked to the flow of information, physical products, and money. Effective supply chain management streamlines these flows and provides companies with the flexibility to respond quickly to unforeseen circumstances and fluctuations in demand. Optimizing the supply chain can reduce overall operating costs and enhance competitiveness in the market.

[0014] In the context of this disclosure, “risk assessment” includes, for example, evaluating the risk of a product being out of stock in the supply chain when an incident affecting the supply chain occurs (e.g., natural disaster, political instability, supplier bankruptcy, technical failure, market fluctuations, etc.), expressed in terms of the magnitude (scale, represented by the scope and monetary value) and frequency (probability) of the threat. The program of the present invention can provide users with information to support such risk assessment.

[0015] The program described in this disclosure includes, for example, a function to simulate changes in inventory levels using manufacturing lead time, initial inventory levels at each layer, etc. as variables. Key functions that can be implemented in the program described in this disclosure include, for example, a function to run a simulation by selecting an inventory definition from a list of options (for example, three months' worth of the previous year's average inventory, three months' worth of the average of the most recent three months, etc.; the method for specifying normal inventory levels at each layer can be selected according to pre-prepared definitions), a risk scenario generation function (providing examples of probability and scope of impact, presenting risks for each category (classifying categories, breaking down causes, and calculating frequency of occurrence), displaying frequency of occurrence and impact from past cases (Fault Tree Analysis, etc.), a function to formulate from scenarios (simulation model generation), a function to connect building blocks at each layer (from manufacturer to medical institution) to create a supply chain network diagram, information feedback from downstream to upstream, etc.), and a risk simulation function (creating mathematical formulas (simulation models)). The program may include functions such as performing calculations based on the provided data, selecting an inventory definition from a list of options, and running a simulation, as well as user interface functions (risk keyword input, scenario selection function, calculation parameter input, simulation execution, stop button, etc.) and visualization functions (displaying the inventory quantity of each layer in a time-series graph, dynamically (animating) displaying fluctuations in inventory quantity based on layer warehouses mapped on a map, and indicating the scope of risk impact with color, etc.). In some embodiments, the program of this disclosure can configure the structure of the supply chain according to risk scenarios, for example, by visualizing the time-dependent changes in inventory according to risk, thereby supporting discussions among stakeholders regarding inventory levels, or by mapping transportation volumes and routes on a map, thereby supporting discussions among stakeholders regarding alternative transportation routes.

[0016] In some embodiments, the program of this disclosure may be a program that causes one or more computers to perform the steps of: obtaining one or more risk keywords or risk-related questions or requests entered by a user; generating prompts to be entered into a large-scale language model based on the risk keywords or risk-related questions or requests; entering the prompts into the large-scale language model to output one or more risk scenarios; and presenting the output one or more scenarios to the user. In this specification, "risk keywords or risk-related questions or requests" may be simply referred to as "queries".

[0017] In some embodiments, the program of the Disclosure may be a program that causes one or more computers to perform the step of obtaining one or more risk keywords or risk-related questions or requests entered by a user. User input may be, for example, keyboard input, voice input, or data input such as text files or images. One or more computers running the program of the Disclosure are equipped with functions for obtaining these inputs.

[0018] In the context of this disclosure, examples of “risk keywords or risk-related questions or requests” include the following: Example 1: "Create a risk scenario in which inventory shortages occur due to an incorrect demand forecast for product A." Example 2: "Create a scenario in which a large-scale earthquake occurs and affects treatments for rare diseases." Example 3: "Create a scenario for when a wholesaler is subjected to a cyberattack and its core systems go down."

[0019] In some embodiments, the program of the present disclosure may be a program that causes one or more computers to perform the step of generating prompts to be input into a Large Language Model (LLM) based on risk keywords or risk-related questions or requests.

[0020] In the context of this disclosure, examples of large-scale language models (LLMs) include GPT (Generative Pre-trained Transformer), BERT (Bidirectional Encoder Representations from Transformers), T5 (Text-to-Text Transfer Transformer), and Llama (Large Language Model Meta AI). As used herein, “Large-scale language model (LLM)” refers to an AI model that can learn from vast amounts of text data and generate human language. These models are widely used in the field of natural language processing (NLP) and are utilized for a wide range of tasks, including text summarization, translation, question answering, and text generation. LLMs are typically trained using datasets containing billions of words collected from books, articles, websites, and other sources available on the internet. These models usually employ a neural network architecture called a multi-layer transformer, which is particularly well-suited for processing contextually rich information. Furthermore, trained models can be adapted to various language-related tasks by being fine-tuned for specific tasks.

[0021] In some embodiments, the LLM used in the program of this disclosure may be fine-tuned for creating risk scenarios. Fine-tuning may involve using risk information related to the supply chain from sources such as past incidents related to the supply chain, internal knowledge and failure records accumulated within the organization, publicly available failure knowledge databases (https: / / www.shippai.org / fkd / index.php), and general news. Retraining the model using the collected information (dataset) can improve performance for the desired task (creating risk scenarios). Furthermore, in some embodiments, the LLM used in the program of this disclosure may be transfer-learned for creating risk scenarios. In transfer learning, the model is retrained for a new task using the weights of a pre-trained model as initial values. Transfer learning can also be considered part of fine-tuning.

[0022] As used herein, a “prompt” input to a Large-Scale Language Model (LLM) refers to text input provided to the model. This prompt serves as a starting point or instruction for the model to generate a response, determining what kind of text to generate or what questions to answer. For example, a prompt may provide direction by instructing the language model to focus on a specific topic or task. The input prompt can determine the context of the generated text and indicate the scope of information the model will refer to. Furthermore, a prompt may instruct the model on the format in which to generate the text (e.g., an answer to a question, a summary, etc.).

[0023] The program described herein can generate prompts to be input to a Large-Scale Language Model (LLM) based on risk keywords or risk-related questions or requests. The prompts may be the risk keywords or risk-related questions or requests entered by the user, or they may be modified versions of the input. The technique of crafting prompts to be input to an LLM so that the model generates desired output is called prompt engineering. In some embodiments, several examples for a task may be included in the prompt, and the model may infer based on these examples. This technique is also called fusion shot learning. Fusion shot learning can improve the model's performance by presenting several examples. Furthermore, in some embodiments, prompts to be input to an LLM can be created using RAG (Retrieval-Augmented Generation). RAG retrieves information related to the user-entered question from a RAG database, combines the original question (prompt) with the retrieved related information, and inputs the question into the LLM. By combining related information, it generates accurate and detailed responses. The fine-tuning described above can also be performed using the RAG function, and can be achieved by accumulating the information mentioned above in the RAG database. Figure 1 shows an example of scenario generation using RAG that can be used in the program of this disclosure. By using RAG, external information can be obtained in real time and questions can be sent to the LLM, which reduces the learning cost of the LLM and makes it possible to reflect the latest information.

[0024] In the context of this disclosure, examples of prompts that may be created based on risk keywords or risk-related questions or requests include the following: Example 1: "Create a risk scenario in which inventory shortages occur due to an incorrect demand forecast for product A." Example 2: "Create a scenario in which a large-scale earthquake occurs and affects treatments for rare diseases." Example 3: "Create a scenario for when a pharmaceutical wholesaler is subjected to a cyberattack and its core systems go down." Example 4: "This illustrates a risk of stockouts occurring when misjudging the planned production volume when launching a new product, Product A." Example 5: "List examples of risks associated with rare disease treatments, which inherently have limited production capacity, where raw materials cannot be procured at the appropriate time." Example 6: "List examples of risks where our factory would be damaged by a natural disaster such as a typhoon, making manufacturing impossible." Example 7: "List examples of risks where logistics functions would be paralyzed due to the disruption of major roads in the xx region caused by an earthquake."

[0025] In some embodiments, the program of the present disclosure may be a program that causes one or more computers to perform the step of inputting prompts into a large language model to output one or more risk scenarios.

[0026] In some embodiments, one or more risk scenarios are generated by entering appropriate prompts into the LLM. Figure 2 shows an example of scenario generation based on a query. In the context of this disclosure, the following results are output as an example of a risk scenario for the prompt in Example 3. Example 3-1: "Due to a failure in the wholesaler's core system, it became impossible to process orders for pharmaceuticals, resulting in the inability to deliver them to medical institutions." Example 3-2: "The core system was down for three days. We attempted to process orders and deliveries without using the system, but because fewer manufacturers accept fax orders, we were only able to place about 30% of the usual orders. It is estimated that it will take about a week to restore the system. The probability of this incident occurring is approximately 0.001 (once in a thousand years)." Example 3-3: "Immediately after the core system went down, the inventory level of insulin (a refrigerated product) became unknown, and it was not possible to confirm the inventory that should have been replenished within 24 hours, resulting in a one-week supply delay for some medical institutions. The probability of this incident occurring is approximately 0.01 (once in 100 years)."

[0027] In some embodiments, the program of this disclosure may be a program that causes one or more computers to perform a step of presenting one or more outputted scenarios to a user. The presentation to the user may be in any format, such as displaying on a display such as an LCD, printing to a printer, or reading text aloud by voice. Figure 3 shows an example of scenario generation in chat format, but the display format is not limited to this.

[0028] In some embodiments, the program of the Disclosure may be a program that causes one or more computers to perform a step of accepting the selection of one or more presented scenarios and / or modification of the content of those scenarios. In some embodiments, the program of the Disclosure may further be a program that causes the computer to perform a step of retraining a machine learning model (LLM) used for scenario generation based on the user's selection of the one or more scenarios and / or modification of the content of those scenarios.

[0029] In some embodiments, the selection of a scenario may be performed manually by the user or automatically by the program. In some embodiments, further prompts may be input to the LLM to generate alternative scenario options, and such input and output may be repeated as needed. In some embodiments, the program of this disclosure may also include a function (reinforcement learning algorithm) that learns the scenario presentation function for subsequent uses based on the results of the user's scenario selection. For example, TD learning or the Actor-Critic algorithm may be used for reinforcement learning.

[0030] Reinforcement learning is a branch of machine learning in which an agent interacts with its environment and learns the optimal action through trial and error. The agent observes the state of the environment and selects an action. The environment rewards the agent for the selected action, and the agent learns based on these rewards. The agent's goal is to find the optimal policy that yields the greatest reward in the long run, and in this process, the agent grows to select better actions while balancing exploration and utilization.

[0031] Temporal Difference Learning (TD) is a reinforcement learning technique in which an agent learns by refining its predictions of future rewards. A key feature of TD learning is that it compares the next observation with the predicted result to calculate the error (TD error), and updates the value of the state based on that error. This method is called "bootstrapping," and it is a mechanism that uses existing predicted values ​​to make new predictions. Because TD learning can update the value function online, real-time learning is possible. TD learning is also the foundation of basic reinforcement learning algorithms such as Q-learning and SARSA, and is an important technique for agents to adapt to the environment and learn better actions.

[0032] The Actor-Critic algorithm is another reinforcement learning technique in which an "Actor" learns action policies, and a "Critic" evaluates those actions, allowing the agent to learn better behaviors. The Actor selects an action, and the Critic evaluates the result and provides feedback, updating both the policy and the value function. This increases the stability of learning and allows for flexible handling of complex problems.

[0033] In some embodiments, the content of the scenario may be modified manually by the user or automatically by the program. In some embodiments, further prompts may be input to the LLM to make the desired modifications to the scenario, and such input and output may be repeated as needed. In some embodiments, the program of this disclosure may also include a function (reinforcement learning algorithm) that learns the scenario presentation function for subsequent uses based on the results of receiving user modifications to the scenario content.

[0034] Figure 4 shows an exemplary risk scenario screen. In some embodiments, a large-scale language model generates one or more candidate scenarios by inputting queries such as risk keywords or documents, and the user can then modify those scenarios. For example, this process is carried out as follows. 1. Enter a query such as risk keywords or documents, and press the scenario generation button to display a list of possible scenarios. 2. Scenarios are created by a large-scale language model (also known as generative AI). 3. Select the checkbox in the "Select Scenario Candidates" section and press the "Select Scenario" button. The content will then be displayed in the "Scenario Modification" box. 4. Modify the contents of the scenario modification box and press the "Confirm Scenario" button.

[0035] Figure 5 shows a flowchart illustrating an exemplary risk scenario generation process. As shown in Figure 5, RAG technology may be used to create prompts. Furthermore, the selection and modification of scenarios may involve multiple interactions with the LLM. In addition, although not shown in this figure, user-submitted scenario modifications may be used to retrain the LLM, thereby enhancing the programmatic scenario presentation function.

[0036] In some embodiments, the program of this disclosure may be a program that causes one or more computers to perform the step of presenting one or more components that can be used to simulate a supply chain, based on a scenario. In some embodiments, these components are presented as a graphical interface (GUI) on a display, for example, in the form of icons. Examples of components include components (icons) representing players such as manufacturers, wholesalers, retailers, and end consumers; components (arrows) representing the flow of materials between players and components (arrows) representing the flow of information between them; components representing the production volume of a manufacturer; components representing the consumption volume of an end consumer; and components representing the inventory levels of wholesalers and retailers. In some embodiments, the program performs analysis based on a scenario and presents the necessary components.

[0037] For example, as shown in Figure 6, a model creation panel consisting of a creation palette and a parts box is provided. The parts box contains one or more parts that can be used in the simulation. The user can visualize and configure a supply chain corresponding to a scenario by selecting parts provided in the parts box and placing them in the creation palette. The user can select the necessary elements from the parts box with an input device such as a mouse and move them to the creation palette by drag and drop. For example, first, icons representing manufacturers, wholesalers, retailers, and end consumers are placed on the creation palette. Next, arrows representing the flow of goods from the manufacturer to the wholesaler can be connected, and arrows representing the flow of information such as order information from the wholesaler to the manufacturer can be connected. Similarly, a network diagram can be created by connecting the relationships between the constituent elements. Alternatively, new parts can be added to the network configuration automatically created by the program, or unnecessary parts can be deleted. In some embodiments, the program according to this disclosure may warn the user of incorrect connections when configuring the supply chain and notify the user that it is not possible to connect the constituent elements.

[0038] In some embodiments, a Large-Scale Language Model (LLM) is used to analyze a scenario and present the supply chain components necessary to represent that scenario. The types of components can be selected from a pre-registered list. Furthermore, in some embodiments, the user is alerted if a connection between components does not make sense and cannot be made. In some embodiments, the components necessary to represent the scenario are learned by the Large-Scale Language Model (LLM), and appropriate components are presented. Since presenting all components may overwhelm the user with too many choices, the program may provide support to facilitate user convenience and enable simulations even for those without simulation expertise.

[0039] In some embodiments, a Large-Scale Language Model (LLM) is used to analyze a scenario and construct a proposed supply chain network configuration to represent that scenario. When creating mathematical formulas corresponding to the network diagram, connections between components can be mainly represented by weighted addition (+) and subtraction (-). Time fluctuations in inventory at intermediate wholesalers and retailers are represented by each item in a formula constructed by adding and subtracting pre-registered elements (formalized descriptions of the behavior of those elements). Furthermore, when multiple inventory components are involved, the resulting formula (model formula) can be expressed as a system of linear equations or a system of differential equations.

[0040] In some embodiments, the program of the Disclosure may be a program that causes one or more computers to perform a step of accepting the selection and / or rearrangement of one or more parts by a user. In some embodiments, the program of the Disclosure may further be a program that causes the computer to perform a step of retraining a machine learning model used for presenting parts based on the selection and / or rearrangement of one or more parts by the user. In some embodiments, the selection and / or rearrangement of parts may be achieved by combining icons representing each part on a creation palette using a mouse or the like. In some embodiments, if the connections between parts are inappropriate, an error is notified and the user is informed that the part cannot be placed in that location.

[0041] In some embodiments, the program of this disclosure can analyze a scenario and extract the components necessary for constructing a supply chain. Extraction may be performed using a vector search method or a semantic search method. When constructing the vector database, for example, a predefined (rule-based) correspondence table between natural language and mathematical formulas may be created and trained, or machine learning may be performed using academic papers or other materials containing mathematical formulas as training data. Mathematical formulas, or each term of a mathematical formula, may be treated as words in natural language and embedded in a vector space. Search may be performed using methods such as cosine similarity by calculating the closeness between the text or words and the mathematical formula or each term of the mathematical formula. In some embodiments, the program of this disclosure may include a function (reinforcement learning algorithm) that learns the function to present for subsequent times based on the results of accepting component selection and / or sorting.

[0042] In some embodiments, the program of this disclosure may be a program that causes one or more computers to perform the step of presenting to a user one or more parameters that can be used in a simulation. In some embodiments, the parameters may be, for example, at least one of the expected values ​​of inventory holding period, inventory turnover rate, inventory disposal / removal amount, inventory disposal / removal rate, average inventory quantity, average inventory ratio, production volume per unit time, production lead time, and transportation costs (transportation time, transportation cost, transportation distance, etc.), transportation route, consumption, consumption per unit time, inventory waste amount, inventory waste rate, inventory waste cost, and frequency (probability). The presentation of parameters can be done, for example, using a machine learning model such as a large-scale language model (LLM) or reinforcement learning. For reinforcement learning, for example, TD learning or the Actor-Critic algorithm may be used.

[0043] In some embodiments, the program of the Disclosure may be a program that causes one or more computers to perform a step of accepting a user's selection and / or modification of the one or more parameters. In some embodiments, the program of the Disclosure may further be a program that causes the computers to perform a step of retraining the machine learning model used to present the parameters based on the selection and / or modification of the one or more parameters. In some embodiments, the parameters may be at least one of the following, for example, inventory holding period, inventory turnover rate, inventory disposal / removal volume, inventory disposal / removal rate, average inventory volume, average inventory ratio, production volume per unit time, production lead time, and transportation costs (transportation time, transportation cost, transportation distance, etc.) and transportation routes.

[0044] In some embodiments, the program of the present disclosure may be a program that causes one or more computers to perform the step of running a supply chain simulation. In some embodiments, the supply chain simulation is based on a scenario generated by the LLM. In some embodiments, the supply chain simulation uses components selected and / or rearranged by the user. In some embodiments, the supply chain simulation uses parameters selected and / or modified by the user.

[0045] Figure 7 shows an example simulation execution screen equipped with a user interface for implementing simulation parameter modification and simulation execution functions. The following functions can be implemented on this screen. 1. The model equation display panel displays the created model equation. 2. The parameter display panel displays parameter information used in the simulation. 3. Set the simulation conditions in the parameter modification panel. 4. When you press the parameter confirmation button, the changes will be reflected in the parameter display. 5. When you press the Run Simulation button, the simulation will start and you will be taken to the simulation results screen. 6. Pressing the back button will return you to the model creation screen. 7. Pressing the top button will return you to the top page.

[0046] In some embodiments, the program of this disclosure may be a program that causes one or more computers to perform a step of presenting the results of a simulation to a user. In some embodiments, the simulation results include, for example, at least one of the expected values ​​corresponding to product inventory quantity, production quantity, production quantity per unit time, transport quantity per unit time, and transport time, transport distance, transport cost, transport route, consumption quantity, consumption quantity per unit time, inventory waste quantity, inventory waste rate, inventory waste cost, and frequency (probability) of each element in the supply chain. These results may include presentation in graphs and tables, as well as displaying transport routes, transport time, transport cost, and inventory quantity at the relevant point on a map in conjunction with a GIS (Geographic Information System). If the results change over time, they may be displayed as animations (videos). The simulation results can be used to determine the optimal inventory quantity or change transport routes as countermeasures against the risk. In addition, they can be used to visualize the magnitude of the impact and to consider risk countermeasures, and to contribute to the creation of a BCP (Business Continuity Plan).

[0047] Figure 8 shows an exemplary simulation results screen equipped with a user interface for reviewing the simulation results. The following functions can be implemented on this screen. 1. The model equation display panel displays the model equation used for execution. 2. The parameter display panel shows the parameter information used during execution. 3. The results display panel shows various information about the simulation results. 4. Pressing the back button will return you to the model creation screen. 5. Pressing the top button will return you to the top page.

[0048] In some embodiments, the program of this disclosure is installed on one or more computers. A step of obtaining one or more risk keywords or risk-related questions or requests entered by the user, A step of generating prompts to input into a large-scale language model based on risk keywords or risk-related questions or requests, The steps include inputting a prompt into a large-scale language model to output one or more risk scenarios, A step of presenting one or more of the outputted scenarios to the user, A step of accepting the user's selection of one or more scenarios and / or modification of their content, Based on the aforementioned scenario, the step of presenting one or more components that can be used in supply chain simulation, A step of accepting the user's selection and / or rearrangement of one or more of the above-mentioned parts, A step of presenting the user with one or more parameters that can be used in the simulation, A step of accepting the user's selection and / or modification of one or more of the parameters, Steps to perform a supply chain simulation, Steps to present the results of the simulation to the user. Regarding programs that execute.

[0049] In some embodiments, the program of the Disclosure may be a program that causes one or more computers to perform the step of saving and / or loading scenarios. In some embodiments, the program of the Disclosure may save generated scenarios to a storage device connected to one or more computers. Also, in some embodiments, the program of the Disclosure may load scenarios saved to a storage device connected to one or more computers into the memory of the computers.

[0050] In some embodiments, the program of the present disclosure may be a program that causes one or more computers to perform the step of selecting a Large Language Model (LLM). In some embodiments, the Large Language Model (LLM) may be selected manually by a user or automatically by a program depending on the desired task.

[0051] In some embodiments, the program of the present disclosure may be a program that causes one or more computers to perform the step of updating a database used by the RAG. In some embodiments, the updating of the database used by the RAG may be performed periodically, for example, daily, weekly, or monthly.

[0052] Figure 9 shows an exemplary program logic structure of the program disclosed herein. Figure 10 shows an exemplary flowchart of the program disclosed herein. Although not explicitly shown in these figures, the program disclosed herein may further have a function (reinforcement learning function) that learns the results of user modifications to the risk scenarios presented by the program and learns the function of presenting risk scenarios in subsequent instances, and a function (reinforcement learning function) that learns the results of user modifications to the components and simulation parameters used in the simulations presented by the program and learns the function of presenting them in subsequent instances. It will be understood by those skilled in the art that the matters described herein as features of the program disclosed herein can be appropriately applied to the devices and systems described below.

[0053] A device to support supply chain risk assessment. This disclosure relates, in one embodiment, to an apparatus for assisting in supply chain risk assessment.

[0054] In some embodiments, the present disclosure relates to an apparatus for assisting supply chain risk assessment, comprising: a query input unit that acquires one or more risk keywords or risk-related questions or requests entered by a user; a prompt generation unit that generates prompts to be input to a large-scale language model based on the risk keywords or risk-related questions or requests; a risk scenario generation unit that inputs the prompts to the large-scale language model and outputs one or more risk scenarios; and a scenario presentation unit that presents the output one or more scenarios to the user.

[0055] In some embodiments, the apparatus of the Disclosure may include a processor that performs various operations and memory connected to the processor. In some embodiments, the apparatus of the Disclosure may be realized by running the program of the Disclosure on a general-purpose computer. In some embodiments, the apparatus of the Disclosure is an apparatus that includes a processor connected to a storage device storing the computer program of the Disclosure and capable of executing instructions of the program.

[0056] In some embodiments, the apparatus relating to this disclosure includes means for inputting data. Examples of means for inputting data include a keyboard and a mouse.

[0057] In some embodiments, the apparatus according to the present disclosure includes a central processing unit (CPU) connected to a keyboard, mouse, etc. for data input, a hard disk, flash memory, etc. as a storage unit, and a memory (storage means) such as ROM, RAM, etc.

[0058] Means for outputting data including prediction results include, for example, monitors and printers. Alternatively, means for storing the data in storage devices such as hard disks, flash memory, ROM, and RAM can also be considered output methods.

[0059] The device relating to this disclosure may include means for storing a program to support the risk assessment of the supply chain relating to this disclosure. Examples of means for storing the program include hard disks and flash memory. Such storage means may be connected by a communication line. That is, the device relating to this disclosure may be part of a system obtained by connecting a device including the means for storing the program via a communication line.

[0060] Figure 11 is a schematic diagram showing an exemplary embodiment of the apparatus according to the present disclosure. In Figure 11, 100 is a computer, comprising a control unit 101, a storage unit 102, a peripheral device I / F unit 103, an input unit 104, a display unit 105, and a communication unit 106, which are connected by a bus 110. Note that this configuration is illustrative, and various configurations can be adopted as appropriate.

[0061] The control unit 101 consists of a CPM (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), etc. The CPU calls programs stored in the memory unit 102, ROM, recording medium, etc., into the work memory area on RAM and executes them, drives and controls each device connected via the bus 110, and realizes the processing that the computer performs. ROM is a non-volatile memory that holds programs and data such as the computer 100's boot program and BIOS. RAM is a volatile memory that temporarily holds programs and data loaded from the memory unit 102, ROM, recording medium, etc., and also has a work area used by the control unit 101 when performing various processing. The memory unit 102 is, for example, an HDD (Hard Disk Drive) and stores programs executed by the control unit 101 and various other data.

[0062] The peripheral device interface (I / F) section 103 is a port for connecting the computer 100 to peripheral devices. The peripheral device interface section 103 consists of USB, Bluetooth, IEEE1394, RS-232C, etc. The connection method to peripheral devices can be wired or wireless. The input section 104 has input devices such as a keyboard, mouse, pointing device, and numeric keypad, and provides operation instructions, action instructions, and data input to the computer 100. The display section 105 is a logic circuit or device driver for displaying images and videos on a display device such as an LCD panel. The input section 104 and the display section 105 can also be configured as an integrated touch display.

[0063] The communication unit 106 has a communication control device, a communication port, etc., and is a wired or wireless communication interface that mediates communication with the network 120. The bus 110 is a communication path that mediates the exchange of control signals, data signals, etc. between each device. The network 120 may be further connected to an external server 130 or a database (or network storage) 140.

[0064] In some embodiments, the apparatus of the Disclosure may further include a scenario operation unit that accepts the selection and / or modification of one or more scenarios by a user. In some embodiments, the apparatus of the Disclosure may further include a scenario retraining unit that retrains a machine learning model based on the user's selection and / or modification of the one or more scenarios.

[0065] In some embodiments, the apparatus of the present disclosure may further include a component presenter that presents one or more components that can be used for supply chain simulation based on a scenario.

[0066] In some embodiments, the apparatus of the Disclosure may further include a component handling unit that accepts the selection and / or rearrangement of one or more components by a user. In some embodiments, the apparatus of the Disclosure may further include a component retraining unit that retrains a machine learning model based on the selection and / or rearrangement of the one or more components by the user.

[0067] In some embodiments, the apparatus of the present disclosure may further include a parameter presentation unit that presents to the user one or more parameters that can be used in a simulation.

[0068] In some embodiments, the apparatus of the Disclosure may further include a parameter manipulation unit that accepts the selection and / or modification of one or more parameters by a user. In some embodiments, the apparatus of the Disclosure may further include a parameter relearning unit that retrains a machine learning model based on the selection and / or modification of the one or more parameters by the user.

[0069] In some embodiments, the apparatus of the present disclosure may further include a simulation execution unit that performs a supply chain simulation.

[0070] In some embodiments, the apparatus of the present disclosure may further include a simulation results presentation unit that presents the simulation results to the user.

[0071] In some embodiments, the simulation results include at least one of the expected values ​​corresponding to production volume, production volume per unit time, product inventory volume, transport volume per unit time, transport time, transport distance, transport cost, and transport route, consumption volume, consumption volume per unit time, inventory waste volume, inventory waste rate, inventory waste cost, and frequency (probability).

[0072] In some embodiments, the apparatus of the present disclosure may further include a scenario input / output unit for saving and / or loading scenarios.

[0073] In some embodiments, the apparatus of the present disclosure may further include an LLM selection unit for selecting a large language model (LLM).

[0074] In some embodiments, RAG (Retrieval-Augmented Generation) may be used to generate prompts.

[0075] In some embodiments, the apparatus of the present disclosure may further include a DB update unit that updates the database used by the RAG.

[0076] A system to support supply chain risk assessment. This disclosure relates, in one aspect, to a system for assisting in supply chain risk assessment.

[0077] In some embodiments, the System of the Disclosure is a system for assisting supply chain risk assessment, comprising one or more computers, the one or more computers comprising memory and a processor connected to the memory, wherein the processor performs the steps of: obtaining one or more risk keywords or risk-related questions or requests entered by a user; generating prompts to be entered into a large language model based on the risk keywords or risk-related questions or requests; entering the prompts into the large language model to output one or more risk scenarios; and presenting the output one or more scenarios to the user. In some embodiments, the memory or storage device of a computer included in the System of the Disclosure stores the Program of the Disclosure, and when the Program is executed, each of the above steps is performed.

[0078] In some embodiments, the processor may further perform a step of accepting a user's selection and / or modification of the one or more scenarios. In some embodiments, the processor may further perform a step of retraining a machine learning model based on the user's selection and / or modification of the one or more scenarios.

[0079] In some embodiments, the processor may further perform the step of presenting one or more components that can be used in supply chain simulation based on the scenario.

[0080] In some embodiments, the processor may further perform a step of accepting a user's selection and / or rearrangement of one or more components. In some embodiments, the processor may further perform a step of retraining a machine learning model based on the user's selection and / or rearrangement of one or more components.

[0081] In some embodiments, the processor may further perform the step of presenting the user with one or more parameters that can be used in the simulation.

[0082] In some embodiments, the processor may further perform a step of accepting the user's selection and / or modification of one or more parameters. In some embodiments, the processor may further perform a step of retraining the machine learning model based on the user's selection and / or modification of one or more parameters.

[0083] In some embodiments, the processor may further perform the step of simulating the supply chain.

[0084] In some embodiments, the processor may further perform the step of presenting the simulation results to the user.

[0085] In some embodiments, the simulation results include at least one of the expected values ​​corresponding to production volume, production volume per unit time, product inventory volume, transport volume per unit time, transport time, transport distance, transport cost, and transport route, consumption volume, consumption volume per unit time, inventory waste volume, inventory waste rate, inventory waste cost, and frequency (probability).

[0086] In some embodiments, the processor may further perform the step of saving and / or loading scenarios.

[0087] In some embodiments, the processor may further perform the step of selecting a Large Language Model (LLM).

[0088] In some embodiments, RAG (Retrieval-Augmented Generation) may be used to generate prompts.

[0089] In some embodiments, the processor may further perform the step of updating the database used by the RAG.

[0090] Non-temporary computer-readable recording medium This disclosure relates, in one embodiment, to a non-temporary computer-readable recording medium storing the program of this disclosure. Examples of computer-readable recording media include, but are not limited to, hard disk drives (HDDs), solid-state drives (SSDs), flash memory (such as USB memory sticks and SD cards), optical discs (such as CDs, DVDs, and Blu-ray® discs), magnetic tapes, floppy disks, and cloud storage.

[0091] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art to which the invention pertains. Any methods and materials similar or equivalent to those described herein may be used for carrying out or testing the invention, but several possible and preferred methods and materials are described herein. All publications referenced herein are incorporated herein by reference, and the methods and / or materials cited in relation to these publications are disclosed and described herein. In the event of any conflict, this disclosure shall prevail over the disclosure of the incorporated publication.

[0092] Where a range of values ​​is given, unless the context clearly indicates otherwise, each intermediate value between the upper and lower limits of that range, up to one-tenth of the lower limit unit, is also specifically disclosed. Each smaller range between any given or intermediate value within the given range and any other given or intermediate value within that given range is also included in this disclosure. The upper and lower limits of these smaller ranges may be independently included in or excluded from that range, and each range that includes either, either, or both of the limit values ​​in the smaller range is also included in the invention, but the limit values ​​specifically excluded in the given range are reserved. Where a given range includes one or both of the limit values, a range that excludes either or both of the included limit values ​​is also included in the invention. The term “about” with respect to numerical values ​​means within 5%.

[0093] The embodiments described herein are intended to be illustrative only, and those skilled in the art will be able to make numerous modifications and alterations without departing from the spirit of the invention. Certain modifications and alterations may yield satisfactory results, though not optimal. All such modifications and alterations are intended to fall within the scope of the invention as defined by the appended claims. Furthermore, any combination of the components disclosed herein, and any conversion of the expressions of this disclosure between methods, apparatus, systems, computer programs, data structures, recording media, etc., are also valid embodiments of this disclosure. Thus, the details described relating to the methods of this disclosure may be applied to systems, computer programs, data structures, recording media, etc.

[0094] This disclosure will be further understood by reference to the examples and drawings herein. These examples are provided solely to illustrate the disclosure set forth in the claims, and the scope of this disclosure is not limited by the illustrated embodiments, which are intended only as examples of a single aspect of this disclosure. Any functionally equivalent method is included within the scope of this disclosure. In addition to those described herein, various modifications of this disclosure will be apparent to those skilled in the art from the foregoing description. Such modifications are intended to be within the scope of the appended claims. [Explanation of Symbols]

[0095] 100... Computers 101... Control Unit 102...Storage section 103...Peripheral Devices I / F Section 104...Input section 105...Display section 106... Communications Department 110...bus 120 Network 130...External Server 140...Database

Claims

1. A program to support supply chain risk assessment, which is installed on one or more computers, A step of obtaining one or more risk keywords or risk-related questions or requests entered by the user, A step of generating prompts to input into a large-scale language model based on risk keywords or risk-related questions or requests, The steps include inputting a prompt into a large-scale language model to output one or more risk scenarios, A step to present one or more of the outputted scenarios to the user. A program that executes something.

2. The aforementioned one or more computers further include: A step that accepts the user's selection of one or more scenarios and / or modification of their content. The program according to claim 1, which causes to execute.

3. The aforementioned one or more computers further include: A step to retrain the machine learning model based on the user's selection of one or more scenarios and / or modification of their content. The program according to claim 2, which causes the execution of the program.

4. The aforementioned one or more computers further include: Based on the aforementioned scenario, the step of presenting one or more components that can be used in supply chain simulation. The program according to claim 1, which causes to execute.

5. The aforementioned one or more computers further include: A step of accepting the user's selection and / or rearrangement of one or more of the above-mentioned parts. The program according to claim 4, which causes the execution of the program.

6. The aforementioned one or more computers further include: A step in which a machine learning model is retrained based on the user's selection and / or rearrangement of one or more of the aforementioned components. The program according to claim 5, which causes the execution of the program.

7. The aforementioned one or more computers further include: A step of presenting the user with one or more parameters that can be used in the simulation. The program according to claim 1, which causes to execute.

8. The aforementioned one or more computers further include: A step of accepting the user's selection and / or modification of one or more of the above parameters. The program according to claim 7, which causes the execution of the program.

9. The aforementioned one or more computers further include: A step of retraining a machine learning model based on the user's selection and / or modification of one or more parameters. The program according to claim 8, which causes the execution of the program.

10. The aforementioned one or more computers further include: Steps to perform a supply chain simulation The program according to claim 1, which causes to execute.

11. The aforementioned one or more computers further include: Steps to present the results of the simulation to the user. The program according to claim 10, which causes to execute

12. The program according to claim 11, wherein the results of the simulation include at least one of the expected values ​​corresponding to production volume, production volume per unit time, product inventory volume, transport volume per unit time, transport time, transport distance, transport cost, and transport route, consumption volume, consumption volume per unit time, inventory waste volume, inventory waste rate, inventory waste cost, and frequency (probability) of occurrence.

13. The aforementioned one or more computers further include: Steps to save and / or load a scenario The program according to claim 1, which causes to execute.

14. The aforementioned one or more computers further include: Steps to select a Large-Scale Language Model (LLM) The program according to claim 1, which causes to execute.

15. The program according to claim 1, wherein RAG (Retrieval-Augmented Generation) is used to generate prompts.

16. The aforementioned one or more computers further include: Steps to update the database used by RAG The program according to claim 15, which causes to execute.

17. A device for supporting supply chain risk assessment, A query input section that retrieves one or more risk keywords or risk-related questions or requests entered by the user, A prompt generation unit that generates prompts to be input into a large-scale language model based on risk keywords or risk-related questions or requests, A risk scenario generation unit that inputs a prompt into a large-scale language model and outputs one or more risk scenarios, Scenario presentation unit that presents one or more outputted scenarios to the user. A device including a device.

18. A system for supporting supply chain risk assessment, comprising one or more computers, The one or more computers include memory and a processor connected to the memory. The aforementioned processor, A step of obtaining one or more risk keywords or risk-related questions or requests entered by the user, A step of generating prompts to input into a large-scale language model based on risk keywords or risk-related questions or requests, The steps include inputting a prompt into a large-scale language model to output one or more risk scenarios, A step to present one or more of the outputted scenarios to the user. A system that executes this.

19. A non-temporary computer-readable recording medium storing the program described in claim 1.

Images