Risk scenario evaluation system and risk scenario evaluation method

Abstract

We will thoroughly assess, during normal times, the scale and duration of damage caused by potential troubles and disasters in the supply chain. [Solution] The risk scenario evaluation system of the present invention comprises: a damage performance structuring unit that generates damage performance structured information in which the damage target, damage scale, and damage period are stored in a format usable as a database, based on damage implementation information stored in natural language text, and a damage classification master that stores the definition method of the damage target, damage scale, and damage period, which are types of damage to be of interest; and a damage performance abstraction unit that generates damage performance abstraction information in which the risk type value is stored in relation to the damage target, damage scale, and damage period, based on the damage performance structured information, based on a risk response master that stores the risk type value, which is an indicator of the magnitude of the impact that the risk type associated with the cause of the damage has on the damage target.

Description

Technical Field

[0001] The present invention relates to a risk scenario evaluation system and a risk scenario evaluation method.

Background Art

[0002] In recent years, along with the increase in geopolitical risks and the like, the risk of supply chain disruption has been increasing. In order to improve the business continuity of a company itself, it is required to be prepared for disasters and geopolitical risks from normal times. In particular, countermeasures against troubles and disasters occurring to suppliers need to be formulated according to the damage caused by troubles and disasters, the magnitude of the damage, and the continuation status of the damage. It is important to predict and evaluate in advance the damage caused by the assumed risks to the supply chain, its scale, and its duration.

[0003] As the background art of this technical field, there is Patent Document 1. The supply chain support system of Patent Document 1 includes a supply chain information acquisition means, a model setting means for setting a supply chain model corresponding individually to normal times and when a risk event occurs, an estimated data derivation means for deriving estimated data estimating the time-series situation in the supply chain, an actual result data acquisition means for acquiring actual result data from the supply chain in time series, a risk judgment means for judging the presence or absence of a risk event occurrence from the ratio of the estimated data and the actual result data, and when a risk event occurs, using a risk response supply chain model to perform a simulation and deriving a supply chain system and a risk cost that minimize the risk loss occurring when the risk event occurs.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] The supply chain support system described in Patent Document 1 estimates the time-series state of the supply chain, determines whether or not a risk has occurred by comparing it with actual supply chain data, and derives countermeasures and risk costs to minimize losses caused by that risk when it occurs.

[0006] However, while the supply chain support system described in Patent Document 1 can formulate countermeasures after a trouble or disaster occurs, it cannot anticipate troubles and disasters that may occur at suppliers during normal times and evaluate the damage, scale of damage, and duration of damage caused by such troubles and disasters. Furthermore, the supply chain support system described in Patent Document 1 cannot evaluate the damage, scale of damage, and duration of damage caused by troubles and disasters for which there is no history of damage.

[0007] Therefore, the present invention aims to comprehensively evaluate the scale and duration of damage caused by troubles and disasters that may occur in the supply chain, even during normal times. [Means for solving the problem]

[0008] The risk scenario evaluation system of the present invention is characterized by comprising: a damage performance structuring unit that generates damage performance structured information in which the damage target, the damage scale, and the damage period are stored in a format usable as a database, based on damage implementation information in which the cause, scale, and duration of damage in the supply chain are stored in natural language text, and based on a damage classification master in which the type of damage of interest is the target of damage, the method for defining the scale of damage, and the method for defining the duration of damage are stored, and a damage performance abstraction unit that generates damage performance abstraction information in which the risk type values, which are indicators of the magnitude of the impact that the risk type associated with the cause of the damage has on the target of damage are stored, in relation to the target of damage, the scale, and the duration of damage, based on the damage performance structured information. Other means will be described within the descriptions of embodiments for carrying out the invention. [Effects of the Invention]

[0009] According to the present invention, the scale and duration of damage caused by troubles and disasters that may occur in the supply chain can be comprehensively evaluated during normal times. More specifically, according to the present invention, it becomes possible for buyers, who are users of the present invention, to estimate during normal times the scale and duration of damage that suppliers may suffer due to troubles and disasters, based on the risks faced by suppliers. This makes it possible for buyers to formulate countermeasures in advance to avoid risks in a manner that is appropriate to the type, scale, and duration of damage expected from the risks faced by suppliers. [Brief explanation of the drawing]

[0010] [Figure 1] This is a functional block diagram of the risk scenario evaluation system. [Figure 2] This is a hardware configuration diagram of the risk scenario evaluation system. [Figure 3] This is a flowchart of the overall process. [Figure 4] Figure 3 is a flowchart showing the detailed processing steps S10. [Figure 5] This figure shows an example of damage history information. [Figure 6] This figure shows an example of a damage classification master. [Figure 7] This figure shows an example of structured information on damage records. [Figure 8] This diagram illustrates the structuring of damage report information. [Figure 9] Figure 3 is a flowchart showing the detailed processing steps S20. [Figure 10] This figure shows an example of a risk response master. [Figure 11] This figure shows an example of abstracted information regarding damage records. [Figure 12] Figure 3 is a flowchart showing the detailed processing steps S30. [Figure 13]It is a diagram showing an example of supply chain information. [Figure 14] It is a diagram showing an example of risk scenario information. [Figure 15] It is a diagram for explaining the filtering in step S320 of FIG. 12. [Figure 16] It is a diagram for explaining the filtering in step S330 of FIG. 12. [Figure 17] It is a diagram showing an example of risk scenario evaluation information. [Figure 18] It is an example of a risk scenario evaluation result screen.

Mode for Carrying Out the Invention

[0011] Hereinafter, embodiments of the present invention (referred to as "the present embodiment") will be described with reference to the drawings.

[0012] (Term Arrangement) In the present embodiment, "cause", "risk", "risk type", and "risk type value" are concepts that are similar to each other but should be clearly distinguished.

[0013] A cause is an event that causes damage, such as "cold wave", "drought", "hurricane", "flood", etc. A cause does not necessarily have to be a natural phenomenon, and it may be an economic event, a political event, a conflict event, a criminal event, etc., or an event such as an infectious disease. In any case, it is important that the term "cold wave" or the like as a cause is described in open data such as news almost as it is.

[0014] A risk is a cause from the perspective of the administrator of the risk scenario evaluation system. Therefore, for example, "flood" is a cause as a term appearing in open data, and is a risk from the perspective of the administrator. The term "cause" includes the meaning of "an event that actually occurred", while the term "risk" includes the meaning of "an event that should be precautionarily noted in combination with an event that appeared in news etc., even if it does not directly appear in news etc.".

[0015] A risk type is a broader concept that groups together risks that cause the same or similar damage. As mentioned above, a risk is a cause from the perspective of the administrator, so a risk type is also a broader concept of the cause. For example, "cold wave," "drought," "hurricane," and "flood," which are listed as causes in open data, are risks from the administrator's perspective, and these risks belong to the risk type "earthquake, flood, drought." When someone faces news of a "cold wave," their imagination is limited to the actual situation of the "cold wave," and it is unlikely that they will consider the "drought," "hurricane," and "flood" that should be addressed together. Risk types fill in this gap in imagination. In this embodiment, risk types are assigned the symbols A, B, C, ...

[0016] The risk type value is an indicator that represents the magnitude of the impact that a risk type has on the target of damage (details below). The risk type value is defined for each combination of target of damage and risk type.

[0017] (Functions of the risk scenario assessment system) Figure 1 is a functional block diagram of the risk scenario evaluation system 1000. The risk scenario evaluation system 1000 includes an information collection management unit 1100, a damage record structuring unit 1200, a damage record abstraction unit 1300, a risk scenario evaluation unit 1400, and a risk scenario evaluation result display unit 1500. These are programs that describe the procedures for information processing. The risk scenario evaluation system 1000 also includes an input / output interface unit 1600 and a data storage unit 2000.

[0018] The information collection and management unit 1100 acquires text data (unstructured data) from open data such as news reports, government and research agency reports, etc., describing the location, cause, scale and duration of damage suffered by companies in the supply chain, etc., in natural language, etc., via a communication device or input screen provided by the input / output interface unit 1600, and stores it in the data storage unit 2000 as damage performance information 2100.

[0019] The information collection and management unit 1100 acquires data on its supply chain constituent companies, location information of constituent companies, and items handled by constituent companies from its own database, EDI (Electronic Data Interchange) system, etc., via a communication device or input screen provided by the input / output interface unit 1600, and stores it in the data storage unit 2000 as supply chain information 2200. The supply chain information 2200 stores suppliers in association with the location of the damage as a condition (filtering condition) for limiting the records of damage history abstraction information.

[0020] The information collection and management unit 1100 extracts key points (country, region, risk, and risk type) from the damage record information 2100 and stores them in the data storage unit 2000 as risk scenario information 2300. The risk scenario information 2300 stores the risk type as a condition (filtering condition) for limiting the records of the damage record abstraction information 2700.

[0021] The information collection and management unit 1100 acquires data containing the type of damage of interest, such as the target of the damage, the method for defining the scale of the damage, and the method for defining the period of the damage, via the input screen provided by the input / output interface unit 1600, and stores it in the data storage unit 2000 as a damage classification master 2400.

[0022] The information collection and management unit 1100 acquires the risk response master 2500, which stores risk type values ​​that are indicators of the magnitude of the impact that risk types, as a higher-level concept of the cause of damage, have on the affected target, via the input screen provided by the input / output interface unit 1600, and stores it in the data storage unit 2000.

[0023] The damage record structuring unit 1200 generates damage record structured information 2600 from the unstructured damage record information 2100, based on the damage classification master 2400, which structures the location of the damage, the target of the damage, the scale of the damage, and the duration of the damage into a format usable as a database, and stores it in the data storage unit 2000.

[0024] The damage history abstraction unit 1300 generates damage history abstraction information 2700 from the damage history structured information 2600, based on the risk response master 2500, and stores the risk type value associated with the location of the damage, the target of the damage, the scale of the damage, and the duration of the damage, and stores it in the data storage unit 2000. "Abstraction" means "to move away from specific examples and generalize" or "to extract the essence common to individual examples."

[0025] The risk scenario evaluation unit 1400 generates risk scenario evaluation information 2800 from the damage history abstraction information 2700, based on the supply chain information 2200 and the risk scenario information 2300, and stores statistical values ​​of the scale of damage and statistical values ​​of the duration of damage for each supplier and each affected target, in association with the supplier and the affected target, and stores it in the data storage unit 2000.

[0026] The risk scenario evaluation result display unit 1500 displays the risk scenario evaluation information 2800 and informs the user of the scale and duration of the damage for each supplier and each affected party.

[0027] (Hardware configuration of the risk scenario assessment system) Figure 2 is a hardware configuration diagram of the risk scenario evaluation system 1000. The risk scenario evaluation system 1000 is connected to each other via a network 30 with terminal devices 40 and information provision devices 50. The risk scenario evaluation system 1000 can be implemented as a general-purpose computer. For this reason, the risk scenario evaluation system 1000 has a CPU (Central Processing Unit) 11, ROM (Read Only memory) 12, RAM (Random Access Memory) 13, auxiliary storage device 14, display device 15, input device 16, media reader 17, and information transmission / reception device 18.

[0028] The CPU 11 is a processor that performs various calculations. To perform calculations, the CPU 11 loads the aforementioned programs from the auxiliary storage device 14 into the ROM 12, thereby realizing the functions of each program.

[0029] The risk scenario evaluation system 1000 may be installed as an application executable on an OS (Operating System) program, for example, by installing the aforementioned program from a portable storage medium to an auxiliary storage device 14 via a media reader 17.

[0030] ROM12 is a memory that stores the program executed by the CPU11, as well as data necessary for the execution of the program. RAM13 is memory that stores other programs and other items necessary for starting the risk scenario evaluation system 1000.

[0031] The auxiliary storage device 14 is, for example, a device such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive). The auxiliary storage device 14 may be implemented as a separate device from the risk scenario evaluation system 1000. In this case, the auxiliary storage device 14 may be implemented as a file server connected to the network 30. The auxiliary storage device 14 may be provided both inside and outside the risk scenario evaluation system 1000, and may share the storage of information. Note that the auxiliary storage device 14 in Figure 2 corresponds to the data storage unit 2000 in Figure 1.

[0032] The display device 15 is, for example, a CRT (Cathode Ray Tube) display, an LCD (Liquid Crystal Display), or an organic EL (Electro-Luminescence) display, and performs the functions of the risk scenario evaluation result display unit 1500 shown in Figure 1. The input device 16 is, for example, a keyboard, mouse, or microphone, and performs the functions of the input / output interface unit 1600 in Figure 1. When the risk scenario evaluation system 1000 is implemented using a so-called server, the display device 15 and the input device 16 are omitted, and their functions are handled by the terminal device 40. The display device 15 and the input device 16 may be integrated as a touch panel.

[0033] The media reader 17 is a device that reads information from portable storage media such as CD-ROMs. The information transmission / reception device 18 is a device that transmits and receives data to and from external devices such as terminal devices 40 via the network 30. Examples include communication equipment that communicates with the network 30, such as a wired LAN or wireless LAN, a dial-up router, or an infrared communication device. The information transmission / reception device 18 performs the functions of the input / output interface unit 1600 shown in Figure 1.

[0034] The risk scenario evaluation system 1000 of this embodiment is a server and uses a network 30, terminal devices 40, and information provision devices 50. Network 30 only needs to enable communication between devices, and its type (LAN, WAN, etc.) is not specified. The terminal device 40 is a computer such as a personal computer or tablet terminal, and has the functions of a display device 15 and an input device 16. It accepts operations from the user and displays the processing results of the risk scenario evaluation system 1000. There may be multiple terminal devices 40.

[0035] The information provision device 50 is a computer such as a server that provides various types of information, and includes a system that provides open data, an EDI system, and a system that provides supply chain information. The risk scenario evaluation system 1000 then acquires various types of information from the information provision device 50 as described below.

[0036] (Flowchart of the overall process) Figure 3 is a flowchart of the overall process. In step S10, the damage performance structuring unit 1200 of the risk scenario evaluation system 1000 structures the damage performance information 2100 and creates damage performance structured information 2600. In step S20, the damage history abstraction unit 1300 of the risk scenario evaluation system 1000 abstracts the damage history structured information 2600 and creates damage history abstraction information 2700. In step S30, the risk scenario evaluation unit 1400 of the risk scenario evaluation system 1000 creates risk scenario evaluation information 2800 from the damage history abstraction information 2700 to predict and evaluate damage for each supplier and each affected entity. In step S40, the risk scenario evaluation result display unit 1500 of the risk scenario evaluation system 1000 displays the risk scenario evaluation information 2800 as the evaluation result.

[0037] Details of step S10 are shown later as Figure 4. Details of step S20 are shown later as Figure 9. Details of step S30 are shown later as Figure 12. The detailed screen for step S40 is shown later as Figure 18.

[0038] (Details of Step S10) Figure 4 is a flowchart showing the detailed processing of step S10 in Figure 3. In step S110, the damage record structuring unit 1200 acquires the damage record information 2100 and the damage classification master 2400 stored in the data storage unit 2000. Although we are still in the middle of Figure 4, the explanation will now move on to Figures 5 and 6.

[0039] (Information on past damage) Figure 5 shows an example of damage history information 2100. Damage history information 2100 describes (stores) the industry to which the supply chain belongs, the damage within the supply chain, the cause of the damage, the scale of the damage, and the duration of the damage in natural language text. Specifically, Figure 5 is a news article about the suspension of production of automotive parts by Company X due to a large-scale power outage caused by a record-breaking cold wave that occurred in North America on February 7, 2021. However, it is surprisingly difficult for users to quickly and accurately understand the gist of such unstructured data.

[0040] The information collection and management unit 1100 acquires this unstructured data from the web, external databases, etc., via the input / output interface unit 1600 through processes such as crawling and scraping, and stores it in the data storage unit 2000 as damage record information 2100.

[0041] (Damage Classification Master) Figure 6 shows an example of the damage classification master 2400. The information collection and management unit 1100 acquires the damage classification master 2400, for example, through the input / output interface unit 1600, either by manual input or data upload. The damage classification master 2400 stores the target of the damage (column 2401), the method for defining the scale of the damage (column 2402), and the method for defining the period of the damage (column 2403) in relation to each other. The "targets of damage" refer to the types of damage (perspectives or viewpoints) that should be considered when analyzing damage in relation to business activities. In this case, these are "parts procurement," "production capacity," "inventory loss," "transportation lead time," "review of transactions," and "supply disruption." These targets of damage are things that manufacturing companies should always be aware of, even during normal times.

[0042] The damage scale column 2402 stores the method for defining the damage scale. For example, "parts procurement" is associated with "enter the amount of parts procured compared to normal times between 0% and 100%." ​​This indicates that the damage performance structuring unit 1200 quantifies the details regarding the damage scale described in the damage performance information 2100 in this concise manner. The damage period column 2403 stores the method for defining the damage period. For example, "parts procurement" is associated with "inputting the number of days for which the amount of parts procured decreased." This indicates that the damage performance structuring unit 1200 quantifies the details regarding the damage period described in the damage performance information 2100 in this concise manner. The explanation returns to Figure 4.

[0043] In step S120, the damage performance structuring unit 1200 structures the unstructured damage performance information 2100 based on the damage classification master 2400 for each piece of damage performance information 2100, and generates damage performance structured information 2600. Although we are still in the middle of Figure 4, the explanation will now move on to Figures 7 and 8.

[0044] (Structured information on damage records) Figure 7 shows an example of the Damage Record Structured Information 2600. The Damage Record Structured Information 2600 stores the following information in an interrelated manner: time (column 2601), company name (column 2602), industry (column 2603), country (column 2604), region (column 2605), cause (column 2606), affected party (column 2607), scale of damage (column 2608), and duration of damage (column 2609). Note that "location" refers to a geographical location and is a concept that includes "country" and "region". In the Damage Record Structured Information 2600, the industry, affected party, scale of damage, and duration of damage are stored in a format that can be used as a database. One record of the Damage Record Structured Information 2600 corresponds to one Damage Record Information 2100.

[0045] Furthermore, for example, if the risk scenario assessment system 1000 is operated in a specific location with a small area, the damage performance structured information 2600 does not need to have a field related to location. The same applies to other information generated from the damage performance structured information 2600. However, in this case, filtering to limit the location (details below) becomes impossible.

[0046] The "Target" section of the Damage Record Structured Information 2600 is the result of the Damage Record Structuring Unit 1200 replacing the corresponding section of the Damage Record Information 2100 with the target of damage in the Damage Classification Master 2400. The "Scale of Damage" section of the Damage Record Structured Information 2600 is the result of the Damage Record Structuring Unit 1200 quantifying the corresponding section of the Damage Record Information 2100 using the definition method for scale of damage in the Damage Classification Master 2400. The "Period of Damage" section of the Damage Record Structured Information 2600 is the result of the Damage Record Structuring Unit 1200 quantifying the corresponding section of the Damage Record Information 2100 using the definition method for duration of damage in the Damage Classification Master 2400. The "Other" section of the Damage Record Structured Information 2600 is the result of the Damage Record Structuring Unit 1200 summarizing the sections of the Damage Record Information 2100 corresponding to time, company name, industry, country, region, and cause.

[0047] (Structuring of damage history information) Figure 8 illustrates the structuring of damage history information. Specifically, Figure 8 shows an example where the damage history structuring unit 1200 uses an LLM (Large Language Model) to structure the damage history information 2100. The damage history structuring unit 1200 inputs the damage history information 2100 and the damage classification master 2400 (which serves as a condition for structuring the damage history information 2100) into the LLM, and extracts the structured damage history information 2600. The LLM is a language model constructed using a large amount of data and deep learning technology.

[0048] When the damage report information 2100, "On February 7, 2021, production of automobiles in the southern United States is expected to be affected" (Figure 8, top), is used as input to LLM, the damage classification master 2400 (Figure 8, middle) becomes the input condition (prompt) for structuring. The output from LLM is the structured damage report information 2600 (Figure 8, bottom). The output includes the date "2021 / 2 / 10", company name "Company X", industry "machine parts", country "United States", region "Texas", cause "cold wave", affected area "production capacity", damage scale "0%", and damage period "45 days". The output itself is nothing more than a list of words for each item. However, the user can quickly understand the key points of the damage from this output. The damage performance structuring unit 1200 extracts, for example, the target of the damage "production capacity", the scale of the damage "0%", and the duration of the damage "45 days" (Figure 8, bottom) from the natural language "expected to completely halt production for about a month and a half" (Figure 8, top).

[0049] When predicting and evaluating damage for each risk scenario (step S30), it is desirable to use methods such as statistical processing and machine learning, as described in step S340 below, in order to estimate the scale and duration of damage that will occur when the risk materializes. However, it is not possible to estimate the scale and duration of damage by performing statistical processing, machine learning, etc. on text data written in natural language, such as the damage history information 2100. Therefore, the damage history structuring unit 1200 has significance in structuring the damage history information 2100, which is text data written in natural language, etc., and converting it into a table format that can be processed by statistical processing, machine learning, etc. The explanation returns to Figure 4.

[0050] In step S130, the damage record structuring unit 1200 filters the damage record structuring information 2600 by company name and deletes duplicate data. For example, in the damage record structuring information 2600 in Figure 7, if there are multiple records with the same company name (field 2602), the damage record structuring unit 1200 deletes all but one duplicate record. Filtering means deleting duplicate or unnecessary information.

[0051] This prevents the apparent increase in the influence of specific damage data 2100 when estimating the scale and duration of damage, even when the same damage data 2100 is obtained from different data sources, thereby preventing a decrease in estimation accuracy. Note that the method for determining data duplication is not limited to this; methods including other fields of the damage data structured information 2600, clustering, etc., may also be used. After the completion of the processing in step S130, the series of processes performed by the damage data structuring unit 1200 in step S10 are terminated.

[0052] (Details of Step S20) Figure 9 is a flowchart showing the detailed processing of step S20 in Figure 3. In step S210, the damage history abstraction unit 1300 retrieves the damage history structured information 2600 and the risk response master 2500 stored in the data storage unit 2000. Although we are still in the middle of Figure 9, the explanation will now move on to Figure 10.

[0053] (Risk response master) Figure 10 shows an example of the risk response master 2500. The information collection and management unit 1100 acquires the risk response master 2500, for example, through the input / output interface unit 1600, either by manual input or data upload.

[0054] The Risk Response Master 2500 stores industry (column 2501), affected area (column 2502), and risk type value (column 2503) in an interconnected manner. The risk type value (column 2503) here is a binary value ("1" or "0") for each risk type. Examples of risk types include "A: Earthquake, flood, drought", "B: Economic conflict, protectionism", "C: Political conflict, demonstrations", "D: Regional armed conflict", and "E: Terrorist attack". Here, "1" indicates that the occurrence of a risk belonging to that risk type will affect the affected area (column 2502). "0" indicates that the occurrence of a risk belonging to that risk type will not affect the affected area (column 2502).

[0055] For example, in the "machine parts" industry, the affected areas are "parts procurement, production capacity, inventory loss, transportation lead time, transaction review, and supply disruption." The risk type value for risk type "A: earthquake, flood, and drought" is "1,1,1,1,0,0." This indicates that earthquakes, floods, and droughts affect parts procurement, production capacity, inventory loss, and transportation lead time, but do not affect transaction review or supply disruption. In addition to the binary values ​​mentioned above, the risk type value may also be a real number that takes continuous values ​​within the range of "0" to "1."

[0056] As mentioned above, the risk types in Figure 10 and the causes in Figure 7 are different concepts. For example, "cold wave" as a cause can be concretely imagined by users as a term described in the news. On the other hand, a user who imagines a cold wave may not simultaneously imagine an earthquake, flood, or drought. However, in the sense that the damage caused by "earthquake, flood, and drought" as a single risk type is similar to the damage caused by "cold wave," "earthquake, flood, and drought" can be said to be a higher-level concept than "cold wave." The role of Risk Response Master 2500 is to fill in the so-called "imaginary gap" for users who can imagine a cold wave but cannot simultaneously imagine an earthquake, flood, or drought.

[0057] Furthermore, the damage history abstraction unit 1300 can consider a cause as a risk and identify the risk type to which that risk belongs. For example, the damage history abstraction unit 1300 can consider the cause "flood" as the risk "flood" and identify the risk type "A: earthquake, flood, drought" to which the risk "flood" belongs. Similarly, the damage history abstraction unit 1300 can consider the cause "cold wave" as the risk "cold wave" and identify the risk type "A: earthquake, flood, drought" to which the risk "cold wave" belongs. In short, risk type "A" represents natural disasters in general. The explanation returns to Figure 9.

[0058] In step S220, the damage history abstraction unit 1300 uses the risk response master 2500 to supplement the damage history structured information 2600 with information on possible risks for each record in the damage history structured information 2600, and generates damage history abstraction information 2700. Although we are still in the middle of Figure 9, the explanation will now move on to Figure 11.

[0059] (Abstract information on damage records) Figure 11 shows an example of the damage history abstraction information 2700. The damage history abstraction information 2700 stores the industry (column 2701), country (column 2702), region (column 2703), affected area (column 2704), damage scale (column 2705), damage period (column 2706), and risk type value (2707) in an interrelated manner. The damage history abstraction information 2700 complements the causes of damage that have actually occurred in the past by including risks that have not yet occurred but may occur in the future. Comparing Figure 11 with Figure 7, the causes in Figure 7 are replaced by risk type values ​​in Figure 11.

[0060] For example, record 2610 in Figure 7 stores the industry "machine parts" and the affected area "production capacity". Record 2504 in Figure 10 stores the industry "machine parts", the affected area "production capacity", and the risk type value "1,1,1,1,1,...". As a result, record 2708 in Figure 11 will store the industry "machine parts", the affected area "production capacity", and the risk type value "1,1,1,1,1,...".

[0061] This allows us to extract, for example, risk categories (natural disasters) such as "earthquakes, floods, and droughts" as higher-level concepts, based on the cause of damage being a cold wave. Alternatively, let's assume, for example, that there has never been armed conflict in Texas, USA. Armed conflict as a cause of damage does not actually exist. However, even in this case, by modifying the contents of Risk Response Master 2500, in extreme cases, it becomes possible to treat damage information caused by a cold wave as damage information that could potentially be caused by armed conflict.

[0062] This is because, in this case, the target of the damage is production capacity. Three scenarios can be considered: production facilities being damaged and production stopping due to a cold wave, production facilities being damaged and production stopping due to an earthquake, and production facilities being damaged and production stopping due to armed conflict. These three scenarios are considered to be the same in that production facilities are damaged and production stops.

[0063] When predicting and evaluating damage for each risk scenario in step S30 with respect to the damage history structured information 2600, if there is no record in the damage history structured information 2600 for a given risk, it is not possible to estimate the scale and duration of damage for that risk for which no record exists. Therefore, by generating damage history abstraction information 2700, even if there is no record in the damage history structured information 2600 for a given risk, other records can be treated as possible damage for that given risk. As a result, in step S340, described later, it becomes possible to predict and evaluate damage for each risk scenario through statistical processing, machine learning, etc. The explanation returns to Figure 9. After the completion of the processing in step S220, the series of processes performed by the damage history abstraction unit 1300 in step S20 are completed.

[0064] (Relationship between cause and risk type) As stated above, the distinction between cause and risk is a matter of perspective, and the two are essentially the same. Furthermore, risk type is a higher-level concept of risk. However, more generally, risk type is sufficient if it is associated with a cause, even if it is not a higher-level concept of the cause. Users want to understand other causes they might not easily notice by using the concept of risk type. Therefore, risk type often becomes a higher-level concept of the cause.

[0065] (Details of Step S30) Figure 12 is a flowchart showing the detailed processing of step S30 in Figure 3. In step S310, the risk scenario evaluation unit 1400 acquires the supply chain information 2200, risk scenario information 2300, and damage history abstraction information 2700 stored in the data storage unit 2000. Although we are still in the middle of Figure 12, the explanation will now move on to Figures 13 and 14.

[0066] (Supply chain information) Figure 13 shows an example of supply chain information 2200. The supply chain information 2200 stores the supplier (column 2201), industry (column 2202), country (column 2203), region (column 2204), and items handled (column 2205) in an interconnected manner. The information collection and management unit 1100 automatically acquires the information necessary to create the supply chain information 2200 from EDI or publicly available company information via the input / output interface unit 1600. The information collection and management unit 1100 stores the completed supply chain information 2200 in the data storage unit 2000. At this time, the information collection and management unit 1100 adjusts the format of the acquired data to match the format of the information in each column of Figure 13, if necessary.

[0067] Generally, suppliers are connected hierarchically, such as first-tier subcontractors, second-tier subcontractors, third-tier subcontractors, and so on. Therefore, the supplier column 2201 may also store the hierarchical distance relative to the company as a buyer, such as "○th tier". Furthermore, a user may perform this embodiment from the position of multiple buyers. Therefore, the supply chain information 2200 may have a separate buyer column that lists the names of the buyers.

[0068] (Risk Scenario Information) Figure 14 shows an example of risk scenario information 2300. Risk scenario information 2300 stores the country (column 2301), region (column 2302), risk (column 2303), and risk type (column 2304) in an interrelated manner. As is clear from Figure 14, one record of risk scenario information 2300 corresponds to one risk scenario. A risk scenario is a risk that the user anticipates in the future, associated with its higher-level concepts of risk type and location.

[0069] The information collection and management unit 1100 may purchase disaster prevention and crisis management information, etc., from insurance companies, evaluation organizations, etc., via the input / output interface unit 1600, and create risk scenario information 2300. The explanation returns to Figure 12.

[0070] In step S320, the risk scenario evaluation unit 1400 filters the supply chain information 2200 by risk type, country, and region for each record of the risk scenario information 2300, and similarly filters the damage history abstraction information 2700. Although we are still in the middle of Figure 12, the explanation will now move on to Figure 15.

[0071] (Filtering in step S320) Figure 15 illustrates the filtering process in step S320 of Figure 12. The risk scenario evaluation unit 1400 filters each of the supply chain information 2200 and the damage history abstraction information 2700 for each record in the risk scenario information 2300. The filtering conditions (values ​​for each item to be kept without being deleted) are the records in the risk scenario information 2300.

[0072] First, let's look at Figure 15(a). One record that makes up the risk scenario information 2300 stores the country "United States", region "Texas", risk "flood", and risk type "A". The risk scenario evaluation unit 1400 filters the supply chain information 2200 using this record as a filtering condition. Compared to the supply chain information 2200a before filtering, the record for the country "Japan" has been deleted from the supply chain information 2200b after filtering. This is because the filtering condition includes the country "United States" but does not include the country "Japan".

[0073] Next, let's look at Figure 15(b). One record that makes up the risk scenario information 2300 stores the country "United States", region "Texas", risk "flood", and risk type "A", similar to Figure 15(a). The risk scenario evaluation unit 1400 filters the damage performance abstraction information 2700 using this record as a filtering condition. Compared to the damage performance abstraction information 2700a before filtering, the record for the country "Mexico" has been deleted from the damage performance abstraction information 2700b after filtering. Furthermore, in the risk type value column, other sub-fields other than "A" have been deleted. This is because the filtering condition includes the country "United States" and risk type "A", but does not include the country "Mexico" and risk types "B, C, D, E, ...". The explanation returns to Figure 12.

[0074] In step S330, the risk scenario evaluation unit 1400 filters the damage history abstraction information 2700 by industry, country, and region for each supplier in the supply chain information 2200. Although we are still in the middle of Figure 12, the explanation will now move on to Figure 16.

[0075] (Filtering in step S330) Figure 16 illustrates the filtering in step S330 of Figure 12. The risk scenario evaluation unit 1400 filters the damage history abstraction information 2700b filtered in step S320 for each record of the supply chain information 2200b filtered in step S320.

[0076] For example, the risk scenario evaluation unit 1400 filters the damage history abstraction information 2700b in Figure 15(b) using one record of the supply chain information 2200b in Figure 15(a) as a filtering condition. One record that makes up the supply chain information 2200b stores the supplier "Company A", industry "Machine Parts", country "United States", region "Texas", and handled item "Motor". Compared to the damage history abstraction information 2700b before filtering, the record for the industry "Raw Materials" has been deleted from the damage history abstraction information 2700c after filtering. This is because the filtering condition includes the industry "Machine Parts" but does not include the industry "Raw Materials".

[0077] The filtered damage history abstraction information 2700c does not have a supplier column. However, the risk scenario evaluation unit 1400 associates supplier "Company A" with the damage history abstraction information 2700c. This is because the user will learn about the potential damage that Company A, as a supplier, might suffer through the damage history abstraction information 2700c in Figure 16. The explanation returns to Figure 12.

[0078] In step S340, the risk scenario evaluation unit 1400 calculates the scale and duration of damage caused by the risk for each affected target, and generates risk scenario evaluation information 2800 based on the supply chain information 2200, risk scenario information 2300, and the calculation results. Although we are still in the middle of Figure 12, the explanation will now move on to Figure 17.

[0079] (Risk Scenario Assessment Information) Figure 17 shows an example of Risk Scenario Assessment Information 2800. Risk Scenario Assessment Information 2800 stores the following information in relation to each other: parts (column 2801), suppliers (column 2802), risks (column 2803), affected objects (column 2804), average damage scale (column 2805), deviation of damage scale (column 2806), average damage period (column 2807), and deviation of damage period (column 2808). The explanation returns to Figure 12.

[0080] The risk scenario evaluation unit 1400 uses the filtered damage history abstraction information 2700c from step S330 to perform statistical processing for each affected target and calculate the average damage scale, the deviation of the damage scale, the average damage period, and the deviation of the damage period. Variance can be used as an indicator of deviation. Note that the calculation method is not limited to this method, and may be calculated using methods such as machine learning.

[0081] The risk scenario evaluation unit 1400 repeats the processing in steps S330 and S340 for each supplier (outer loop), and in step S340, it repeats statistical processing for each affected entity (inner loop). As a result, for example, the two records in the damage history abstraction information 2700c in the lower part of Figure 16 (records for Company A) are statistically aggregated into the first record in the risk scenario evaluation information 2800 in Figure 17. After the processing in step S340 is completed, the series of processes performed by the risk scenario evaluation unit 1400 ends in step S30.

[0082] (Screen example) Figure 18 shows an example of the risk scenario evaluation results screen 3000. In step S40, the risk scenario evaluation results display unit 1500 displays the risk scenario evaluation results screen 3000 on the display device 15 or terminal device 40. The contents of the risk scenario evaluation results screen 3000 are almost the same as the risk scenario evaluation information 2800. From the risk scenario evaluation information 2800 in Figure 17, for example, the following can be seen.

[0083] The component "motor" is supplied by supplier "Company A". When the risk "flood" materializes, the potential damages could include "production capacity," "transportation lead time (LT)," and "inventory losses." When the target of the damage is production capacity, the average scale of the damage is 30%, the deviation of the scale of the damage is ±10%, the average duration of the damage is 30 days, and the deviation of the duration of the damage is ±15 days.

[0084] (Effects of the embodiment) The Risk Scenario Evaluation System 1000 allows users (buyers) to anticipate risks faced by suppliers, which could not be quantitatively evaluated through conventional manual methods, and to estimate the scale and duration of damage that suppliers would suffer from troubles and disasters, even during normal times. Furthermore, it prevents omissions and oversights in supplier risk management tasks that were previously performed manually by buyers due to the buyer's subjective perception, and enables risk management for risks that the buyer could not have anticipated. As a result, buyers can formulate countermeasures in advance to avoid risks in a manner that is appropriate to the expected targets, scale, and duration of damage based on the risks faced by suppliers.

[0085] It should be noted that the present invention is not limited to the embodiments described above, and various modifications are included. For example, the embodiments described above are described in detail for the purpose of clearly illustrating the present invention, and the present invention is not necessarily limited to having all the configurations described. Furthermore, it is possible to add, delete, or replace some of the configurations of the embodiments described above with other configurations.

[0086] Furthermore, each of the aforementioned configurations, functions, processing units, processing means, etc., may be implemented in hardware, either partially or entirely, by designing them as integrated circuits, for example. Alternatively, each of the aforementioned configurations, functions, etc., may be implemented in software by a processor interpreting and executing a program. Information such as programs, tables, and files that implement each function can be stored in recording devices such as memory, hard disks, SSDs, or recording media such as IC cards, SD cards, or DVDs. Also, control lines and information lines are shown only if deemed necessary for explanation, and not all control lines and information lines are necessarily shown in the product. In practice, it can be assumed that almost all configurations are interconnected. [Explanation of symbols]

[0087] 1000 Risk Scenario Assessment System 1100 Information Gathering and Management Department 1200 Damage Record Structuring Department 1300 Damage Record Abstraction Department 1400 Risk Scenario Assessment Department 1500 Risk Scenario Evaluation Result Display Section 1600 Input / Output Interface Section 2000 Data Storage Unit 2100 Damage Record Information 2200 Supply Chain Information 2300 Risk Scenario Information 2400 Damage Classification Master 2500 Risk Response Master 2600 Structured Information on Damage Records 2700 Abstract information on damage history 2800 Risk Scenario Assessment Information

Claims

1. The cause, scale, and duration of damage in the supply chain are recorded in natural language text from damage history information. Based on a damage classification master in which the type of damage to be of interest, the target of the damage, the method for defining the scale of the damage, and the method for defining the period of the damage are stored, A damage performance structuring unit generates damage performance structured information in which the aforementioned affected targets, the scale of the damage, and the period of the damage are stored in a format usable as a database. Based on a risk response master in which risk type values, which are indicators of the magnitude of the impact that the risk type associated with the cause of the damage has on the affected object, From the aforementioned structured information on damage records, A damage performance abstraction unit generates damage performance abstraction information in which the risk type value is stored in relation to the aforementioned target of damage, the aforementioned scale of damage, and the aforementioned period of damage, A risk scenario evaluation system characterized by having the following features.

2. From the aforementioned abstract information on damage records, As a condition for limiting the records of the aforementioned damage history abstraction information, the risk type is based on the risk scenario information in which it is stored, and the supplier is based on the supply chain information in which it is stored. The system includes a risk scenario evaluation unit that generates risk scenario evaluation information in which statistical values ​​of the scale of the damage and statistical values ​​of the duration of the damage are stored in relation to the supplier and the affected object. A risk scenario evaluation system according to claim 1, characterized by the following:

3. The system includes a risk scenario evaluation result display unit that displays the aforementioned risk scenario evaluation information. The risk scenario evaluation system according to claim 2, characterized by the following:

4. The aforementioned victims are: This includes at least one of the following: parts procurement, production capacity, inventory losses, transportation lead times, trade review, and supply suspension. A risk scenario evaluation system according to claim 1, characterized by the following:

5. The aforementioned damage performance structuring unit is: When generating the aforementioned structured information on damage records, Large Language Models (LLM) are used. A risk scenario evaluation system according to claim 1, characterized by the following:

6. The aforementioned risk scenario assessment information is, The statistical values ​​for the scale of damage include the average and deviation of the scale of damage, and the statistical values ​​for the period of damage include the average and deviation of the period of damage. The risk scenario evaluation system according to claim 2, characterized by the following:

7. The damage history structuring section of the risk scenario evaluation system is: The cause, scale, and duration of damage in the supply chain are recorded in natural language text from damage implementation information. Based on a damage classification master in which the type of damage to be of interest, the target of the damage, the method for defining the scale of the damage, and the method for defining the period of the damage are stored, The system generates structured information on damage records in which the aforementioned target of damage, the scale of damage, and the period of damage are stored in a format that can be used as a database. The damage history abstraction unit of the aforementioned risk scenario evaluation system is: Based on a risk response master in which risk type values, which are indicators of the magnitude of the impact that the risk type associated with the cause of the damage has on the affected object, From the aforementioned structured information on damage records, To generate damage history abstraction information in which the risk type value is stored in relation to the aforementioned target of damage, the aforementioned scale of damage, and the aforementioned period of damage, A risk scenario assessment method characterized by the following.

Images