Risk assessment system, risk assessment method, and risk assessment program

Abstract

The present invention relates to a risk assessment system, a risk assessment method, and a risk assessment program. [Solution] The risk assessment system 1 comprises a storage unit, a detection unit, a name matching processing unit, and a determination unit. The storage unit stores name matching conditions and determination conditions. The detection unit detects interfering elements from the input content related to exports based on predetermined conditions and identifies appropriate characters, which are the input content from which the interfering elements have been removed. The name matching processing unit performs name matching processing related to export risk based on the name matching conditions and appropriate characters to identify name matching information and obtain related information of the name matching information. The determination unit determines the export risk based on the determination conditions and related information.

Description

Technical Field

[0001] The present invention relates to a risk determination system, a risk determination method, and a risk determination program.

Background Art

[0002] Exports of goods, technologies, etc., as well as the provision of technologies to non-residents (deemed exports) are regulated globally under Japanese laws such as the "Foreign Exchange and Foreign Trade Act." For this reason, companies and research institutions expend a great deal of effort in verifying whether export targets and business partners are subject to regulations.

[0003]

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] By the way, when the user who manages exports and the applicant are different, the applicant may add interfering elements to the input content or use common names or jargon in order to escape export management checks, but Patent Document 1 does not consider such cases.

[0006] This invention has been made in view of the problems of the prior art described above, and its purpose is to provide a novel technology that more effectively determines export risk from the input content related to each export control. [Means for solving the problem]

[0007] To solve the above problems, the present invention provides a risk assessment system for determining export risk, The risk assessment system comprises a storage unit, a detection unit, a name matching processing unit, and a determination unit. The aforementioned storage unit stores the matching conditions and the determination conditions. The detection unit detects interfering elements from the export-related input content based on predetermined conditions, identifies appropriate characters which are the input content from which the interfering elements have been removed, The name matching processing unit performs the name matching process related to the export risk based on the name matching conditions and the appropriate characters to identify the name matching information, and obtains related information of the name matching information. The determination unit determines the export risk based on the determination conditions and the related information.

[0008] Furthermore, the present invention relates to a risk determination method for which a risk determination system for determining export risk is executed, The risk assessment system comprises a storage unit, a detection unit, a name matching processing unit, and a determination unit. The memory unit stores the matching conditions and the determination conditions, The detection unit detects interfering elements from the input content related to exports based on predetermined conditions, and identifies appropriate characters which are the input content from which the interfering elements have been removed. The name matching processing unit performs a name matching process related to the export risk based on the name matching conditions and the appropriate characters to identify the name matching information and obtain related information of the name matching information. The determination unit includes the step of determining the export risk based on the determination conditions and the related information.

[0009] Furthermore, the present invention relates to a risk assessment program for determining export risk, The computer is configured to function as a storage unit, a detection unit, a name matching processing unit, and a determination unit. The aforementioned storage unit stores the matching conditions and the determination conditions. The detection unit detects interfering elements from the export-related input content based on predetermined conditions, identifies appropriate characters which are the input content from which the interfering elements have been removed, The name matching processing unit performs the name matching process related to the export risk based on the name matching conditions and the appropriate characters to identify the name matching information, and obtains related information of the name matching information. The determination unit determines the export risk based on the determination conditions and the related information.

[0010] This configuration allows for the detection of interfering elements from the input content, and efficient determination of export risk through appropriate data matching. For example, by detecting intentional whitespace characters, removing them, and then performing further data matching, export risk can be determined with high accuracy.

[0011] In a preferred embodiment of the present invention, the name matching processing unit performs a name matching process related to export risk to identify the name matching information based on the items of the input content relating to export and / or the name matching conditions for each language of the input content, and the appropriate characters, and obtains a plurality of related pieces of information of the name matching information for each item.

[0012] This configuration allows for appropriate data matching processes to be performed for each item, enabling efficient assessment of export risks.

[0013] In a preferred embodiment of the present invention, the storage unit includes detection conditions, Based on the detection conditions, the detection unit detects the interfering characters constituting the interfering element and / or the arrangement of said interfering characters in the input content from the input content relating to the export, and further calculates the interfering nature of the interfering element. When the interference level is equal to or higher than a predetermined threshold, the clustering processing unit further acquires the correlation information between the clustering information and the input content.

[0014] With such a configuration, it is possible to acquire correlation information including the original input content from the interference level.

[0015] In a preferred form of the present invention, the storage unit includes detection conditions. Based on the detection conditions, the detection unit detects interfering characters that constitute the interference element and the arrangement of the interfering characters in the input content regarding the export, and further calculates the interference level of the interference element. When the interference level is equal to or higher than a predetermined threshold, the clustering processing unit further acquires the correlation information based on similar words of the appropriate characters for which the clustering processing has been executed.

[0016] With such a configuration, it is possible to acquire correlation information including similar words of appropriate characters from the interference level.

[0017] In a preferred form of the present invention, the detection conditions include interference kana character conditions. Based on the interference kana character conditions, the detection unit detects kana character terms that can be converted into Chinese characters and constitute the interference element from the input content regarding the export, and further calculates the interference level. When the interference level is equal to or higher than a predetermined threshold, the clustering processing unit acquires the correlation information between the clustering information and the input content.

[0018] With such a configuration, it is possible to appropriately perform risk determination on the input content that attempts to avoid detection of determination in kana notation.

[0019] In a preferred form of the present invention, the clustering processing unit searches for a plurality of pieces of the correlation information, calculates the degree of correlation between the content of each of the plurality of pieces of the correlation information and the appropriate characters, and acquires the top predetermined number of pieces of the correlation information having a high degree of correlation among the plurality of pieces of the correlation information and the determination conditions.

[0020] By adopting such a configuration, it is possible to obtain and determine information with relatively high relevance from a vast amount of information, which is also conducive to improving processing efficiency and reducing time.

[0021] In a preferred form of the present invention, the risk determination system further comprises a generation unit. The generation unit generates an explanatory text of the export risk based on the input content, the related information, and the export risk.

[0022] By adopting such a configuration, it is possible to present the reason for determination to the user and efficiently perform the work of export management.

Advantages of the Invention

[0023] According to the present invention, by detecting interference elements in the input content and performing predetermined processing, a novel technology related to the risk determination system can be provided.

Brief Description of the Drawings

[0024] [Figure 1] A block diagram of the system configuration according to an embodiment of the present invention is shown. [Figure 2] An example of the hardware configuration of an information processing apparatus and a terminal according to an embodiment of the present invention is shown. [Figure 3] A flowchart of the processing procedure of the risk determination system according to an embodiment of the present invention is shown. [Figure 4] An example of each piece of information used in the risk determination system according to an embodiment of the present invention is shown. [Figure 5] A schematic image of the clustering process executed by the risk determination system according to an embodiment of the present invention is shown. [Figure 6] An example of the clustering conditions used in the risk determination system according to an embodiment of the present invention is shown. [Figure 7] An image diagram of a partial processing procedure of the risk determination system according to an embodiment of the present invention is shown. [Figure 8]An example of the judgment conditions used in the risk assessment system according to one embodiment of the present invention is shown. [Figure 9] An example of the display processing result on a terminal relating to one embodiment of the present invention is shown. [Modes for carrying out the invention]

[0025] Further details will be provided below with reference to the attached drawings. Preferred embodiments are shown in the drawings. However, many different embodiments are possible and are not limited to those described herein.

[0026] For example, in this embodiment, the configuration and operation of the risk assessment system will be described, but similar effects can be achieved by the execution method (steps), apparatus, computer program, etc. The program in this embodiment may be provided as a non-transient recording medium that can be read by a computer, or it may be provided so that it can be downloaded from an external server.

[0027] Furthermore, in this embodiment, "part" may also include, for example, hardware resources implemented by circuits in a broad sense, and the information processing of software that can be specifically realized by these hardware resources.

[0028] In this embodiment, "information" can be represented, for example, by the physical value of a signal representing voltage or current, the high or low value of a signal as a set of binary bits consisting of 0s or 1s, or by a quantum superposition (so-called qubit), and communication and calculations can be performed on a circuit in a broad sense.

[0029] In a broad sense, a circuit is a circuit realized by appropriately combining circuits, circuits (including processors and memory). That is, it includes CPUs (Central Processing Units), GPUs (Graphics Processing Units), LSIs (Large Scale Integration), ASICs (Application Specific Integrated Circuits), FPGAs (Field-Programmable Gate Arrays), etc.

[0030] <System Configuration> Figure 1 is a block diagram showing the system configuration according to one embodiment of the present invention. As shown in Figure 1, the risk assessment system 1 includes an information processing device 10 and a database DB. The risk assessment system 1 is configured to communicate with multiple user terminals 2 (reference numerals 2(a) to 2(c) in Figure 1), various external DBs 3 (reference numerals 3(a) to 3(c)), and a generation device 4 via a network NW.

[0031] The information processing device 10 operates as a server, the user terminal 2 is a terminal used by export managers of companies and organizations to view export risk assessments, etc., and the external DB 3 is a database from which the information processing device 10 acquires information related to government and organizational export regulations. In this specification, the term external DB 3 is not limited to relational databases, but is also used in a broad sense to include information sources accessible from outside, such as websites, news sites, and search engine results published on the internet.

[0032] The generation device 4 is an external server device or the like that controls the processing by the generation model. Alternatively, the information processing device 10 may store the generation model and generate responses without using external devices. Furthermore, the risk assessment system 1 can also be configured with a generation unit 104 (described later) without using a generation model.

[0033] In this embodiment, the network NW is an IP (Internet Protocol) network, but there are no restrictions on the type of communication protocol, nor on the type or size of the network.

[0034] Furthermore, general-purpose server computers or personal computers can be used as the information processing device 10 and the generation device 4. It is also possible to implement the functional components described later on multiple computers to constitute the risk assessment system 1.

[0035] User terminal 2 can be a smartphone, tablet, personal computer, wearable device, etc. User terminal 2 stores a risk assessment program for the user. In this embodiment, information related to risk assessment is displayed on user terminal 2.

[0036] <Hardware Configuration> Figure 2(a) shows an example of the hardware configuration of the information processing device 10. The information processing device 10 comprises a control unit 11, a storage unit 12, and a communication unit 13 as its hardware configuration.

[0037] The control unit 11 includes one or more processors such as a CPU, and controls the entire operation of the information processing device 10 by executing the risk determination program, OS, and other applications according to the present invention.

[0038] The storage unit 12 is an HHD, SSD, ROM, RAM, etc., and stores the risk determination program according to the present invention and data used by the control unit 11 when executing processing based on the program. The control unit 11 executes processing based on the risk determination program stored in the storage unit 12, thereby realizing the functional configuration described later.

[0039] The communication unit 13 performs communication control with the network NW and provides inputs necessary for operating the information processing device 10, as well as outputs related to the operation results.

[0040] Figure 2(b) shows an example of the hardware configuration of terminal 90 (user terminal 2 in Figure 1). Terminal 90 comprises a control unit 91, a storage unit 92, a communication unit 93, an input unit 94, and an output unit 95 as its hardware configuration.

[0041] The control unit 91 of the terminal 90 includes one or more processors such as a CPU and controls the entire operation of the terminal 90. The storage unit 92 of the terminal 90 is an HDD, SSD, ROM, RAM, etc., and stores the user risk assessment program described above, as well as data used by the control unit 91 when executing processing based on the program.

[0042] The communication unit 93 of terminal 90 controls network communication. The input unit 94 of terminal 90 is a touch panel, mouse, keyboard, etc., which inputs user operation requests to the control unit 91. The output unit 95 of terminal 90 is a display, etc., which displays the results of processing by the control unit 91.

[0043] The terminal 90 may be configured to access the risk assessment system 1, which is provided as a web application, by running a web application instead of the user's risk assessment program.

[0044] <Functional Configuration> The information processing device 10 provides the risk assessment system 1 by executing a risk assessment program. As shown in Figure 2(a), the information processing device 10 has a functional configuration comprising a detection unit 101, a name matching processing unit 102, a determination unit 103, a generation unit 104, and a display processing unit 105. These are the concrete realization of information processing by software (stored in the storage unit 12) by hardware (control unit 11, etc.).

[0045] The detection unit 101 detects various information from the input content received from the user terminal 2 or the like. The detection unit 101 detects interfering elements. In this embodiment, the detection unit 101 detects interfering elements from the input content related to exports based on predetermined conditions and identifies the appropriate characters which are the input content from which the interfering elements have been removed. The detection unit 101 can also calculate the degree of interference of the interfering elements.

[0046] The data matching processing unit 102 identifies data matching information. In this embodiment, the data matching processing unit 102 identifies data matching information by performing data matching processing related to export risk based on data matching conditions and appropriate characters, and obtains related information for said data matching information. The data matching processing unit 102 performs data matching processing related to export risk based on data matching conditions that have items in the input content and / or for each language of said input content, and appropriate characters.

[0047] Furthermore, the name matching processing unit 102 acquires related information based on the interference level calculated by the detection unit 101. The name matching processing unit 102 can also acquire multiple pieces of related information and calculate the degree of correlation between the content of each piece of related information and the appropriate character.

[0048] The determination unit 103 determines the export risk. In this embodiment, the determination unit 103 determines the export risk based on the determination conditions and related information. For example, the determination unit 103 calculates a score from the degree of agreement between the determination conditions and the related information, and determines the risk, such as "high alert," based on the score.

[0049] The generation unit 104 generates a response. In this embodiment, the generation unit 104 generates an explanation of the export risk based on the input content, related information, and export risk. For example, the generation unit 104 generates an explanation of the export risk based on the content of the related information website.

[0050] The display processing unit 105 processes the display screen of the user terminal 2, etc. In this embodiment, the display processing unit 105 processes the determination of the determination unit 103 and the export risk explanation text generated by the generation unit 104, etc.

[0051] <database DB> The database DB in Figure 1 stores input document information for the documents to be entered, item information for the document items, various detection conditions for the detection unit 101 to detect interfering elements, matching information identified after performing the matching process, judgment conditions for determining export risk, and other various information necessary for risk determination. Some or all of this information may be stored in the storage unit 12, etc., or some of this information may be stored in another database, etc.

[0052] The following explanation of the risk assessment system 1 and the processing performed by each functional element will be given with reference to Figures 3-9.

[0053] <Overview of the Risk Assessment System> The risk assessment system 1 according to this embodiment is a system that reduces the burden on the user and determines the export risk of the subject. For example, when a company exports a product or technology, it can determine the risk of it falling under domestic and international regulations based on its name, specifications, and destination. In the case of universities, etc., it can also determine the export risk based on the nationality and background of foreigners, etc., as a measure against "deemed export" when technology is illegally provided to foreigners such as international students.

[0054] Furthermore, it addresses malicious intent on the part of the export applicant to circumvent detection by conventional export control systems. By detecting and appropriately handling such disruptive elements, export risks can be accurately assessed.

[0055] Furthermore, the risk assessment system 1 goes beyond simply comparing the input content with the regulatory list; it also acquires relevant information, including web searches, enabling it to determine indirect export risks that cannot be anticipated based solely on the input content. While this embodiment is primarily intended for use in Japanese, it is compatible with languages ​​from around the world.

[0056] <Detection of interfering elements> Figure 3 is a flowchart showing the processing procedure of a risk assessment system 1 according to one embodiment. First, at S301 in Figure 3, the detection unit 101 of the information processing device 10 detects interfering elements from the input content received from the user. Specifically, it detects interfering elements based on the input content of the document entered by the user and the detection conditions.

[0057] In this embodiment, the user inputs documents related to export control. For example, document files such as export declarations, classification determinations, and resumes within an organization are assumed. These documents have items for each type of content to be entered, and the detection unit 101 detects elements that interfere with the input content for each item based on the detection conditions. In addition, the user can also input reference data such as drawings of the exported goods and safety data sheets.

[0058] Figures 4(a) to 4(g) show examples of various information and detection conditions stored by the storage unit 12 according to this embodiment. Input document information is managed by a document ID for each document, as shown in Figure 4(a). Item information is managed by an item ID, including items and document IDs, as shown in Figure 4(b). For example, the detection unit 101 identifies a document from the file name of the input document, and identifies an item from the name and location of the item described in the document.

[0059] As an example, the detection unit 101 first detects interfering characters based on the interfering character conditions. As shown in Figure 4(c), the interfering character conditions are managed by a condition ID, which includes the character code of the interfering character. Interfering characters include character codes for identifying characters that are not normally used and that humans or computers may not be able to correctly identify or may misidentify. For example, the commonly used Latin letter "A" and the Greek letter "A" are identified by different character codes. Such characters are called homoglyphs, and because they are difficult for the human eye to distinguish and computers also recognize them as different characters, they can be misused to evade keyword detection.

[0060] Furthermore, in addition to the commonly used whitespace characters (full-width / half-width spaces created by the space key), there are many other types, including "zero-width characters" that have no visible width. Inserting such special whitespace characters into input content makes it difficult for conventional systems that perform specific string pattern matching to detect them correctly.

[0061] The risk assessment system 1 can also detect such characters as interfering elements by storing them as interfering character conditions.

[0062] In addition, as shown in Figures 4(d) to (e), the memory unit 12 stores interference placement conditions and interference kana character conditions. Interference placement conditions are conditions for detecting interference elements from the placement patterns of predetermined characters (e.g., interference characters) in the input content. Interference placement conditions also include conditions for detecting interference elements when interference characters are placed near a specific string (e.g., a restricted keyword). Interference kana character conditions are conditions for detecting interference elements when part or all of a predetermined word written in kanji is written in kana characters.

[0063] Furthermore, the storage unit 12 may store semantic interference conditions for detecting interference elements when there is inconsistency between an item and its input content, as shown in Figures 4(f) to 4(g), and document interference information for detecting interference elements by combining the input content of multiple items in a document. The detection unit 101 can also identify interference elements based on these conditions and various export-related information stored in the storage unit 12, as well as well-known statistical models.

[0064] If the risk assessment system 1 finds no interfering elements in the input (NO in Figure 3S302), it executes the data matching process described later. On the other hand, if an interfering element is detected in S301 (YES in S302), it performs processing related to that interfering element before the data matching process.

[0065] <Removal of interfering elements> In S303, the detection unit 101 removes the detected interfering elements and identifies appropriate characters. For example, blank characters and zero-width characters included in the interfering character conditions may be deleted, and in the case of homoglyphs or the like, they may be replaced with characters of the character code usually used.

[0066] <Calculation of interference property> In S304, the detection unit 101 calculates the interference property. The calculated interference property is used when acquiring related information described later.

[0067] Specifically, the interference property is calculated based on the interference scores (Figs. 3(c) to (g)) associated with each interference condition stored in the storage unit 12. For example, when one interfering character "A (U+0391)" is detected for a certain input content, "10" is added as the interference score. In one detection condition, when a plurality of interfering elements are detected, the one with the largest interference score among the plurality of interfering elements may be calculated as the interference score for the detection condition.

[0068] The detection unit 101 calculates the interference property based on the interference scores of the interfering elements detected in the entire input content. The interference property may be calculated, for example, in an additive form, or one or more detection conditions may be weighted and added together to calculate the interference property.

[0069] In this embodiment, as an example, the subsequent flow changes depending on whether the calculated interference property exceeds predetermined thresholds a_1 and a_2 (a_1 < a_2) (S305). If the interference property is less than a_1, or if it is a_1 or more and a_2 or less, the next process is advanced. If the interference property exceeds a_2, the input content (document) is regarded as already dangerous and the subsequent process is stopped. At this time, the detection unit 101 may send a notification requesting confirmation to the user side as a major incident.

[0070] <Execution of alignment processing> In S306, the clustering processing unit 102 of the information processing apparatus 10 executes clustering processing. Here, the clustering processing refers to the preliminary processing for accurately executing the identification of clustering information and the acquisition of related information, which will be described later. The clustering processing unit 102 executes clustering processing on appropriate characters and the input content determined as NO in S302.

[0071] FIG. 5 shows an example of the clustering processing in the present embodiment. As shown in the figure, the clustering processing unit 102 performs clustering processing based on different clustering conditions for each language and each item. For example, after performing preprocessing common to all items (such as converting full-width English characters to half-width characters and normalizing character codes), specific clustering processing is executed for corporate names, personal names, product names, etc. Also, instead of performing clustering processing at once, for example, like the personal name in the figure, by dividing the processing into multiple steps, the clustering processing can be accurately performed.

[0072] For example, the clustering processing unit 102 executes clustering processing common to all items. As an example, first, full-width characters of English letters and symbols are replaced with half-width characters, and uppercase letters are unified into lowercase letters. In the case of Japanese, conversely, half-width katakana are replaced with full-width hiragana. Next, characters that can be combined on Unicode are normalized by NFC (Normalization Form Canonical Composition), etc. For example, the characters "か" + "゛" are combined as "が".

[0073] Next, different clustering processing is executed for each item. When the item is the name of the export destination (corporate name), "株式会社" is replaced with "(株)", "有限公司" in Chinese is replaced with "Co.,Ltd", and "OOO (meaning limited liability company) in Russia is replaced with "LLC,". In the case of a personal name, honorifics in each country ("Mr.", "Dr.", "様", "氏", etc.) are removed, and the order of the surname and given name is swapped if necessary. When converting to English, it can be converted to "shi" etc. corresponding to the Hepburn system, or converted to the Kunrei system. It can also correspond to the pinyin in Chinese.

[0074] <Identification of Clustering Processing> In S307, the name matching processing unit 102 identifies the name matching information. Figure 6 shows an example of the name matching information stored in the storage unit 12 in this embodiment. The name matching processing unit 102 identifies the name matching information corresponding to the correct character for which the name matching process in S306 was performed.

[0075] For example, the corporate name / export destination name matching information is managed by a name matching ID, which includes an item ID, specific name, name after name matching processing (similar terms), address, group companies, etc. Furthermore, it may be associated with identification information such as a flag that identifies which list the specific name belongs to, or it may include name matching information that does not belong to any list. For example, a flag may be assigned to a list if the name listed in that list matches the name matching information in whole or partially.

[0076] Here, "list" includes Japan's "Export Trade Control Order Annex 1" and "Foreign User List," which are also used in determining export risks as described later, the U.S. Entity List, notifications from the Ministry of Economy, Trade and Industry and trade organizations in various countries, and other black / white lists of products, people, and export destinations created independently by the organization to which the user belongs. Some or all of these may be stored in the storage unit 12, or they may be retrieved from an external DB3 or web server as needed.

[0077] For example, if the result of the name matching process in S306 is "ABC Inc.," the name matching processing unit 102 identifies "American Big Company Inc.," as the specific name, along with other similar words, addresses, group companies, and the relevant list (List C in Figure 6). The same process is applied to the name matching information for other items.

[0078] The data matching processing unit 102 can also translate the input content into multiple languages ​​to identify the data matching information. For example, it can perform data matching processing based on Japanese input content, translate (convert) it into multiple foreign languages, and then identify the data matching information. By identifying the data matching information in multiple foreign languages ​​for a single input content, export risk assessment can be performed with greater accuracy.

[0079] <Searching using data matching information / Obtaining related information / Calculating relevance / Determining the degree of relevance> In steps S308 to S312 of Figure 3, the data matching processing unit 102 performs a search for related information to obtain it, and further calculates the degree of relevance of the obtained related information. The determination unit 103 then determines the export risk.

[0080] In this embodiment, if the interference level calculated in S305 is less than a_1, a search is performed using the specific name of the identified name matching information (S308). If it is between a_1 and a_2, the search is performed not only using the specific name but also including similar words associated with the name matching information (enhanced search, S309). Specifically, the name matching processing unit 102 inputs the specific name of the name matching information as a search query and executes the search. In addition, the search may be performed based on various information associated with the name matching information. Furthermore, the name matching processing unit 102 may execute the search using a search query that includes the interference elements detected in S301. If there is no name matching information to identify, the search is performed using appropriate characters after name matching processing. When searching with similar words, these similar words are not limited to those associated with the name matching information, but may also include those generated by, for example, generative AI or RAG (Retrieval-Augmented Generation).

[0081] In this embodiment, related information is retrieved using a "hybrid search" that combines keyword search, vector search, graph database search, etc. The search targets include databases, websites, external DBs, etc., and the retrieved related information includes URLs, document files, drawing files, etc. The data matching processing unit 102 retrieves a predetermined number of related information items based on the search content items. Figure 7 shows an image of the processing S308 to S311 in this embodiment.

[0082] As shown in the figure, the data matching processing unit 102 searches for related information, for example, by company name / export destination, and retrieves 200 entries from websites, etc. Based on the content of the related information, the data matching processing unit 102 calculates the degree of relevance and scores each piece of related information. The data matching processing unit 102 then sorts the various pieces of information in order of relevance and retrieves the top 50 pieces of related information with the highest relevance. Note that the number of items to search and the number of items to retrieve may differ for each item in the document being searched. Also, if there are multiple pieces of related information with the same degree of relevance, the sorting may be prioritized based on the creation date or update date of the related information.

[0083] Relevance can be calculated using methods such as the density of regulated keywords in the text, the frequency of occurrence of search query content, the degree of agreement between the estimated category of the search query and the document category, and the frequency of occurrence of numerical values ​​related to thresholds. In addition, relevance can be calculated using factors such as the degree of agreement of the document structure, the authority of the information (whether or not it is government information), and the recency of the update date.

[0084] Furthermore, the detection of interfering elements in S301 may change the ratio of search results per item and the parameters used to calculate relevance. For example, in the case of enhanced search, the weighting of relevance calculation based on the density of restricted keywords can be increased.

[0085] Then, in S312, the data matching processing unit 102 determines the export risk based on the acquired related information and the determination conditions. Figure 8 shows an example of the determination conditions according to this embodiment. The determination conditions are conditions for determining the export risk by comparing the content of the acquired related information with the various words described in each list stored in the storage unit 12. Specifically, the determination conditions are divided into list regulations and catch-all regulations, and the determination is made based on each item related to each regulation.

[0086] Here, the information in the list stored in the memory unit 12 is not limited to the original text data of courts and notices themselves, but may also be extracted and normalized to improve the accuracy and speed of matching, and converted and prepared into a format suitable for computer reading. Alternatively, the system may be configured to train a generative AI with this list and judgment conditions to determine export risk.

[0087] The determination of compliance with list regulations is made based on the item name, specification values ​​(specs, or numerical values ​​listed in drawings or safety data sheets), technology name, etc., included in the acquired relevant information, and the list of regulations in other countries, as well as the determination criteria for evaluating the degree of agreement. For example, in the case of determining the compliance of an item name and specification values, it is first determined whether the item is included in the list, and if so, whether the specification values ​​exceed the regulatory threshold. If the item is included in the list and exceeds the regulatory threshold, a determination score of 100 is calculated; otherwise, a score lower than 100 is calculated.

[0088] The judgment score calculated here is not limited to a numerical value; for example, it may be evaluated using four ranks established according to a unique criterion. For example, ranks S, A, B, and C could be set in descending order of concern, and a rank could be assigned if the product name matches the list and the specification value exceeds the threshold, A if it does not, B for a partial match, and C if neither applies.

[0089] In the case of determination by technology name, it is determined whether the technology in question is included in the list, and if it is included in the list, it is determined whether it is a manufacturing technology, a design technology, or a technology used, and a determination score is calculated.

[0090] The determination of whether or not an item is subject to catch-all regulations is made based on the destination country, intended use, names of trading partners (including corporations and universities), and personal names of the acquired relevant information, as well as the foreign user list, black / white list, military keyword list, and determination criteria. For example, in the case of a destination country, the determination is made based on whether or not the country is included in the list of countries of concern or in the list of countries receiving preferential export control. In the case of determination based on intended use, the determination is made based on whether or not military keywords are included in the content of the relevant information, or the number of such keywords. In the case of companies and universities, the determination is made not only based on whether or not they are included in the foreign user list, etc., but also on whether or not the company has a capital relationship with a company listed on the list. In the case of personal names, in addition to determining the person whose name is in question, the determination is made based on whether or not the company or university to which the person currently or previously belongs / is involved is included in the foreign user list, military-related organization, or black / white list.

[0091] The determination unit 103 calculates the overall export risk score by selecting the highest score among the determined determination scores. For example, if the determination score based on list determination is 50 and the determination score based on catch-all regulations is 75, the overall score will be determined to be 75.

[0092] Furthermore, the risk assessment according to this embodiment can assess not only risks corresponding to current regulations, but also risks that may become subject to regulations in the future. For example, even if an export product is not currently on the list, if a company or country related to that export product is included in the list, it may become subject to regulations in the future, and therefore the future risk (future risk assessment score) will be relatively larger (higher score) than the current risk assessment. <Generate the answer> In step S313 of Figure 3, the generation unit 104 of the information processing device 10 generates an explanatory text for the determination. Specifically, the generation unit 104 generates the explanatory text for the determination based on the input content, related information, and the determination.

[0093] Furthermore, in this embodiment, the name matching processing unit 102 acquires related information, and the determination unit 103 determines the input content. The name matching processing unit 102 acquires related information based on the correct characters that have been processed for name matching, and the determination unit 103 determines the export risk of the input content based on the determination conditions and the related information, and stores the determination in the storage unit 12, linking it to the input content. The generation unit 104 can use this information to generate an explanatory text for the determination. For example, the generation unit 104 can extract text data included in the content of the related information, etc., by text mining.

[0094] The generation unit 104 transmits the input content, related information, export risk, and the determination conditions used to determine the export risk to the generation device 4. The explanatory text generated by the generation device 4 is then displayed by the display processing unit 105 (described later), and the display processing result is transmitted to the user terminal 2.

[0095] The generation device 4 is configured to include a natural language processing (NLP) model. Preferably, the generation device 4 is configured to include a large language model (LLM). The natural language processing model enables the processing of data input as natural language by a computer. The type of natural language processing model used in this embodiment is not limited, but examples include ChatGPT (Chat Generative Pre-trained Transformers), BERT (Bidirectional Encoder Representations from Transformers), Gemini, and Claude. It is also possible to configure the generation device 4 using multiple devices. Furthermore, the information processing device 10 can be configured to include the generation device 4.

[0096] In this embodiment, the generation unit 104 inputs the input content, related information, export risk, and the determination conditions used to determine the export risk to the generation device 4, and generates the determination reason based on the output of the generation model. However, it is not always necessary to use the generation model, and the generation unit 104 of the information processing device 10 can also generate the answer support. For example, the information processing device 10 can store a predetermined algorithm or table for generating the answer support in the storage unit 12, and the generation unit 104 can use this algorithm or table to generate the determination reason.

[0097] <Display Processing> In S314 of Figure 3, the display processing unit 105 processes the input content to display the export risk and a description of the export risk. Figure 9 shows an example of the display processing result of the judgment result screen 200 on the user terminal 2. As shown in the figure, the judgment result screen 200 of the user terminal 2 consists of an input content display area 201, a judgment result display area 202, and a description display area 203.

[0098] The input content display area 201 displays the input content of documents, etc., entered into the information processing device 10, item by item. If the input content contains interfering elements, the appropriate characters with those interfering elements removed may be displayed.

[0099] The judgment result display area 202 displays the export risk of the input content determined by the judgment unit 103. In the same area 8, the export risk of the input content is displayed for comprehensive, list control, and catch-all control. In this figure, usability is improved by using numerical values ​​and pie charts for each export risk, but other configurations are also possible.

[0100] In the explanatory text display area 203, an explanatory text is displayed that explains why the export risk of the input content was determined as it was. In Figure 9, the explanatory text is displayed in separate areas: the list control explanatory text display area 203a and the catch-all control explanatory text display area 203b. However, the explanatory text may also be displayed in a single display area.

[0101] For example, the explanatory text display area 203 displays information such as which related information corresponds to which list or judgment condition. If the related information is a URL, the URL and the relevant section of the content displayed when accessing that URL are extracted and displayed. Similarly, in the case of a document file, the file name and the relevant section may be extracted and displayed. Furthermore, if the language of the related information is different, the generation unit 104 may translate it into a corresponding language (Japanese in this embodiment) and generate the explanatory text.

[0102] In addition, as shown in the judgment result display area 202 of Figure 9, if an interfering element is detected or if the input content matches a list, a flag (mark) may be added and the result displayed. For example, when the detection unit 101 detects an interfering element from the input content, a flag may be added to the input content, and the display processing unit 105 may perform display processing based on that flag.

[0103] Furthermore, by selecting "Details" for each area, more detailed evaluation results (what factors were used to award points) or more detailed explanatory text (such as a list of related information obtained) may be displayed.

[0104] As described above, the risk assessment system 1 according to the present invention can appropriately determine the export risk related to the input content by performing predetermined processing using the input content and matching conditions.

[0105] In this embodiment, we have described risk determination, which determines export risk from input content. However, the same effects as the present invention can be obtained when the risk determination system 1 is used in fields other than export risk determination.

[0106] In this embodiment, the processes of the detection unit 101, the name matching processing unit 102, and the determination unit 103 are processes performed within this system by the execution of the program of the present invention by hardware. However, it may also be a process in which a prompt is sent to the generation device 4 containing instructions to perform each process on the input content, interference conditions, appropriate characters, name matching conditions, and name matching identification information, and the processing results are obtained. Alternatively, this system may have a generation AI, and the generation AI may perform each process. Furthermore, different generation AIs may be used to execute each process, or multiple processes may be executed on a single generation AI. [Explanation of symbols]

[0107] 1. Risk Assessment System 2 User terminals 3 External DB 4 Generator 10 Information Processing Devices 11 Control Unit 12 Storage section 13 Communications Department 90 devices 91 Control Unit 92 Memory section 93 Communications Department 94 Input section 95 Output section 101 Detection unit 102 Processing Unit 103 Judgment section 104 Generation part 105 Display Processing Unit 200 Judgment result screen 201 Input content display area 202 Judgment result display area 203 Description display area NW Network

Claims

1. A risk assessment system for determining export risks, The risk assessment system comprises a storage unit, a detection unit, a name matching processing unit, and a determination unit. The aforementioned storage unit stores the matching conditions and the determination conditions. The detection unit detects interfering elements from the export-related input content based on predetermined conditions, identifies appropriate characters which are the input content from which the interfering elements have been removed, The name matching processing unit performs the name matching process related to the export risk based on the name matching conditions and the appropriate characters to identify the name matching information, and obtains related information of the name matching information. The determination unit determines the export risk based on the determination conditions and the related information. Risk assessment system.

2. The matching processing unit performs matching processing related to export risk based on the matching conditions for each item of the input content relating to the export and / or for each language of the input content, and the appropriate characters, to identify the matching information, and obtains multiple pieces of related information for each item of the matching information. The risk determination system according to claim 1.

3. The storage unit includes detection conditions, Based on the detection conditions, the detection unit detects the interfering characters constituting the interfering element and / or the arrangement of said interfering characters in the input content from the input content relating to the export, and further calculates the interfering nature of the interfering element. The name matching processing unit further acquires the name matching information and the input content and related information if the interference level is above a predetermined threshold. The risk determination system according to claim 1.

4. The storage unit includes detection conditions, Based on the detection conditions, the detection unit detects the interfering characters constituting the interfering element and the arrangement of said interfering characters in the input content from the input content relating to the export, and further calculates the interfering nature of the interfering element. The name matching processing unit further obtains the related information based on similar words of the appropriate characters for which the name matching process was performed, if the interference level is above a predetermined threshold. The risk determination system according to claim 1 or 3.

5. The aforementioned detection conditions include the condition for interfering kana characters, The detection unit, based on the interference kana character conditions, detects kana character terms that can be converted to kanji and constitute the interference element from the input content related to the export, and further calculates the interference level. The name matching processing unit, when the interference level is above a predetermined threshold, acquires the name matching information and the input content and related information. The risk determination system according to claim 4.

6. The name matching processing unit searches for multiple related information, calculates the degree of relevance between the content of each of the multiple related information and the appropriate character, and obtains the determination condition and a predetermined number of the top related information with the highest degree of relevance from among the multiple related information. The risk determination system according to claim 1.

7. The risk assessment system further comprises a generation unit, The generation unit generates an explanation of the export risk based on the input content, the related information, and the export risk. A risk assessment system according to any one of claims 1 to 3.

8. A risk assessment system for determining export risks is implemented, and this is a risk assessment method, The risk assessment system comprises a storage unit, a detection unit, a name matching processing unit, and a determination unit. The memory unit stores the matching conditions and the determination conditions, The detection unit detects interfering elements from the input content related to exports based on predetermined conditions, and identifies appropriate characters which are the input content from which the interfering elements have been removed. The name matching processing unit performs a name matching process related to the export risk based on the name matching conditions and the appropriate characters to identify the name matching information and obtain related information of the name matching information. The determination unit includes the step of determining the export risk based on the determination conditions and the related information, Risk assessment method.

9. A risk assessment program for determining export risks, The computer is configured to function as a storage unit, a detection unit, a name matching processing unit, and a determination unit. The aforementioned storage unit stores the matching conditions and the determination conditions. The detection unit detects interfering elements from the export-related input content based on predetermined conditions, identifies appropriate characters which are the input content from which the interfering elements have been removed, The name matching processing unit performs the name matching process related to the export risk based on the name matching conditions and the appropriate characters to identify the name matching information, and obtains related information of the name matching information. The determination unit determines the export risk based on the determination conditions and the related information. Risk assessment program.

Images