Data encryption processing system for large-scale file processing based on LLM and RAG when transmitting and receiving network data

Abstract

The present invention discloses a technical concept comprising a system for processing data encryption during data transmission and reception between networks for file sourcing and analysis of a corporate DB based on AI-based LLM and RAG, which includes: a network connection unit that verifies the user's access rights through a user terminal, proceeds with a network connection to the corporate DB, and manages the processing of network connections between the corporate DB, network shared storage, and local storage; a cryptography processing unit that performs encryption processing on the network connection connected through the network connection unit according to a preset encryption processing method for data encryption between terminals; and a data transmission and reception unit that enables data transmission between terminals when encryption processing is performed through the cryptography processing unit.

Description

Technology Field

[0001] The present invention relates to a system for encrypting data during data transmission and reception between networks, and more specifically, to a technical field concerning a system for encrypting data during the data processing process when selectively transmitting and receiving files from a corporate database to network shared storage and local storage for large-scale file processing for tax audits by the National Tax Service, investigations by the Ministry of Justice, or other large-scale data / document file investigations, based on AI (Artificial Intelligence)-based LLM (Large Language Model) and RAG (Retrieval-Augmented Generation). Background Technology

[0003] During tax audits or accounting audits, a large volume of corporate transaction data, financial statements, and supporting documents exists, but most are managed in the form of unstructured data. Existing accounting systems possess differing formats and incompatible structures, making data integration difficult; furthermore, the high volume of manual verification items leads to excessive consumption of time and costs. Systems based on OCR or simple search methods have low error detection rates because they fail to accurately grasp accounting context or numerical relationships. Additionally, the lack of capabilities to track the time-series continuity and data consistency of audited materials creates a risk of manipulation, while instability in inter-system integration degrades the overall quality of the audit.

[0004] Recently, innovation in accounting audit and tax data processing based on AI and automation technologies has been actively underway. Systems are being developed that utilize natural language processing to understand the context within accounting documents and perform automatic classification and verification by precisely recognizing tables and numbers using deep learning-based OCR. Audit efficiency is being enhanced through the introduction of RPA automation for repetitive tasks, machine learning models for detecting abnormal transactions, and cloud-based data management platforms. Furthermore, efforts to prevent data tampering and forgery are being made using blockchain-based transaction tracking technology, while the establishment of systems for early risk identification and reporting through real-time audit logic is being pursued in parallel.

[0005] The AI ​​accounting and audit automation market is growing rapidly globally, driven by trends toward digital transformation and enhanced transparency. It is expected to expand to approximately $13 billion by 2025, with an average annual growth rate of over 10%. Key growth drivers include the introduction of electronic tax systems in various countries, strengthened corporate internal controls, and mandatory ESG information disclosure. In particular, not only large accounting firms but also mid-sized companies are expanding the adoption of AI audit solutions. Domestically, the markets for related software, data analysis, and blockchain audit technologies are projected to expand rapidly, driven by the National Tax Service's policy to advance computerized tax audits. Prior art literature

[0007] Korean Patent Publication No. 10-2014-0005878 Korean Registered Patent Publication No. 10-1658964 Korean Registered Patent Publication No. 10-1804139 The problem to be solved

[0008] The data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to the present invention has been devised to solve the conventional problems described above, and presents the following problem to be solved.

[0009] First, we aim to prevent data hacking that may occur during the replication of large amounts of data in the AI-based LLM and RAG processing.

[0010] Second, we aim to prevent the unauthorized external leakage of data between networks through third-party terminals or external connections during mass data replication.

[0011] Third, we intend to allow selective data access only to authorized users.

[0012] The problems solved by the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0014] A data encryption processing system for large-scale file processing based on LLM and RAG according to the present invention during data transmission and reception between networks has the following means for solving the problem to be solved above.

[0015] A data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to the present invention may be characterized by comprising: a network connection unit that verifies the user's access rights through a user terminal, proceeds with a network connection to the corporate DB, and manages the network connection between the corporate DB, network shared storage, and local storage; a password processing unit that performs password processing on the network connection connected through the network connection unit according to a preset password processing method for data encryption between terminals; and a data transmission and reception unit that allows data transmission between terminals to proceed when password processing is performed through the password processing unit.

[0016] In the case of a data encryption processing system for network data transmission and reception for large-scale file processing based on LLM and RAG according to the present invention, the network connection unit may be characterized by including: a target DB connection unit that checks the user's access rights through a user terminal and mutually connects the corporate DB and the network shared storage through the network; and a local network connection unit that mutually connects the network shared storage where the target information was uploaded and the local storage when the corporate DB and the network shared storage are mutually connected through the target DB connection unit and the target information is uploaded to the network shared storage.

[0017] In the case of a data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to the present invention, the encryption processing unit may be characterized by comprising: i) in a network connection between the enterprise DB and the network shared storage, the enterprise DB is set as the sender and the network shared storage is set as the receiver; ii) in a network connection between the network shared storage and the local storage, the network shared storage is set as the sender and the local storage is set as the receiver; 1) the receiver generates a pair of public and private keys; 2) the receiver transmits the public key to the sender, the sender encrypts the data to be transmitted with the receiver's public key, the sender transmits the encrypted ciphertext to the receiver, and the receiver decrypts the received ciphertext with its own private key.

[0018] In the case of a data encryption processing system for network data transmission and reception for large-scale file processing based on LLM and RAG according to the present invention, the encryption processing unit, when proceeding with a network connection to the corporate DB after verifying the user's access rights through the user terminal, may be characterized by including a TLS / SSL encryption processing unit that: i) mediates for the user terminal to transmit a connection request to the corporate DB; ii) mediates for the corporate DB to present the corporate DB certificate to the user terminal; iii) requests for the user terminal to verify the certificate of the corporate DB; iv) requests for the user terminal to generate an arbitrary session key to be used for symmetric key communication; v) requests for the user terminal to encrypt the session key using the public key of the corporate DB; vi) mediates for the user terminal to transmit the encrypted session key to the corporate DB; vii) requests for the corporate DB to decrypt the session key encrypted with the private key of the corporate DB; and viii) requests for the user terminal and the corporate DB to exchange messages, encrypt, and decrypt using the same session key. there is.

[0019] In the case of a data encryption processing system for network data transmission and reception for large-scale file processing based on LLM and RAG according to the present invention, the encryption processing unit may be characterized by including a file encryption processing unit that performs file encryption processing on the target information according to a preset encryption processing method and stores it on the network shared storage. Effects of the invention

[0021] The data encryption processing system for large-scale file processing based on LLM and RAG according to the present invention, configured as described above, provides the following effects during data transmission and reception between networks.

[0022] First, through the encryption of manganese data, data leakage is prevented even from arbitrary external hacking.

[0023] Second, through the optional application of triple encryption, data is protected through encryption even when a third party or external access attempt is made between the networks.

[0024] Third, data preservation ensures the authenticity of evidence during tax audits, forensic investigations, and search and seizure, thereby preserving its admissibility and consequently securing the reliability of the data results.

[0025] The effects of the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below. Brief explanation of the drawing

[0027] Figure 1 is a conceptual diagram of an entire frame to which a data encryption processing system is applied during data transmission and reception between networks for large-scale file processing based on LLM and RAG according to one embodiment of the present invention. FIG. 2 is a conceptual diagram of a network encryption processing section in which a data encryption processing system is applied during network data transmission and reception for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 3 is a conceptual diagram illustrating that encryption processing is applied between a network shared storage and a local storage among data encryption processing systems for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 4 is a block diagram illustrating the main configuration of a data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 5 is a block diagram illustrating a network connection unit and its sub-configuration among a data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 6 is a block diagram illustrating a encryption processing unit and its sub-configuration among a data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 7 is a conceptual diagram illustrating the function of an asymmetric encryption processing unit among a data encryption processing system for transmitting and receiving data between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 8 is a conceptual diagram illustrating an asymmetric encryption processing unit functioning between a corporate DB and a network shared storage in a data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 9 is a block diagram illustrating a file encryption processing unit and its sub-configuration among a data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 10 is a conceptual diagram for explaining the function of an array matrix processing unit among a data encryption processing system for transmitting and receiving data between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 11 is a conceptual diagram illustrating the function of a random selection inversion unit in a data encryption processing system for transmitting and receiving data between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 12 is a conceptual diagram illustrating the function of a random matrix application unit in a data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. Specific details for implementing the invention

[0028] For large-scale file processing based on LLM and RAG according to the present invention, the data encryption processing system for data transmission and reception between networks may be subject to various modifications and may have various embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the technical spirit and scope of the present invention.

[0029] FIG. 1 is a conceptual diagram of an entire frame to which a data encryption processing system is applied during data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 2 is a conceptual diagram of a network encryption processing portion to which a data encryption processing system is applied during data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 3 is a conceptual diagram illustrating that encryption processing is applied between a network shared storage and a local storage within a data encryption processing system during data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 4 is a block diagram illustrating the main configuration of a data encryption processing system during data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 5 is a block diagram illustrating a network connection unit and its sub-configurations within a data encryption processing system during data transmission and reception between networks for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 6 is a block diagram illustrating an encryption processing unit and its sub-configuration in a data encryption processing system for network data transmission and reception for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 7 is a conceptual diagram explaining the function of an asymmetric encryption processing unit in a data encryption processing system for network data transmission and reception for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 8 is a conceptual diagram illustrating that an asymmetric encryption processing unit in a data encryption processing system for network data transmission and reception for large-scale file processing based on LLM and RAG according to an embodiment of the present invention functions between an enterprise DB and a network shared storage.FIG. 9 is a block diagram illustrating a file encryption processing unit and its sub-configurations in a data encryption processing system for network data transmission and reception based on LLM and RAG for large-scale file processing based on LLM and RAG according to an embodiment of the present invention. FIG. 10 is a conceptual diagram explaining the function of an array matrix processing unit in a data encryption processing system for network data transmission and reception based on LLM and RAG according to an embodiment of the present invention. FIG. 11 is a conceptual diagram explaining the function of a random selection inversion unit in a data encryption processing system for network data transmission and reception based on LLM and RAG according to an embodiment of the present invention. FIG. 12 is a conceptual diagram explaining the function of a random matrix application unit in a data encryption processing system for network data transmission and reception based on LLM and RAG according to an embodiment of the present invention.

[0030] In the case of a data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to the present invention, as illustrated in FIG. 1, it relates to a system for encrypting data during the data processing process when selectively transmitting and receiving files from a corporate DB (10) to a network shared storage (101) and a local storage (102) for large-scale file processing from a corporate DB (10) for the purpose of investigating a large amount of data / document files from a corporate DB (10) from the perspective of a specific agency, such as the Board of Audit and Inspection, the National Tax Service, or the Ministry of Justice, for the purpose of conducting an audit, tax investigation, or search and seizure, etc. between networks, such as between the corporate DB (10) and the network shared storage (101) or between the network shared storage (101) and the local storage (102).

[0031] In addition, during the process of accessing and connecting from the user terminal (1) to the corporate DB (10), security processing during the network connection process with the user terminal (1) is handled, along with encryption processing for the transmitted files / data.

[0032] Here, the user terminal (10) is a terminal carried by the user and may be a smartphone, such as Apple’s iPhone or Samsung Electronics’ Galaxy series, a tablet PC, such as Apple’s iPad or Samsung Electronics’ Galaxy Tab, or a laptop PC, that is, a laptop or a desktop PC.

[0033] The user terminal (10) is configured to be a wired or wireless network and to perform various computational functions through the installation of its own applications or other programs via internal storage media such as HDD, SSD, or other memory, and is not limited to a specific brand or form of terminal.

[0034] In the case of a data encryption processing system for large-scale file processing based on LLM and RAG according to the present invention, the process of uploading target information from RAG, that is, corporate DB (10), to network shared storage (101) is carried out, and then the process of uploading target information to be used as a target for investigation / investigation, etc. on network shared storage (101) is carried out, and then a mechanism for uploading / transmitting two-stage data files (target information, target information) as described above is established to conduct investigation / audit / analysis on the part of the AI-based LLM (Large Language Model) regarding the target information through detailed AI and language learning.

[0035] To this end, for large-scale file processing based on LLM and RAG according to the present invention, a data encryption processing system for data transmission and reception between networks may include a network connection unit (100); an encryption processing unit (200) and a data transmission and reception unit (300), as shown in FIG. 4.

[0036] Here, the meaning of 'unit' or the suffix "bu" is not limited to a single unit of a machine, device, or part, but is to conceptually define a functional unit that is computationally executed through an application or program.

[0037] In the case of the network connection unit (100), it is configured to verify network connection / access rights to the corporate DB (10) through the user's browser, application, or other connection program via the user terminal (1), proceed with network connection to the corporate DB (10), and manage the processing of network connections between the corporate DB (10), network shared storage (101), and local storage (102).

[0038] In the case of the corporate DB (10), it is a general term for a DB owned by a company subject to audit / investigation / inquiry, or a DB in which the company stores / backs up its own internal information, transaction information, etc., and the network shared storage (101) is connected to the corporate DB (10) via wired / wireless networking. Subsequently, all information subject to investigation, that is, the target information, is uploaded in bulk or in large volume from the corporate DB (10) to ensure authenticity (see FIG. 2). This will be described later.

[0039] Subsequently, the local storage (102) connects to the network shared storage (101), and the local storage (102) may have multiple units based on one network shared storage (101). The local storage (102) also undergoes a process of individually uploading target information to be investigated / audited / analyzed via LLM from the target information of the network shared storage (101) and conducting investigation / auditing / analysis (see FIG. 3), and this part will also be described later.

[0040] The user terminal (1) is allowed access to the corporate DB (10), as well as to the network shared storage (101) and local storage (102) where the target information is uploaded.

[0041] In the case of the password processing unit (200), the configuration is such that data encryption is performed between the terminal-user terminal (1), corporate DB (10), network shared storage (101), and local storage (102) on the network connection connected through the network connection unit (100) according to a ‘pre-set password processing method’.

[0042] In the case of the data transmission and reception unit (300), when encryption processing is performed through the encryption processing unit (200), the configuration allows data transmission to proceed between the terminal-user terminal (1), corporate DB (10), network shared storage (101), and local storage (102).

[0043] As illustrated in FIG. 5, the network connection unit (100) may include a target DB connection unit (110) and a local network connection unit (120).

[0044] In the case of the target DB connection part (110), as shown in FIG. 5, the user's access rights are verified through the user terminal (1), and after the network connection between the user terminal (1) and the corporate DB (10) is completed, the corporate DB (10) and the network shared storage (101) are mutually network connected.

[0045] The local network connection unit (120) is configured such that the corporate DB (10) and the network shared storage (101) are interconnected through the target DB connection unit (110), and when target information is uploaded to the network shared storage (101), the network shared storage (101) on which the target information is uploaded is interconnected with the local storage (102). Subsequently, among the target information in the network shared storage (101), only the target information is selected and temporarily stored in the local storage (102), and then an audit / investigation / inquiry can be conducted based on LLM as described above.

[0046] As illustrated in FIG. 6, the cryptographic processing unit (200) may include an asymmetric key cryptographic processing unit (210).

[0047] In the case of the asymmetric key encryption processing unit (210), as a first method of a preset encryption processing method, as shown in FIGS. 7 and 8, i) in the network connection between the corporate DB (10) and the network shared storage (101), the corporate DB (10) is set as the sender and the network shared storage (101) is set as the receiver, and ii) in the network connection between the network shared storage (101) and the local storage (102), the network shared storage (101) is set as the sender and the local storage (102) is set as the receiver.

[0048] In the case of the asymmetric key encryption processing unit (210), as illustrated in FIGS. 7 and 8, 1) the set recipient generates a pair of public key and private key, 2) the set recipient transmits the public key to the set sender, the sender encrypts the data to be transmitted with the recipient's public key, the sender transmits the encrypted ciphertext to the recipient, and the recipient decrypts the received ciphertext with their own private key.

[0049] In the case of the encryption processing unit (200), as shown in FIG. 6, it may include a TLS / SSL encryption processing unit (220).

[0050] As a second method of a preset encryption processing method, in the case of the TLS / SSL encryption processing unit (220), when proceeding with a network connection to the corporate DB (10) after verifying the user's access rights through the user terminal (1), i) the user terminal (1) mediates to transmit a connection request to the corporate DB (10), ii) the corporate DB (10) mediates to present the corporate DB certificate to the user terminal (1), iii) the user terminal (1) requests verification of the certificate of the corporate DB (10), iv) the user terminal (1) requests the generation of an arbitrary session key to be used for symmetric key communication, v) the user terminal (1) requests encryption of the session key using the public key of the corporate DB (10), vi) the user terminal (1) mediates to transmit the encrypted session key to the corporate DB (10), vii) the corporate DB (10) requests decryption of the session key encrypted with the private key of the corporate DB (10), and viii) the user terminal (1) and the corporate DB (10) is a configuration that requests to send and receive messages using the same session key, and to encrypt and decrypt them.

[0051] In the case of the password processing unit (200), as shown in FIG. 6, it may include a file password processing unit (230).

[0052] In the case of the file password processing unit (230), the configuration is such that the target information is encrypted using the third method of the pre-set password processing method and stored on the network shared storage (101).

[0053] As illustrated in FIG. 9, the file password processing unit (230) may include an array matrix processing unit (231); a random selection inversion unit (232) and a random matrix application unit (233).

[0054] The array matrix processing unit (231) is configured to process the target information into an array of binary numbers of 1s and 0s, i.e., digital information, and then convert it into an M*N matrix. Although a 5*6 matrix is ​​shown as an example in FIG. 10, the size of the M*N matrix can be set to any size depending on the size of the target information.

[0055] In the case of the random selection inversion unit (232), the row and column are randomly selected x times, and the corresponding numbers are inverted so that 1 is 0 and 0 is 1, and the original information that was selected x times and inverted is stored separately.

[0056] Subsequently, as illustrated in FIG. 12, in the case of the random matrix application unit (233), an N*M random matrix is ​​applied to the matrix that has been randomly selected and inverted by the random selection inversion unit (232) to obtain the final matrix. Since the information of the N*M random matrix is ​​also shared only between the transmitting end and the receiving end, if the N*M matrix is ​​not known, the original information cannot be known from the final matrix.

[0057] The scope of rights of the present invention is determined by the matters described in the patent claims, and the parentheses used in the patent claims are not intended for optional limitation but for clear components, and the descriptions within the parentheses should also be interpreted as essential components. Explanation of the symbols

[0059] 1: User terminal 10: Corporate DB 100: Network Connection Unit 101: Network Shared Storage 102: Local Storage 110: Target DB Connection 120: Local network connection 200: Password processing unit 210: Asymmetric key cryptography unit 220: TLS / SSL cryptography unit 230: File password processing unit 231: Array matrix processing unit 232: Random selection inversion section 233: Random matrix application section 300: Data Transmission / Reception Unit

Claims

Claim 1 A system for processing data encryption during mutual data transmission and reception between networks for file sourcing and analysis of an enterprise DB based on AI (Artificial Intelligence)-based LLM (Large Language Model) and RAG, comprising: a network connection unit that verifies a user's access rights through a user terminal, proceeds with a network connection to the enterprise DB, and manages the processing of network connections between the enterprise DB, network shared storage, and local storage; a password processing unit that performs password processing for data encryption between terminals on the network connection established through the network connection unit according to a preset password processing method; and a data transmission and reception unit that enables data transmission between the terminals when password processing is performed through the password processing unit, wherein the network connection unit includes a target DB connection unit that verifies a user's access rights through a user terminal and mutually connects the enterprise DB and the network shared storage. The system includes a local network connection unit that interconnects the network shared storage with the network shared storage where the target information is uploaded, and when the target information is uploaded to the network shared storage through the target DB connection unit, the network shared storage where the target information is uploaded is interconnected with the local storage. The encryption processing unit includes a file encryption processing unit that performs file encryption processing on the target information according to a preset encryption processing method and stores it on the network shared storage. The file encryption processing unit includes an array matrix processing unit that processes the target information into digital information, which is an array of binary numbers of 1 and 0, and then converts it into an M×N matrix; and a random selection inversion unit that performs random selection of rows and columns of the M×N matrix X times and inverts the selected numbers.A data encryption processing system for network data transmission and reception based on LLM and RAG for large-scale file processing, characterized by including a random matrix application unit that extracts encrypted information by multiplying an N×M random matrix by a random selection inversion unit that inverts the matrix. Claim 2 delete Claim 3 A data encryption processing system for data transmission and reception between networks for large-scale file processing based on LLM and RAG according to claim 1, wherein the encryption processing unit comprises: i) setting the corporate DB as the sender and the network shared storage as the receiver in a network connection between the corporate DB and the network shared storage; ii) setting the network shared storage as the sender and the local storage as the receiver in a network connection between the network shared storage and the local storage; 1) the receiver generates a pair of public and private keys; 2) the receiver transmits the public key to the sender, the sender encrypts the data to be transmitted using the receiver's public key, the sender transmits the encrypted ciphertext to the receiver, and the receiver decrypts the received ciphertext using its own private key. Claim 4 A data encryption processing system for inter-network data transmission and reception based on LLM and RAG for large-scale file processing, wherein, in the case of the encryption processing unit proceeding with a network connection to the corporate DB after verifying the access rights of the user through the user terminal, i) mediates for the user terminal to transmit a connection request to the corporate DB, ii) mediates for the corporate DB to present a corporate DB certificate to the user terminal, iii) requests for the user terminal to verify the certificate of the corporate DB, iv) requests for the user terminal to generate an arbitrary session key to be used for symmetric key communication, v) requests for the user terminal to encrypt the session key using the public key of the corporate DB, vi) mediates for the user terminal to transmit the encrypted session key to the corporate DB, vii) requests for the corporate DB to decrypt the session key encrypted with the private key of the corporate DB, and viii) requests for the user terminal and the corporate DB to exchange messages, encrypt, and decrypt using the same session key. Claim 5 delete

Images