Transaction process management method, transaction process management system, transaction process management program, and transaction process verification method

Abstract

In multi-party transactions, the lack of a neutral record-keeping infrastructure makes it impossible to prevent alteration of transaction facts. To address this challenge, a transaction process management method is provided in which a processing unit (S1), connected to terminal devices (T1, T2) via a network (N) and belonging to no administrator, executes all processing within a protected execution environment. The processing unit converts received data into integer values, irreversibly fixes the record with a cryptographic summary value, generates a synchronous proof value based on both inputs, makes a determination and automatically fixes the record using only integer arithmetic comparison processing, cryptographically concatenates each result as an append-only record, releases the record as read-only when technical conditions are met, generates and provides an attribution proof value, and records the input and output of each operation to enable independent re-verification.

Description

[Technical Field]

[0001] The present invention relates to a transaction process management method, a transaction process management system, a transaction process management program, and a transaction process verification method. More particularly, the present invention relates to a technique for objectively recording facts occurring at each stage of transactions between a plurality of parties, preventing falsification, structurally eliminating disputes, and enabling verification of the legitimacy of processing. [Background Art]

[0002] In transactions between a consignor and a trustee in the manufacturing industry, quality-related disputes may occur at each stage from receiving specifications to completing delivery. Conventionally, these disputes have been mainly addressed by the following methods.

[0003] First, there is a method based on paper-based inspection records. An inspector enters measurement values into a paper form, and both the consignor and the trustee confirm the values. This method allows easy falsification of records, and does not provide a means for resolving disputes over claims and counterclaims.

[0004] Second, there is a method using a database system. Measurement data is stored in a database and managed electronically. However, it is impossible to eliminate the possibility that the database administrator (usually either the consignor or the trustee) may modify the data.

[0005] Third, methods using blockchain technology have been proposed. In blockchain, a plurality of nodes on a distributed network achieve consensus and prevent falsification of transaction records. However, blockchain has the following problems. - High computational cost for consensus formation (such as Proof of Work / Proof of Stake) - Since each node belongs to a party (miner / validator), the independence of node operators cannot be guaranteed - Since smart contracts execute arbitrary programs, the transparency of decision logic cannot be ensured - When using floating-point arithmetic, rounding errors between the CPU / GPU / OS may prevent the determination result from being reproduced.

[0006] Fourth, immutable ledgers (e.g., Amazon QLDB) have been proposed. While these maintain a history of data changes, they cannot eliminate the possibility that the ledger administrator (service provider) could manipulate the data. Furthermore, they lack synchronization proof based on input from both parties, and automatic detection limited to integer operations. It should be noted that Amazon QLDB discontinued its service in July 2025, creating a gap in the market for centralized append-only ledgers.

[0007] Fifth, a deterministic multi-party state transition language (e.g., Digital Asset's DAML) has been proposed. DAML is a general-purpose programming language that does not depend on a blockchain platform and enables deterministic state transitions to be executed among multiple parties. However, DAML differs from the present invention in the following respects. - DAML is a general-purpose programming language and does not have the constraint of excluding floating-point operations. - DAML provides a program execution environment and does not have the concept of an independent computing entity (a processing unit that does not belong to the administrator of any terminal device) separate from all parties to a transaction. - DAML does not have a mechanism to structurally enforce comparison processing using only integer arithmetic and automatic state transitions. - DAML does not have a protection mechanism or proof-of-concept generation to technically guarantee the independence of the processing unit.

[0008] Sixth, transaction deterministic data processing architectures have been proposed in the field of financial transactions (e.g., U.S. Patent No. 10,637,967, U.S. Patent No. 10,326,862). These architectures guarantee the determinism of transaction processing in high-frequency financial exchanges by having an arbitrator select the processing results of multiple redundant processors. However, these arbitrators select results from redundant processors and do not perform verification decisions as an independent third party separate from the parties to the transaction. Furthermore, they lack features such as limiting comparison processing to integer arithmetic only, generating synchronous proof values ​​based on inputs from both sides, a mechanism for read-only access in emergencies, and generating constitutive proof values ​​that allow external verification of the calculation process.

[0009] Seventh, in the financial industry, there is a known practice of storing currency data as integers rather than floating-point numbers (e.g., "cents" rather than "dollars"). However, this practice concerns the format of data storage and does not provide a transaction process management method that includes comparison processing using only integer arithmetic, automatic state transitions based on the results, execution of processing by independent arithmetic entities, and independent reverifiability of calculations.

[0010] Furthermore, the issue of reproducibility in floating-point arithmetic is not merely a theoretical problem. In 2025, an attack exploiting rounding errors in IEEE 754 floating-point arithmetic occurred in the decentralized financial protocol Balancer, resulting in the loss of approximately $128 million in assets. This incident demonstrated that transaction decisions relying on floating-point arithmetic can cause real economic damage.

[0011] Eighth, there is the practice of record keeping by an independent and trusted third party (TTP). Notaries, escrow services, and certification authorities (CAs) manage or authenticate records as independent third parties from the parties to the transaction. However, these traditional TTPs have the following problems: (i) the independence of the third party is guaranteed only by contract or law, not by technology (the possibility of the third-party manager manipulating the data cannot be ruled out); (ii) they do not have a verification decision algorithm that relies solely on integer arithmetic; (iii) they do not have an automatic state transition based on the decision result; and (iv) terminal devices do not have a means to independently re-verify the legitimacy of the processing.

[0012] Ninthly, the timestamp protocol specified in RFC 3161 is known. The Time-Stamp Authority (TSA), as an independent third party, assigns timestamps to cryptographic summaries of data and manages timestamp tokens in a chain structure. However, the TSA provides proof of time and does not perform verification of the data content. In other words, the TSA does not have comparison processing using only integer arithmetic, generation of a synchronous proof value based on both inputs, automatic state transitions based on the judgment result, or independent reverifiability of the process.

[0013] Tenth, version control systems based on hash chains (e.g., Git) are known. Git uses cryptographic summary values ​​of SHA-1 or SHA-256 to manage each commit in an append-only chain structure, making tampering detectable. However, Git is intended for version control and does not have the concept of a processing unit independent of the transaction parties, verification decisions based solely on integer arithmetic, synchronization proof based on inputs from both parties, or automatic state transitions based on the decision results.

[0014] Eleventh, in the field of medical information systems, the Break-the-Glass (BTG) access pattern is known. BTG is a method for accessing medical records in an emergency by bypassing normal authentication, and the fact of access is audited afterward. However, conventional BTG has the following problems: (i) the determination of emergency conditions depends on human subjectivity (access is possible simply because it is judged to be an emergency), (ii) emergency conditions are not technically predefined, and (iii) the legitimacy of access depends solely on a post-access human audit.

[0015] Twelfth, large-scale internet platform operators (marketplace operators, cloud service providers, social media operators, etc.) centrally manage transaction data and records between users on their own servers. On these platforms, transaction judgment and management (calculation of product evaluation scores, evaluation of seller creditworthiness, price calculation, service level determination, etc.) are performed based on algorithms unilaterally set by the platform operator. Users have no technical means to verify whether the algorithm is processing their data fairly. Furthermore, because platform operators have full access rights and the right to modify the database, the possibility of intentionally or negligently altering transaction records cannot be structurally eliminated. This centralized data management structure has the problem of a "single point of confidence," where all parties to a transaction must unconditionally trust the platform operator, and the independence of the platform operator itself is not technically guaranteed.

[0016] Thirteenth, internet-based criminal fraud is becoming a serious social issue. Identity theft, special fraud (so-called "ore-ore fraud"), where perpetrators impersonate others to conduct transactions; fraud where perpetrators impersonate relatives or acquaintances to request money via telephone; fraud where perpetrators impersonate police officers, tax officials, or social security officials to defraud people of personal information or money; and romance fraud, where perpetrators use false identities and romantic feelings to defraud people of money, all exploit common technological vulnerabilities. Specifically, (i) there is no cryptographic means to prove that the identity claimed by the other party in a transaction is genuine; (ii) all inputs from both parties during the transaction process are not recorded in an immutable format; and (iii) there is no means for the parties to independently re-verify the transaction's decision-making process. Traditional identity verification methods (passwords, SMS authentication codes, image transmission of identification documents, etc.) can be stolen, forged, or bypassed through social engineering. If these credentials are obtained by an attacker, there is no technical means to retrospectively prove that both legitimate parties actually participated in the transaction.

[0017] Fourteenth, with the rapid development of artificial intelligence technology, the accuracy of deepfakes (fake images, videos, and audio generated by AI) has dramatically improved, reaching a level where it is difficult for humans to distinguish between genuine and fake content by sight and hearing. AI-generated content is being misused for fraud, fake news, defamation, and political information manipulation. Furthermore, biometric authentication (fingerprint authentication, facial recognition, iris recognition, voiceprint authentication, etc.) is based on pattern matching of input data and registered data, and therefore has a structural vulnerability that allows authentication to be bypassed if AI generates sufficiently sophisticated fake biometric information. In other words, all authentication methods based on pattern matching (biometric authentication, image authentication, voice authentication, handwriting authentication, etc.) will irreversibly lose reliability as AI technology evolves. Conventional technology does not have a means to guarantee the genuine participation of transaction parties and to verify the authenticity of content based solely on the mathematical irreversibility of cryptographic computation, without relying on pattern matching authentication.

[0018] Fifteenth, the Coalition for Content Provenance and Authenticity (C2PA) has been proposed as a standard for proving the origin and authenticity of content (C2PA Technical Specification 2.2, 2025). C2PA is an industry group in which major technology companies such as Adobe, Microsoft, Intel, Arm, Google, and Meta participate, and it enables the tracking of content origin by embedding hash values ​​and digital signatures in the content's metadata (generation date and time, generation device, editing history, etc.). C2PA Specification 2.2 supports attribution proof by TEE as an option. However, C2PA differs structurally from the present invention in the following respects. - C2PA is a metadata standard for content, not a method for managing the transaction process. - C2PA does not have the concept of a synchronous proof value based on input from both parties (only the signature of one of the content creators). - C2PA does not have the concept of comparison processing using only integer arithmetic or automatic state transitions. - C2PA's Claim Generator is the content creator's own device and does not have the concept of a processing unit independent of all parties to the transaction. - C2PA only provides signature verification and does not have the concept of independent reverification based on integer input and output values. - C2PA does not have the concept of read-only open access based on technical conditions.

[0019] None of the above conventional technologies can simultaneously satisfy the following seven requirements. (1) The processing is performed by an operating entity independent of all parties to the transaction, and such independence is guaranteed by a technical protection mechanism and verifiable by an attribution value. (2) Execution of a verification and judgment algorithm using only integer arithmetic, eliminating floating-point rounding errors. (3) Automatic state transitions based on the judgment result (fixing of records without human intervention) (4) Provision of read-only access based on predetermined technical conditions (5) Each terminal device can independently re-verify processing based on input values and output values of integer arithmetic operations (6) Generation of a cryptographic proof value based on inputs from both parties to a transaction ensures a synchronization proof that cannot be generated by only one party, and technically guarantees the genuine participation of the counterparty to the transaction (7) The genuine participation of transaction parties and the authenticity of content are guaranteed based only on the mathematical irreversibility of cryptographic computation, without relying on authentication methods based on pattern matching (including biometric authentication)

Prior Art Documents

Patent Documents

[0020] (Summary of Comparison with Prior Art) The present invention differs from any of Cited Document Group A (decentralized consensus blockchain), Cited Document Group B (centralized append-only ledgers represented by Amazon QLDB), and Cited Document Group C (standalone hardware trusted execution environment technologies such as Intel SGX, ARM TrustZone, and AMD SEV) in the following points.

[0021] First, unlike the distributed nodes in Cited Document Group A, which are configured by each party or their stakeholders (miners, validators), the processing device (S1) of the present invention is configured as a "computing entity independent of all parties to the transaction", and its independence is technically guaranteed by a protected execution environment (TEE), and can be independently verified by a terminal device through an attestation value. This structurally eliminates the problems of unavoidable consensus formation costs (Proof of Work / Proof of Stake) and the party dependency of node operators in Cited Document Group A.

[0022] Secondly, the "setting based on mutual agreement of a predetermined minimum unit" in step (a) of the present invention and the "comparison processing using only integer arithmetic" in step (c) of the present invention are not disclosed in any of the cited document group A (a smart contract environment that does not restrict floating-point arithmetic) and cited document group B (an append-only ledger that only performs data storage and does not have determination processing). Processing using only integer arithmetic eliminates the floating-point rounding error specified in IEEE 754 in principle, and has technical significance in preventing actual economic damage such as the rounding error attack on the Balancer protocol that occurred in 2025 (Non-Patent Document 5).

[0023] Thirdly, the synchronization proof value H(H(d1)∥H(d2)) in step (b) of the present invention is a cryptographic proof value that cannot be generated without inputs from both parties to a transaction, and is not disclosed in any of the cited document group B (a record generated by only one party's input), C2PA described in Non-Patent Document 6 (which only uses a signature from one content creator), and RFC 3161 timestamp described in Non-Patent Document 2 (which only provides time proof for single data). This mathematically eliminates "he said / she said" type disputes and spoofed transactions by one party.

[0024] Fourthly, the "append-only chain including the cryptographic digest value of the immediately preceding record" in step (d) of the present invention is formally similar to version control systems such as Git (see paragraph 0013 of the specification), but differs in that it is structurally combined with steps (a) to (c) and steps (f) and (g). That is, the chain structure of the present invention is inseparably combined with (i) integration based on the mutually agreed minimum unit, (ii) the synchronization proof value based on inputs from both parties, (iii) automatic state transition by integer arithmetic determination, (iv) execution within a protection mechanism and generation of configuration proof values, and (v) independent re-verification by recording integer arithmetic input values and output values, so that a citation of Git lacking these elements cannot derive all elements of the present invention.

[0025] Fifth, the "read-only access based on predetermined technical conditions recorded in a record specifically for appending" in step (e) of the present invention is partially similar in purpose to the Break-the-Glass (BTG) access pattern in the medical information system field (paragraph 0014 of the specification). However, while conventional BTG relies on a human subjective judgment (a declaration that "it has been judged to be an emergency"), the present invention differs in that it relies on technical verification of technical conditions (absence of authenticated access within a predetermined period or success of threshold authentication). This structurally eliminates the intervention of subjective judgment.

[0026] Sixth, the "independent re-verification of the terminal device by recording the input and output values ​​of each integer operation into an append-only record" in step (g) of the present invention fundamentally differs from the selection of results between redundant processors in U.S. Patents 10,637,967 and 10,326,862 (Patent Documents 1 and 2, Redundant Processor Arbitration Architecture in Financial Transactions). The arbitrators in Patent Documents 1 and 2 perform result selection between redundant processors and do not perform verification by a third party independent of the trading parties. The present invention enables the terminal device to independently re-execute integer operations outside the processing unit (S1), and avoids the circular reasoning of self-verification by the processing unit itself.

[0027] Seventh, unlike the DAML described in Non-Patent Document 3, the present invention has (i) the structural elimination of floating-point arithmetic, (ii) the concept of an arithmetic entity independent of the transaction parties, (iii) independence guaranteed by a hardware protection mechanism, and (iv) external verifiability through attribution values. DAML provides deterministic state transitions as a general-purpose programming language, but it does not structurally possess the above elements.

[0028] Eighth, the present invention differs from the C2PA described in Non-Patent Document 6 in terms of its structural purpose. C2PA is a "metadata standard for content origins" and deals with only one of the content creators' signatures. On the other hand, the present invention targets "transaction process management" and structurally enforces synchronous certification based on inputs from both parties in a transaction, determination by integer arithmetic, and automatic state transitions. While C2PA Specification 2.2 deals with TEE attribution as an "option," the present invention differs structurally in that it makes the TEE attribution value a mandatory requirement of step (f) and specifies that all processing in steps (a) through (e) and step (g) is performed within the TEE.

[0029] As described above, by referring to any of the above cited documents and prior art, either individually or in combination, it is not possible to derive a configuration that simultaneously satisfies all the elements of steps (a) through (g) of claim 1 of the present invention, and such combinations are not easily conceivable to a person skilled in the art.

[0030] [Patent Document 1] U.S. Patent No. 10,637,967 [Patent Document 2] U.S. Patent No. 10,326,862 [Non-patent literature]

[0031] [Non-patent document 1] IEEE 754-2019, IEEE Standard for Floating-Point Arithmetic [Non-patent Document 2] IETF RFC 3161, Internet X.509 Public Key Infrastructure Time-Stamp Protocol (TSP) [Non-patent Document 3] Digital Asset Holdings, LLC, "The DAML Smart Contract Language", 2024 [Non-Patent Literature 4] Amazon Web Services, Inc., "Amazon QLDB Developer Guide", 2024 (Service discontinued in July 2025) [Non-Patent Document 5] Balancer Labs, "Post-Mortem: Rounding Vulnerability in Balancer Protocol V2", 2025 [Non-Patent Literature 6] Coalition for Content Provenance and Authenticity (C2PA), "C2PA Technical Specification", Version 2.2, 2025 [Overview of the Initiative] [Problems that the invention aims to solve]

[0032] This invention has been made in view of the problems of the prior art described above, and aims to solve the following problems.

[0033] The first challenge is ensuring the objectivity of transaction records. In conventional technologies, the objectivity of the records is not structurally guaranteed because the record keeper belongs to one of the parties to the transaction. Furthermore, even in conventional technologies that use an independent third party (TTP, TSA, etc.), the independence of that third party is guaranteed only by contract or law, and the possibility that the third party keeper may technically manipulate the data cannot be ruled out.

[0034] The second challenge is ensuring the reproducibility of the judgment results. Conventional judgment methods that use floating-point arithmetic can result in rounding errors between different processors, potentially leading to different judgment results for the same data.

[0035] The third challenge is the structural elimination of quality disputes. In conventional technology, the fixing of records based on judgment results depends on human operation, which creates opportunities for falsification of records after the judgment.

[0036] The fourth challenge is the verifiability of the processing. In conventional processing by an independent third party, there is no means for the trading parties to independently verify that the third party has correctly performed the processing. Self-verification by the processing device itself is circular reasoning and is not a technically effective verification method.

[0037] The fifth challenge is the lack of independence of platform operators and the prevention of impersonation of transaction parties. On large-scale internet platforms, transaction judgment processing is performed on the platform operator's own servers, so the objectivity of the judgment cannot be structurally guaranteed. Furthermore, in transactions conducted over the internet, there is no cryptographic means to prove that both parties to a transaction actually participated in the transaction, making it technically impossible to eliminate impersonation fraud. [Means for solving the problem]

[0038] To solve the above problems, the present invention provides a transaction process management method that is executed in a system comprising at least one first terminal device, at least one second terminal device, and a processing device having a processor and a storage device that does not belong to the administrator of any of the terminal devices, wherein each of the terminal devices and the processing device are connected via a network.

[0039] The transaction process management method of the present invention includes the following steps.

[0040] In step (a), the processing device sets a predetermined minimum unit based on the agreed data received from both the at least one first terminal device and the at least one second terminal device, records the predetermined minimum unit in the storage device, divides each data received from the at least one first terminal device by the predetermined minimum unit to convert it into an integer value, generates a cryptographic summary value for each integer value, and irreversibly fixes the record in the storage device. Since the setting of the predetermined minimum unit is based on the agreement of both parties, it is structurally prevented for one party to arbitrarily manipulate the precision. As a result, the input data is integerized with the precision of the minimum unit, and rounding errors caused by floating-point arithmetic are eliminated in principle.

[0041] In step (b), the processing unit calculates a cryptographic summary value of the concatenation of the cryptographic summary value of each acknowledgment data received from the at least one second terminal device with the corresponding cryptographic summary value from step (a), thereby generating synchronous proof values ​​that cannot be generated without input from both corresponding terminal devices. This mathematically eliminates "he said, she said" disputes.

[0042] In step (c), the processing device converts the deliverable data received from the at least one second terminal device into an integer value, performs a comparison process using only integer arithmetic, outputs a determination result of whether it matches or not, automatically fixes the delivery record for the transaction section if it matches, and automatically fixes the negotiation record for the transaction section if it does not match. As a result, the process from determination to record fixing is performed automatically without human intervention.

[0043] In step (d), the processing device sequentially concatenates the results of steps (a) through (c) as append-only records that include the cryptographic summary value of the immediately preceding record, and records them in the storage device. This makes it possible to detect any tampering with the record at any point in time as a break in the chain.

[0044] In step (e), the processing device, when a predetermined technical condition previously recorded in the append-only record is met, releases at least a portion of the record in read-only mode and adds data identifying the fact of the release operation and the satisfied technical condition to the concatenation of the append-only record.

[0045] Here, the predetermined technical conditions are, for example, (i) the continuation of a state in which no authenticated access is performed from any terminal device included in the system within a predetermined period, or (ii) the success of threshold authentication using a predetermined number or more of emergency keys from a plurality of pre-registered emergency keys. Since both conditions are irreversibly recorded in the append-only record at the start of the transaction, they cannot be changed afterward. Furthermore, the satisfaction of the conditions is technically verifiable and does not depend on subjective human judgment. This structurally eliminates the problem of "who determines it to be an emergency" in conventional Break-the-Glass (BTG) access.

[0046] In step (f), the processing unit executes all the processes of steps (a) through (e) as an independent processing entity from any administrator of any terminal device constituting the system, within an execution environment having a protection mechanism that prevents access to the processing unit's calculation process and stored contents by any administrator of any terminal device constituting the system. Furthermore, the processing unit generates an attribution value indicating the integrity of the program code executed in the execution environment and provides it to at least one of the terminal devices. Note that step (f) defines the execution environment for steps (a) through (e) and step (g), and is not executed after step (e) in the chronological order of processing. The execution environment with the protection mechanism defined in step (f) is continuously established from the start of system operation, and all the processes of steps (a) through (e) and step (g) are executed under its protection.

[0047] The aforementioned protection mechanism is implemented, for example, as a hardware-isolated Trusted Execution Environment (TEE, e.g., Intel SGX, ARM TrustZone, AMD SEV). The TEE (i) makes the program code and data within the TEE unreadable and unmodifiable by any software outside the TEE (including operating systems and hypervisors), and (ii) generates an attestation containing a cryptographic summary of the program code executed within the TEE, allowing a remote verifier to cryptographically verify that the intended program is being executed within the TEE. This ensures the independence of the processing unit (S1) not only by contractual agreement but also technically at the hardware level. Even the operator of the data center where S1 is installed cannot read or modify the program code and data within the TEE.

[0048] By providing an attribution value, each terminal device can remotely verify that the processing unit (S1) is executing the intended program code. The terminal device verifies the received attribution value with the TEE manufacturer's public key and confirms that the cryptographic summary value of the program code contained in the attribution value matches a pre-agreed value.

[0049] In step (g), the processing device records the input and output values ​​of each integer operation in steps (a) to (c) in the append-only record. This allows each of the terminal devices to independently re-execute the integer operation based on the recorded input values ​​and verify their agreement with the output values. The verification of the integrity of the chain structure in step (d) can be performed independently by the terminal devices by sequentially recalculating the cryptographic summary values ​​of all records from the beginning and confirming their agreement with the recorded cryptographic summary values ​​(see claim 5). Furthermore, the satisfaction of the technical conditions in step (e) can also be independently confirmed by the terminal devices by comparing the condition definitions and records of satisfaction, which are irreversibly recorded in the append-only record.

[0050] The effect of step (g) is based on the determinism of integer arithmetic. Integer arithmetic produces the same output for the same input in any processor. Therefore, if any terminal device re-executes integer arithmetic based on the input value and arithmetic procedure recorded in the append-only record, it can confirm that it matches the recorded output value. If a discrepancy is detected, it is proven that the processing unit (S1) did not perform the processing correctly. This verification is performed independently outside of the processing unit (S1), so S1 does not fall into circular reasoning by verifying itself.

[0051] Furthermore, the present invention provides a transaction process management system and a transaction process management program that implement the above transaction process management method. In addition, the present invention also provides a transaction process management method and system that does not include step (e) (read-only access based on technical conditions). In embodiments that do not include step (e), steps (a) to (d), step (f), and step (g) realize objectivity of transaction records, reproducibility of judgment results, structural exclusion of disputes, and verifiability of processing. This embodiment is applicable in transaction environments where emergency read-only access is not required (for example, when separate access means are secured between the trading parties). The present invention also provides a method for verifying a transaction process executed by a terminal device. This verification method involves obtaining an append-only record from the processing device, re-executing integer arithmetic based on the recorded input value to verify consistency with the output value, recalculating the cryptographic summary value of the chain to verify integrity, and verifying the proof value to confirm that the processing device executed the intended program within an execution environment with a protection mechanism. Furthermore, the present invention differs structurally from all of the above-mentioned Patent Documents 1 to 2 and Non-Patent Documents 1 to 6 in that it does not possess any technical structure unique to the present invention (such as a physical enforcement mechanism, a dedicated record for appending, or independent verification using constructive proof values). None of the above-mentioned Patent Documents 1 to 2 or Non-Patent Documents 1 to 6 simultaneously possess all the necessary and sufficient components of the present invention, and the present invention is not easily conceived from these prior art documents. [Effects of the Invention]

[0052] According to the present invention, the following effects can be obtained.

[0053] Firstly, the objectivity of transaction records is structurally and technically guaranteed. Because the processing unit performs all processing within an execution environment with a protection mechanism and provides the terminal device with proof-of-concept values, the independence of the processing unit is guaranteed not only contractually but also technically, and can be verified by the terminal device.

[0054] Secondly, the reproducibility of the judgment results is guaranteed. By employing comparison processing using only integer arithmetic and eliminating floating-point arithmetic, the same judgment results can be obtained on any processor.

[0055] Thirdly, quality disputes are structurally eliminated. Since the process from determining compliance / non-compliance to fixing delivery records or consultation records is performed automatically, there is no opportunity for humans to manipulate records after the determination.

[0056] Fourth, it becomes possible to verify transaction details after the fact. The chain structure of records dedicated to appending information cryptographically guarantees the chronological integrity of all records.

[0057] Fifth, it enables business continuity in emergencies. By opening the system to read-only access based on predetermined technical conditions, records can be viewed under objective conditions that do not rely on subjective judgment, and such operations are recorded.

[0058] Sixth, the verifiability of the process is guaranteed. Since the input and output values ​​of each integer operation are recorded in a dedicated append record, each terminal device can independently re-execute the integer operation to verify the correctness of the process. Because this verification is performed outside the processing unit, it avoids the circular reasoning of self-verification.

[0059] Seventh, objectivity in setting the minimum unit is guaranteed. Since the predetermined minimum unit is set and recorded based on agreed data from both terminal devices, the possibility of one party arbitrarily manipulating the accuracy is structurally eliminated. Eighth, integrated management of multi-party, multi-dimensional transactions becomes possible. By independently performing judgments on multiple evaluation dimensions and using automatic control based on inter-dimensional dependencies and multi-dimensional integrated judgment, complex transaction structures can be structurally managed. When combined with multi-party transactions, the multi-dimensional integrity of the entire global supply chain can be verified in a single step. [Brief explanation of the drawing]

[0060] [Figure 1] This figure shows the overall configuration and network independence (terrestrial lines, 5G, LEO satellite communication, submarine cables) of a transaction process management system according to an embodiment of the present invention. [Figure 2] This figure shows the overall configuration of the processing pipeline consisting of six steps, K1 through K6. [Figure 3] This figure shows a comparison of the trust models of the conventional type (party server) and the present invention (independent processing unit with a protected execution environment). [Figure 4] This figure shows the processing flow of setting the minimum unit and data validation freeze (K1: DVF) based on mutual agreement in step (a). The process of converting raw data V_raw to an integer value V_int by dividing it by a predetermined minimum unit μ is shown in abstract terms. [Figure 5] This figure shows the computation flow of the synchronous proof value generation (K2: SPG) in step (b). [Figure 6] This figure shows the processing flow of integer arithmetic decision and automatic state transition (K3: AST) in step (c). It shows the universal representation using abstract variables V_actual, V_standard, difference Δ, and threshold T. [Figure 7] This figure shows the data structure and tamper detection mechanism of the append-only record chain (K4: TDC) in step (d). [Figure 8] This figure shows the processing flow and automatic recording of operation records for read access (K5: ERA) based on technical conditions in step (e). [Figure 9] This figure shows the independence of the arbitrator (K6: ABI) and the generation and verification of the constructive proof value within the execution environment with a protection mechanism in step (f). [Figure 10] This figure shows the end-to-end sequence of all steps in the transaction process management method as an industry-independent abstraction. [Figure 11] This is a comparative diagram showing the structural differences between blockchain architecture and the architecture of the present invention. [Figure 12] This is a landscape diagram showing the position of the present invention relative to the prior art, along two axes: the technical assurance of the independence of the processing entity and the presence or absence of automatic state transitions. [Figure 13] This figure shows the difference in reproducibility between comparison processing using floating-point arithmetic (IEEE 754) and comparison processing using only integer arithmetic, represented by mathematical abstractions (V, μ, n, ε). [Figure 14] This figure shows the mapping of the seven defense statements of the present invention to Section 101 of the U.S. Patent Act (Alice two-stage test). [Figure 15] This figure shows the chain configuration of multiple transaction segments and the generation of integrated verification values ​​in a multi-party global supply chain. [Figure 16] This diagram shows a network and device-independent architecture, demonstrating that PMG can operate from any terminal device such as a PC, smartphone, tablet, or IoT sensor device over various transports including terrestrial internet, 5G / millimeter wave, LEO satellite communications (Starlink / OneWeb / Kuiper), submarine cables, and mobile cellular communications (3G / 4G / LTE). [Figure 17] This figure shows the audit trail created by recording cryptographic summary values ​​and version information of the inference model in an append-only record. [Figure 18] This diagram shows the transaction architecture with mega AI companies, illustrating the application of integer arithmetic restriction criteria in four types of transactions: GPU computing resource procurement, AI model licensing, training data procurement, and API SLA compliance. [Figure 19] This figure shows the synchronization configuration of append-only records using multiple geographically distributed processing nodes, along with verification of the consistency of the attribution values ​​of each node. [Figure 20] This figure shows the application configuration of the present invention in pandemic response (COVID-19, etc.), government procurement, and crisis response by international organizations (WHO / UNICEF / World Bank, etc.). [Figure 21] This figure shows the fault-tolerant architecture of the present invention against force majeure events (such as war, riots, natural disasters, and economic sanctions). [Modes for carrying out the invention]

[0061] (See Figure 1) Hereinafter, embodiments of the present invention will be described with reference to Figures 1 to 21. Figure 1 shows the overall configuration, and Figures 2 to 21 show the detailed structure and processing flow of each component. Embodiments of the present invention will be described in detail below with reference to the drawings. Note that the following embodiments are not limiting to the present invention, and various modifications are possible within the scope of the technical concept of the present invention.

[0062] (See Figure 2) The main terms used in this specification are defined as follows: "Integer arithmetic" means operations where both input and output are integer values, including addition, subtraction, multiplication, integer division, modulo operations, absolute value operations, comparison operations, and bitwise operations, and does not include floating-point arithmetic as defined in IEEE 754. "Cryptographic summary value" means the output of a cryptographic hash function that generates a fixed-length output value from arbitrary-length input data, including but not limited to SHA-256. "Execution environment with protection mechanisms" means an execution environment that has mechanisms to prevent program code and data within the execution environment from being read and modified by any software outside the execution environment, including but not limited to hardware-based Trusted Execution Environments (TEEs). "Append-only record" means a data structure in which only the addition of new records is permitted, and existing records cannot be modified or deleted. "Independent arithmetic entity" means an entity that does not belong to the administrator of any terminal device constituting the system and performs processing while preventing access to the processing and stored contents by the administrator of any terminal device. A "synchronous proof value" is a value calculated using the cryptographic summary values ​​of two or more different input data, and is a value that cannot be generated without all of those two or more input data. An "attribute proof value" is a value that includes the cryptographic summary value of program code executed within an execution environment, and is a value that allows a remote verifier to cryptographically verify that the intended program is being executed within that execution environment. A "predetermined technical condition" is a condition that is recorded in an append-only record at the start of a transaction, and whose satisfaction is technically verifiable by the system and does not depend on human subjective judgment.

[0063] (See Figure 3) Referring to Figure 1, the overall configuration of the transaction process management system 100 according to an embodiment of the present invention will be described. Referring to Figure 2, the processing pipeline of the present invention consists of six steps: K1 (data verification freeze), K2 (synchronous proof generation), K3 (automatic state transition), K4 (tamper detection chain), K5 (read access based on technical conditions), and K6 (arbiter independence and proof of authenticity). Referring to Figure 3, in a conventional party server configuration, the record manager belongs to the parties to the transaction and independence is not technically guaranteed, whereas in the present invention, the processing device (S1) performs all processing within an execution environment with a protection mechanism and provides proof of authenticity values ​​to the terminal device, which is structurally different.

[0064] (See Figure 4) The transaction process management system 100 comprises at least one first terminal device (T1), at least one second terminal device (T2), a processing unit (S1), and a network (N) connecting them.

[0065] (See Figure 5) The first terminal device (T1) is a terminal device used by the client (e.g., the ordering party) and is used for transmitting specification data, receiving acknowledgments, displaying judgment results, and verifying constructive proof values. T1 has a processor, storage device, communication interface, and input / output device.

[0066] (See Figure 6) The second terminal device (T2) is used by the contractor (e.g., the subcontractor) and is used for receiving specifications, sending confirmation response data, sending deliverable data, displaying judgment results, and verifying the constructive proof values. T2 has the same configuration as T1.

[0067] (See Figure 7) The processing unit (S1) is an independent computing entity that does not belong to the administrator of any terminal device constituting system 100. S1 has a processor, a storage device for storing append-only records, and an execution environment with a protection mechanism. S1 is implemented, for example, as a server installed in an independent third-party data center. Neither the administrator of T1 nor the administrator of T2 has management authority over S1. Furthermore, due to the execution environment with a protection mechanism of S1, even the operator of the data center where S1 is installed cannot read or modify the computing process or stored contents of S1.

[0068] (See Figure 8) Execution environments with S1 protection mechanisms are implemented, for example, as Intel SGX (Software Guard Extensions), ARM TrustZone, or AMD SEV (Secure Encrypted Virtualization). These hardware-based TEEs provide a cryptographically isolated region (enclave) within the processor, protecting program code and data within the enclave from any software outside the enclave (including operating systems, hypervisors, and firmware). Furthermore, the TEE has remote attestation capabilities, generating an attribution value containing a cryptographic summary of the program code executed within the enclave, allowing remote terminal devices to verify the integrity of S1 processing. While side-channel attacks against TEEs (e.g., Spectre, Foreshadow) are known, TEE manufacturers continuously provide microcode updates and software mitigation measures, and terminal devices can verify that the TEE's firmware and microcode versions are up-to-date through verification of the attribution value.

[0069] (See Figure 9) Network (N) is a communication network connecting T1, T2, and S1, and can be, for example, the Internet, a VPN, or a dedicated line.

[0070] (See Figure 10) Refer to Figure 4 to explain the process of step (a).

[0071] Prior to the commencement of a transaction, both T1 and T2 transmit agreed data regarding a predetermined minimum unit to S1. S1 verifies the agreed data from both parties and, only if the agreed data matches, sets the predetermined minimum unit based on that data and records this setting in an append-only record. If the agreed data from both parties does not match, S1 does not set a minimum unit and records the fact of the mismatch in an append-only record. This ensures that the predetermined minimum unit is set objectively based on the agreement of both parties and is not subject to arbitrary manipulation by either party.

[0072] T1 transmits specification data to S1. The specification data includes numerical data that defines the quality of the traded item. For example, in the procurement of GPU computing resources, this includes performance values ​​such as TFLOPS, memory bandwidth, and uptime; in the vaccine supply chain, it includes quality values ​​such as storage temperature, transport temperature, and active ingredient concentration; and in semiconductor manufacturing, it includes physical quantities such as wafer thickness, wiring width, and resistance.

[0073] The S1 processor, within an execution environment with a protection mechanism, converts each received data into an integer value by dividing it by a predetermined minimum unit. For example, dividing a GPU computing power of 312.5 TFLOPS by the minimum unit 0.1 TFLOPS (μ = 0.1) yields the integer value V_int = 3125. Similarly, dividing a vaccine storage temperature of -70.0°C by the minimum unit 0.1°C (μ = 0.1) yields the integer value V_int = -700. This conversion converts floating-point numbers to integers, allowing all subsequent operations to be performed using only integer arithmetic. If the result of the division is not an integer (for example, when V_raw = 312.7 and μ = 0.5, V_raw / μ = 625.4), the fractional part is truncated to obtain an integer value (floor(V_raw / μ)). The fact of truncation and the value before truncation are recorded in an append-only record and are subject to independent re-verification in step (g). Because the truncation rules are clearly defined in advance, the same integer value can be obtained on any processor.

[0074] The processor in S1 generates a cryptographic summary value for each integer value. In this embodiment, the SHA-256 hash function is used. The generation of the cryptographic summary value makes it possible to detect tampering with the original data.

[0075] S1 irreversibly fixes the record in the storage device. Here, "irreversibly fixed" is guaranteed by the combination of the append-only record structure in step (d) (a data structure that makes it impossible to modify or delete existing records) and the execution environment with a protection mechanism in step (f) (which prevents any administrator of any terminal device from accessing the stored contents). In other words, the append-only record structure prevents tampering at the software level, and the execution environment with a protection mechanism prevents tampering at the hardware level, thus achieving double protection. After fixing, it is impossible to modify or delete the record. Also, corresponding to step (g), S1 records the input values ​​(V_raw and μ) and output values ​​(V_int) of the integer conversion in the append-only record. This allows T1 and T2 to independently re-execute and verify the V_raw / μ = V_int operation.

[0076] Refer to Figure 5 to explain the process in step (b).

[0077] T2 receives the specifications and sends its interpretation to S1 as acknowledgment data.

[0078] The processor of S1 generates a synchronization proof value by the following calculation within an execution environment with a protection mechanism. - Let H(d1) be the cryptographic summary value generated in step (a). - Let H(d2) be the cryptographic summary value of the acknowledgment data. - Calculate the synchronous proof value = H(H(d1) || H(d2)) (|| represents bit concatenation)

[0079] This synchronous proof value cannot be generated without input from both T1 (d1) and T2 (d2). Since H(d2) is unknown from T1 alone, and H(d1) is unknown from T2 alone, the synchronous proof value cannot be generated from either side alone. This mathematically proves that both sides have confirmed the same specifications. S1 records the input values ​​H(d1) and H(d2) and the output value used to generate the synchronous proof value in an append-only record (corresponding to step (g)).

[0080] Refer to Figure 6 to explain the process of step (c).

[0081] T2 sends the deliverable data (inspection data of the manufacturing results) to S1.

[0082] The processor of S1 converts the artifact data into an integer value by dividing it by the same predetermined minimum unit as in step (a) within the execution environment which has a protection mechanism.

[0083] The S1 processor performs comparison processing using only integer arithmetic. Specifically, it calculates the absolute difference between the integer value of the output data and the integer value fixed in step (a). If this absolute value is less than or equal to a predetermined integer threshold, it determines that there is a match; if it exceeds the threshold, it determines that there is a mismatch.

[0084] For example, if the baseline value of GPU computing power is V_int = 3125 (= 312.5 TFLOPS / 0.1 TFLOPS), the measured value is V_int_actual = 3100 (= 310.0 TFLOPS / 0.1 TFLOPS), and the threshold is T = 50 (= ±5.0 TFLOPS), then the difference Δ = |3100 - 3125| = 25, and since 25 ≤ 50, it is determined to be a match. S1 records the input values ​​(V_int = 3125, V_int_actual = 3100, T = 50) and output values ​​(Δ = 25, determination = match) of this comparison process in an append-only record (corresponding to step (g)). T1 or T2 can independently recalculate |3100 - 3125| = 25, 25 ≤ 50 using the recorded input values ​​and verify the agreement with the output value of S1.

[0085] Because this comparison process consists solely of integer arithmetic, the same result can be obtained on any processor. Since floating-point arithmetic is not used, rounding errors between the CPU, GPU, and OS do not occur in principle.

[0086] If a match is found, S1 automatically fixes the delivery record for that transaction interval. If a mismatch is found, S1 automatically fixes the negotiation record for that transaction interval. This transition from determination to record fixing is performed automatically without human intervention.

[0087] If the deliverable data includes multiple inspection items (for example, four items: GPU computing power, memory bandwidth, uptime, and response delay), the processor in S1 performs a comparison process for each inspection item individually, and determines that the transaction interval is a match only if all inspection items match.

[0088] Refer to Figure 7 to explain the process of step (d).

[0089] The processor of S1 records the results of each step (a) through (c) in the storage device as append-only records. Each record contains a cryptographic summary of the previous record.

[0090] An example of a record structure is shown below. - entry_id: sequential number of the record - step: Identifier of the corresponding step - prev_hash: Cryptographic summary of the previous record (NULL for the first record) - timestamp: Coordinated Universal Time (UTC) timestamp - run_id: A unique identifier assigned to each transaction unit. - data_hash: Cryptographic summary of the data in the record. - action: The type of operation performed - input_values: Input values ​​for integer operations corresponding to step (g) - output_values: Output values ​​of integer operations corresponding to step (g) - attestation_hash: Cryptographic summary of the proof value of step (f)

[0091] The prev_hash of the first record is NULL. Each subsequent record contains the cryptographic summary of the entire preceding record as its prev_hash. This allows any modification, deletion, or insertion of a record at any point in time to be detected as a break in the chain.

[0092] Chain integrity is verified as follows: The processor in S1 sequentially recalculates the cryptographic summary value of all records from the beginning and checks for a match with the prev_hash already recorded in each record. If all matches, chain integrity is confirmed; if there are no matches, tampering is detected. Furthermore, based on step (g), T1 or T2 can independently recalculate the output_values ​​from the input_values ​​of each record and verify a match with the recorded values. This external verification is an independent verification method that does not fall into the circular reasoning where S1 verifies itself.

[0093] Refer to Figure 8 to explain the process in step (e).

[0094] In this invention, emergency access is automatically controlled based on predetermined technical conditions, rather than relying on subjective human judgment as in the conventional Break-the-Glass (BTG) method. These technical conditions are irreversibly recorded in the dedicated append record at the start of the transaction and cannot be changed afterward.

[0095] The first form of the technical condition is a time-limited condition. If authenticated access is not performed from any terminal device included in the system for a predetermined period (e.g., 72 hours), S1 automatically releases the record for that transaction period in read-only mode. This allows access to the record even when operations are impossible due to an accident involving the person in charge, a natural disaster, etc. The length of the time limit is agreed upon by both parties at the start of the transaction and is recorded in a record dedicated to appending.

[0096] The second form of the technical requirement is threshold authentication. At the start of a transaction, multiple emergency keys (e.g., five emergency keys) are pre-registered, and access is granted only if authentication is successful using a predetermined number (e.g., three or more) of these emergency keys. Each emergency key is distributed to, for example, each of the transaction parties, the insurance company, the arbitration body, and the regulatory authority. Access cannot be granted by any single key holder; a predetermined number of agreements are required.

[0097] In either configuration, the scope of read-only access can be limited to records of transaction intervals related to technical conditions. This prevents records of unrelated transactions from being unnecessarily accessed. S1 appends the facts of the access operation (access date and time, type of conditions satisfied, access scope, access requester, etc.) to the concatenation of the read-only record.

[0098] Referring to Figure 9, the process of step (f) will be explained. S1 (processing unit) executes all of the processes of steps (a) through (e) as an independent computing entity that does not belong to the administrator of T1, the administrator of T2, or the administrator of any terminal device included in system 100, within an execution environment with a protection mechanism.

[0099] Specifically, S1 is installed in an independent third-party data center (for example, a data center of a cloud service provider). Neither the administrator of T1 (the client) nor the administrator of T2 (the contractor) has management authority over S1. In addition, S1's protected execution environment (TEE) prevents even the data center operator from reading or modifying S1's computation process and stored contents. S1's processor performs all hash operations, synchronization proof generation, integer arithmetic checks, chain management, and access control within the TEE. T1 and T2 only input data and display results; they do not perform any computational processing.

[0100] The TEE of S1 generates an attribution value and sends it to T1 and T2, respectively. The attribution value includes a cryptographic summary of the program code executed within the TEE, the TEE's hardware identifier, and the TEE manufacturer's signature. T1 and T2 verify the signature of the attribution value using the TEE manufacturer's public key and confirm that the cryptographic summary of the program code matches a pre-agreed value. This allows T1 and T2 to remotely cryptographically verify that S1 is executing the intended program. The attribution value is also recorded in an append-only record.

[0101] This configuration ensures, both contractually and technically, that all processing is performed independently of the parties to the transaction. While in conventional blockchains each node belongs to either a miner or a validator, in this invention, S1 functions as a single logical operating entity technically isolated from all parties within an execution environment with a protection mechanism. In conventional trusted third parties (TTPs, notaries, escrows, etc.), the independence of the third party is guaranteed only contractually or legally, whereas in this invention, it is technically guaranteed by hardware-level protection mechanisms and proof values. Furthermore, the independence of the processing unit (S1) can be technically verified externally by a terminal device through verification of the proof value. Specifically, the terminal device verifies the signature of the proof value using the TEE manufacturer's public key and confirms that the cryptographically summarized value of the program code contained in the proof value matches a pre-agreed value. Since the signature of the proof value can only be generated with the TEE manufacturer's private key, it is cryptographically proven that the processing unit is executing the intended program within an execution environment with a protection mechanism. This makes it possible to verify the independence of the processing unit through externally observable technical means (receiving and verifying the signature of the proof value), which also contributes to the detection of infringement.

[0102] In the following embodiment, all transaction data is represented using abstract variables. V_raw represents raw data (real values), μ represents a predetermined minimum unit, V_int = V_raw / μ represents an integer value, Δ represents the absolute value of the difference, and T represents a predetermined integer threshold. This representation is not limited to a specific industry and is common to all industrial fields to which the present invention can be applied.

[0103] Example 1: Industry-independent basic flow (matching system, abstract variable representation)

[0104] Referring to Figures 1, 4 through 9, and 10, the basic flow, which is independent of the industry, will be explained. In this embodiment, the client (T1) requests the contractor (T2) to deliver the deliverables, and the processing unit (S1) is an independent computing entity on the cloud that does not belong to either T1 or T2's administrator, and executes all processing within an execution environment with a protection mechanism.

[0105] Prior to the start of a transaction, both T1 and T2 send the minimum unit of agreement data to S1. For example, if T1 sends "μ = 0.1" and T2 also sends "μ = 0.1", S1 confirms the agreement between the two, sets μ = 0.1 as the predetermined minimum unit, and records it in a record dedicated to appending data.

[0106] In step (a), T1 sends n specification data items (V_raw1, V_raw2, ..., V_rawn) to S1. S1 divides each V_rawi by a predetermined minimum unit μi within the TEE to convert it into an integer value V_inti = V_rawi / μi. S1 generates a SHA-256 cryptographic summary value H(V_inti) for each V_inti and irreversibly fixes the record in the storage device. S1 records the input values ​​(V_rawi, μi) and output values ​​(V_inti) for each conversion into an append-only record (corresponding to step (g)).

[0107] In step (b), T2 sends the interpretation of the specification to S1 as acknowledgment data. S1 concatenates the cryptographic summary H(d1) from step (a) and the cryptographic summary H(d2) of the acknowledgment data within the TEE to generate a synchronous proof value H(H(d1) || H(d2)). This value cannot be generated without input from both T1 and T2, mathematically proving that both have confirmed the same specification.

[0108] In step (c), T2 sends the deliverable data (V_raw_actual1, V_raw_actual2, ..., V_raw_actualn) to S1. S1 divides each V_raw_actuali by the same minimum unit μi within the TEE to convert it to an integer value V_int_actuali and calculates the difference Δi = |V_int_actuali - V_inti|. If Δi ≤ Ti (a predetermined integer threshold) for all i, S1 determines it to be a "match" and automatically fixes the delivery record for that transaction interval. If Δi > Ti for any i, S1 determines it to be a "mismatch" and automatically fixes the negotiation record for that transaction interval. S1 records all input and output values ​​of the comparison process in a record for appending (corresponding to step (g)). T1 or T2 can independently re-execute the same integer operation using the recorded input values ​​and verify the agreement with S1's output values. Since this comparison process consists solely of integer arithmetic, the same result is guaranteed on any processor, including x86, ARM, and CUDA.

[0109] In step (d), S1 concatenates the results of steps (a)-(c) as append-only records. Each record contains the cryptographic summary value prev_hash of the previous record, and records the UTC timestamp, transaction identifier run_id, input value, output value, and cryptographic summary value of the proof value.

[0110] In step (e), if there is no authenticated access from any terminal device within a specified period (e.g., 72 hours) due to an accident involving the person in charge, a natural disaster, a communication disruption, etc., S1 automatically releases the record for that transaction section in read-only mode and appends the fact of the release operation (release date and time, satisfaction conditions, release scope) to the concatenation of the append-only record.

[0111] In step (f), all the processes in steps (a)-(e) above are performed by S1 as an independent operating entity within the TEE, separate from all other parties. S1's TEE generates an attribution value and provides it to T1 and T2. T1 and T2 verify the attribution value with the TEE manufacturer's public key and confirm that S1 is executing the intended program.

[0112] In step (g), since the input and output values ​​of each integer operation in steps (a)-(c) are recorded in a record dedicated to appending, T1 or T2 can at any time independently re-execute the integer operation using the input values ​​to verify the validity of S1's decision.

[0113] Example 2: Procurement of GPU computing resources for a mega AI company (matching system, global)

[0114] An example of applying the present invention to a transaction with a mega AI company will be described with reference to Figures 1, 4 through 9, and 18. This example describes a case where technology company X (T1) procures GPU computing resources from GPU computing resource provider Y (T2) (e.g., cloud GPU service providers such as NVIDIA, Google, Microsoft, and Amazon). S1 is an independent processing unit that does not belong to either X or Y, and performs objective determination of contract compliance within the TEE.

[0115] Prior to the commencement of trading, both Company X and Company Y send the minimum agreed-upon data (e.g., computing power μ = 0.1 TFLOPS, bandwidth μ = 0.1 GB / s, uptime μ = 1 second) to S1, and S1 confirms and sets the agreement between both parties.

[0116] In step (a), Company X (T1) sends contract specification data to S1. The specification data includes GPU computing power V_raw1 (in TFLOPS), memory bandwidth V_raw2 (in GB / s), GPU uptime V_raw3 (in seconds), and tolerances for each item. S1 converts each V_rawi within the TEE to an integer value V_inti by dividing it by a predetermined minimum unit μi (e.g., a computing power of 312.5 TFLOPS is divided by μ=0.1 TFLOPS to convert it to the integer value 3125). S1 generates a cryptographic summary value for each integer value using SHA-256 and irreversibly fixes the record. The input and output values ​​are recorded in an append-only record.

[0117] In step (b), Company Y (T2) sends an acknowledgment of its interpretation of the specifications to S1 as confirmation data. S1 generates a synchronous proof value H(H(d1) || H(d2)) within the TEE. This value cannot be generated without input from both Company X and Company Y, and mathematically proves that both parties have agreed on the contract specifications for the GPU computing resources.

[0118] In step (c), Company Y (T2) sends actual data on GPU computing resources (V_raw_actual1, V_raw_actual2, V_raw_actual3) to S1. S1 converts each actual value to an integer V_int_actuali within the TEE and calculates the difference Δi = |V_int_actuali - V_inti|. If Δi ≤ Ti for all items, S1 determines it to be a "match" and automatically fixes the performance record for that contract segment. If Δi > Ti for any item (for example, the actual value is 3050 against a contracted computing capacity of 3125, with a difference of 75 exceeding the threshold of 50), S1 determines it to be a "mismatch" and automatically fixes the negotiation record for that contract segment. All input and output values ​​are recorded in an append-only record. Company X or Company Y can independently re-verify using the recorded input values.

[0119] In step (d), S1 concatenates the results of steps (a)-(c) as append-only records. Each record includes a UTC timestamp, contract identifier, input value, output value, and proof value. Since the procurement of GPU computing resources is conducted internationally, the uniqueness of the time is guaranteed by the UTC timestamp, even if companies X and Y are located in different time zones.

[0120] In step (e), for example, if X Company's AI development project is in an urgent situation (service failure, security incident, etc.) and X Company's contract representative is unavailable, when a pre-set time-limited condition (e.g., 72 hours of no authenticated access) is met, S1 automatically releases the record for that contract section in read-only mode and adds a record of the release operation.

[0121] In steps (f) and (g), all of the above processing is performed by S1 as an independent computational entity within the TEE, and the attribution values ​​are provided to companies X and Y. Neither company X nor company Y has management rights over S1, and the legitimacy of S1's processing can be technically verified by the TEE's attribution values ​​and independent re-verification of integer operations. A similar flow can be applied to AI model licensing agreements, training data procurement agreements, and API SLA compliance.

[0122] Example 3: Global supply chain, multi-party transactions, and satellite link compatibility

[0123] The application of the present invention in a multi-party global supply chain will be explained with reference to Figures 1, 15, 16, and 19. This embodiment corresponds to all of claims 11 to 17 (Group C: Transaction Structure Extension).

[0124] This embodiment describes a case where a North American ordering company A (T1) orders parts from a Southeast Asian manufacturing company B (T2), and an independent European inspection company C performs quality inspections. The processing unit (S1) is an independent computing entity not belonging to any of the three companies and consists of multiple geographically distributed processing nodes (North American node, European node, Asian node). Each processing node executes the same program code within the TEE, and the execution of the same processing is verified by the matching of the proof-of-concept values ​​of each node. Each processing node maintains an identical copy of an append-only record and is synchronized via a multiplexed network of terrestrial internet, LEO satellite communication (Starlink, OneWeb, Kuiper, etc.), and submarine cables (corresponding to claim 21).

[0125] Here, S1 consists of multiple physical nodes, but since all nodes execute the same program code within the TEE and the proof-of-concept values ​​match, S1 functions as a single logical operating entity. This is structurally different from blockchains where different node operators run different software. The distributed nodes of this invention aim to ensure redundancy, not consensus. Since all processing nodes execute the same program code within the TEE and produce the same integer arithmetic output for the same input, replication is realized as synchronous replication following a single write order, and in principle, result discrepancies between nodes do not occur.

[0126] The transaction structure is as follows: - Transaction interval α: Company A (T1) → Company B (T2): Ordering and manufacturing of parts - Trading interval β: Company B (functioning as T1 in another trading interval) → Company C: Quality inspection Here, Company B functions as T2 in transaction interval α, and as the first terminal device (T1) in another transaction interval (transaction interval β) (corresponding to claim 11). Inspection organization Company C is a third-party inspection terminal device G that functions as a third terminal device different from both the first and second terminal devices, and Company C's identification information is recorded in an append-only record (corresponding to claim 15). The protection mechanism in step (f) also prevents Company C's administrator from accessing the processing device (S1) (corresponding to claim 12).

[0127] In step (a), Company A (T1) sends component specification data (V_raw, μ, T for each item) to S1. S1 converts each V_raw to an integer value V_int within the TEE and generates and fixes a cryptographically summarized value. Additionally, a cryptographically summarized value of the specification data for raw materials supplied from Company A to Company B is generated and fixed (corresponding to claim 13). All input and output values ​​are recorded in a record dedicated to appending. Communication between T1 and S1 operates over any transport, including LEO satellite communication (corresponding to Figure 16).

[0128] In step (b), Company B (T2) sends acknowledgment data to S1, and S1 generates a synchronization proof value within the TEE. Although Company A in North America and Company B in Southeast Asia are located in different time zones, the uniqueness of the time is guaranteed by the UTC timestamp.

[0129] In step (c), a comparison process using only integer arithmetic is performed on each of the multiple deliverable data separated by process, and the judgment result for each process is recorded individually in an append-only record (corresponding to claim 14). The deliverable data is received from a third-party inspection terminal device G (Company C in this embodiment) (corresponding to claim 15). All input and output values ​​are recorded and independent re-verification is possible.

[0130] If S1 determines that there is a mismatch in the transaction interval α, S1 notifies the terminal device related to the relevant transaction interval β of the fact of the mismatch (corresponding to claim 16). This allows mismatch information to be shared in real time across the entire supply chain spanning three continents.

[0131] S1 generates an integrated verification value by concatenating the final cryptographic summary values ​​of each of the append-only records for transaction interval α and transaction interval β (corresponding to claim 17). This integrated verification value enables the integrity of the entire global supply chain spanning multiple transaction intervals to be verified at once.

[0132] In step (e), even if the normal communication infrastructure is disrupted due to a natural disaster or other event at a manufacturing base in Southeast Asia, the records of S1 can be accessed via LEO satellite communication (such as Starlink), and when a pre-set time-limited condition or threshold authentication is met, S1 automatically releases the records in read-only mode, and the fact of the release operation is added. The release scope is limited to the records of the transaction interval related to the technical conditions (corresponding to claim 22).

[0133] In steps (f) and (g), all of the above processing is performed by S1 as an independent operating entity within the TEE, and provides the proof-of-concept values ​​to all terminal devices. Neither Company A, Company B, nor Company C has any control over S1.

[0134] Example 4: Pandemic Response, Government Procurement, and Force Majool Fault Tolerance

[0135] The application of the present invention in pandemic response, government procurement, and force majeure events will be illustrated with reference to Figures 1, 4 through 9, 20, and 21. This embodiment corresponds to claims 18 through 22 (Group D: Technological Extensions).

[0136] This embodiment describes a case where an international organization P (e.g., a procurement agency such as the WHO, UNICEF, or World Bank, T1) entrusts the manufacturing and supply of vaccines to a vaccine manufacturing company Q (T2). IoT sensor devices such as temperature sensors and humidity sensors are connected to Q's manufacturing line (corresponding to claim 18). S1 is an independent processing unit that does not belong to either P or Q, and consists of geographically distributed processing nodes (Geneva, New York, Singapore, etc.), with each node executing the same program code within the TEE.

[0137] In step (a), Company P (T1) sends vaccine quality specification data (storage temperature V_raw1, transport temperature V_raw2, active ingredient concentration V_raw3, etc.) to S1. S1 divides each V_rawi within the TEE by a predetermined minimum unit μi to convert it to an integer value V_inti, and generates and fixes a cryptographically summarized value. For example, a storage temperature of -70.0°C is divided by μ=0.1°C to convert it to an integer value of -700. All input and output values ​​are recorded in a record for appending only.

[0138] In step (b), Company Q (T2) sends its interpretation of the specifications to S1 as confirmation response data, and S1 generates a synchronous proof value within the TEE.

[0139] In step (c), output data is automatically acquired from the IoT sensor device on Q's manufacturing line (corresponding to claim 18) and transmitted to S1 via T2. S1 performs comparison processing using only integer arithmetic within the TEE. When an inference model (AI model) is used to calculate the reference value for the comparison processing in step (c), referring to Figure 17, S1 divides the output of the inference model within the TEE by a predetermined minimum unit μ to convert it to an integer value and uses it as the reference value for the comparison processing. As a result, even if the internal calculations of the inference model include floating-point numbers, the comparison processing itself is performed using only integer arithmetic, and the determinism of the determination result is maintained. S1 records the cryptographic summary value (SHA-256 hash value) and version information of the inference model in an append-only record, and the calculations by the inference model and the conversion of its output to an integer value are included in the target of independent re-execution in step (g) (corresponding to claim 19). In independent re-execution, the terminal device generates output using the same inference model (identity verified by cryptographic summary values), and verifies that the result of converting this output to an integer value using the same minimum unit μ matches the integer value recorded by S1. This allows for post-verification of the identity of the AI ​​model used for the determination and the validity of the determination result.

[0140] In step (d), S1 concatenates and records each result as an append-only record. Each record is affixed with the electronic signature of the terminal device or processing device involved in the generation of the record (corresponding to claim 20). Multiple processing nodes of the processing device (S1) maintain identical copies of the append-only record, and are synchronized among all processing nodes via a multiple redundant network including a satellite link (corresponding to claim 21). Input values, output values, and attribution values ​​are recorded in each record.

[0141] In step (e), even if normal communication infrastructure is disrupted due to force majeure events such as lockdowns caused by pandemics, wars, riots, natural disasters, or economic sanctions, S1 can be accessed via LEO satellite communication, and S1 automatically releases the records in read-only mode when pre-configured technical conditions (time-limited conditions or threshold authentication) are met. The scope of release is limited to records of the transaction interval related to the technical conditions (corresponding to claim 22). In the case of threshold authentication, for example, an emergency key is distributed to five parties: Company P (WHO), Company Q (vaccine manufacturer), insurance company, international arbitration body, and supervisory authority, and access is permitted when authentication is successful for three or more parties (corresponding to claim 24). This provides objective evidence for insurance claims, international arbitration, and contract resumption from append-only records. Force majeure fault tolerance has a four-layer structure. Layer 1 (record survival): Records survive even when a specific area is destroyed by geographically distributed nodes. Layer 2 (access): Step (e) operates via satellite link. Layer 3 (Dispute Resolution): Dedicated append records serve as objective evidence. Layer 4 (Network Fault Tolerance): Multiple redundant configurations of terrestrial, satellite, and mobile communications eliminate single points of failure.

[0142] In steps (f) and (g), all of the above processing is performed by S1 as an independent computing entity within the TEE, and provides the proof-of-concept values ​​to all terminal devices. Neither Company P nor Company Q has management authority over S1. All input and output values ​​of integer operations are recorded in append-only records, allowing for independent re-verification by Company P and Company Q. Similar flows can be applied to government procurement of supplies and international data sharing, in addition to pandemic response.

[0143] Example 5: Platform transaction protection and identity theft prevention

[0144] Referring to Figure 1, an embodiment of applying the present invention to transaction process management and identity theft prevention on an internet platform will be described. This embodiment describes a case where a buyer U1 (T1) and a seller U2 (T2) conduct a transaction on a marketplace provided by platform operator R. The processing unit (S1) is an independent processing unit that does not belong to R, U1, or U2, and executes all processing as a computing entity independent of the platform operator within the TEE.

[0145] In this embodiment, in conventional platform trading, transaction determination and record management are performed on the server of platform operator R, making it impossible to rule out the possibility that R may alter transaction records, and users have no means to verify R's determination algorithm. In the present invention, the processing unit (S1) performs all processing within an execution environment (TEE) that has a protection mechanism independent of R's server, and no administrator of any terminal device, including R, can access the calculation process and stored contents of S1.

[0146] In step (a), the buyer U1 (T1) sends product specification data (V_raw and μ for each item such as product specifications, quantity, and price) to S1. S1 divides each V_raw by a predetermined minimum unit μ within the TEE to convert it to an integer value V_int and generates a cryptographically summarized value to irreversibly fix the record. All input and output values ​​are recorded in a record for appending only.

[0147] In step (b), seller U2 (T2) sends specification confirmation data to S1, and S1 generates a synchronization certificate value H(H(d1) || H(d2)) within the TEE. This synchronization certificate value cannot be generated without both U1's input data (d1) and U2's input data (d2). If each terminal device communicates with S1 using cryptographic authentication means (digital certificate, public key cryptography, etc.) and digital signatures are added to each record (corresponding to claim 20), an attacker would need to obtain the private key of the person being impersonated in order to impersonate U1 or U2. Since the authentication data processed within S1's TEE is not readable from the outside, the theft of authentication information by a man-in-the-middle attack is also eliminated. Therefore, the existence of the synchronization certificate value functions as cryptographic evidence that both legitimate parties to the transaction actually participated in the transaction process. This technically eliminates the possibility of an attacker impersonating one of the legitimate parties to complete a transaction in impersonation fraud, special fraud (e.g., "ore-ore" fraud), public institution impersonation fraud, and romance fraud.

[0148] In step (c), seller U2 (T2) sends deliverable data (inspection data of delivered products) to S1. S1 performs a comparison process using only integer arithmetic within the TEE to determine whether there is a match or a mismatch. This determination is not performed by the algorithm of platform operator R, but by an integer arithmetic logic based on criteria agreed upon in advance by both parties within S1's TEE, thus structurally eliminating arbitrary manipulation of the determination result by R.

[0149] In step (d), S1 concatenates and records all results as append-only records. No entity, including platform operator R, can modify or delete append-only records within the TEE. This makes it technically impossible for the platform operator to tamper with transaction records, manipulate seller ratings, or selectively delete transaction history. Steps (e) through (g) are performed in the same manner as in Example 1.

[0150] The configuration of this embodiment provides the following benefits: Firstly, the "single point of confidence" problem caused by the centralized data management of the platform operator is resolved by the independence of the processing unit (S1) and the protection mechanism of the TEE. Secondly, the genuine participation of both parties to a transaction is cryptographically guaranteed by the synchronous proof value, and impersonation fraud is technically eliminated. Thirdly, the chain structure of append-only records makes the entire record of the transaction process tamper-proof, and evidence of fraudulent activity is preserved. Fourthly, independent reverification of integer operations allows users to verify the determination of any entity, including the platform operator, and the problem of algorithmic opacity is resolved.

[0151] Example 6: Content Authenticity Proof and AI Forgery Resistance Architecture

[0152] Referring to Figure 1, an example of applying the present invention to content authenticity verification and AI forgery resistance will be described. With the development of artificial intelligence technology, the accuracy of deepfakes (AI-generated fake images, videos, and audio) has improved, and the probability that all authentication methods based on pattern matching, including biometric authentication (fingerprints, faces, irises, voiceprints, etc.), will be bypassed by AI has increased. This embodiment describes a content authenticity guarantee configuration based on cryptographic computation that does not rely on pattern matching authentication.

[0153] In this embodiment, the terminal device (T1) of the content creator (reporter, photographer, creator, etc.) and the terminal device (T2) of the content verifier or distributor are connected via a processing device (S1). S1 is an independent processing device that does not belong to the administrator of either terminal device and performs all processing within the TEE. The content creator's T1 includes a shooting device or content creation device (e.g., a smartphone, camera, PC, etc. equipped with a TEE) that has an execution environment with a protection mechanism.

[0154] In step (a), the content creator (T1) sends content data (images, videos, audio, text, etc.) to S1. S1 generates a cryptographic summary value of the content data within the TEE and irreversibly fixes the record along with the content metadata (generation date and time, identification information of the generating device, location information, etc.). This cryptographic summary value will be completely different even if only one bit of the content is changed, so AI-induced content modification (deepfake processing, text alteration for fake news generation, etc.) is immediately detected.

[0155] In step (b), the content verifier or distributor (T2) sends confirmation data of receipt of the content to S1, and S1 generates a synchronization certificate value within the TEE. This synchronization certificate value cannot be generated without input from both the content creator (T1) and the content verifier (T2). Here, authentication of each terminal device is not based on pattern matching (biometric authentication, etc.), but on an electronic signature using a secret key stored within the TEE. The secret key is unreadable from the outside due to the TEE's hardware protection mechanism, and it is cryptographically impossible for an AI to guess the secret key (based on the computational difficulty of the elliptic curve discrete logarithm problem). Therefore, even if an era arrives in which AI can completely forge biometric information such as fingerprints, faces, irises, and voiceprints, it will be mathematically impossible to forge the synchronization certificate value based on the secret key in the TEE, and the authenticity guarantee of the present invention will be maintained.

[0156] The trust model in this embodiment represents a fundamental shift from a "pattern matching" type to a "cryptographic computation" type. Pattern matching authentication (fingerprint authentication, facial recognition, voiceprint authentication, etc.) is a method that determines the similarity between input data and registered data, and can be breached if the AI ​​generates false data with accuracy that deceives the similarity determination. In contrast, the cryptographic computation trust model of the present invention has three layers: (i) the one-way nature of cryptographic hash functions such as SHA-256 (it is computationally difficult to reverse-calculate the input from the output), (ii) the confidentiality of the private key of the public-key cryptography scheme (it is possible to generate a signature from the private key, but it is computationally difficult to deduce the private key from the signature), and (iii) hardware isolation of the TEE (the private key does not leave the TEE). These layers constitute a trust foundation that cannot be breached by mathematical principles, no matter how much the computational power of the AI ​​improves.

[0157] Steps (c) through (g) are performed in the same manner as in Example 1. After the content is delivered, any viewer's terminal device can independently verify that the content has not been altered since its creation and that the genuine participation of the content creator is cryptographically proven by (i) recalculating the cryptographic summary value of the content and verifying its match with the recorded value, (ii) verifying the integrity of the chain, and (iii) verifying the proof value. This verification can be performed automatically by an application on a smartphone, and even general users without specialized knowledge can determine the authenticity of the content simply by checking the display of the verification results (e.g., "Verified" or "Unverified").

[0158] The configuration of this embodiment provides the following benefits. Firstly, since the entire process from content generation to distribution is recorded in an unalterable manner by a chain structure of append-only records, it becomes possible to detect deepfakes (AI-generated fake images, videos, and audio) and fake news. Secondly, since the trust model is based on cryptographic computation rather than pattern matching authentication, the authenticity guarantee according to the present invention is maintained by mathematical principles no matter how advanced AI technology becomes (even if fingerprint forgery, facial forgery, voiceprint forgery, etc., become completely possible). Thirdly, since general users can automatically perform independent verification using smartphone applications, etc., it becomes possible to determine the authenticity of content without specialized knowledge. Fourthly, when the content generation device is equipped with TEE, hardware-level authenticity certification is provided at the time of content generation, making it possible to detect any AI processing after generation.

[0159] Interim report and warning

[0160] During the period between step (a) and step (c), interim report data may be sent from T2 to S1. S1 converts the interim report data into an integer value within the TEE and records the comparison result with a predetermined warning threshold (e.g., an integer value corresponding to a predetermined percentage of the tolerance) in an append-only record. If the warning threshold is exceeded, S1 notifies the relevant terminal device of the warning. This allows for the detection of signs of quality deviation before the final inspection (corresponding to claim 8).

[0161] Embodiment of the operating mechanism (corresponding to claims 6, 7, 9, and 10)

[0162] Referring to Figure 7, the operational mechanism of the append-only record chain is explained. Each record in step (d) includes a Coordinated Universal Time (UTC) timestamp indicating the time the record was created (corresponding to claim 6). Since the timestamp is based on an international standard, the uniqueness of the time is guaranteed even if the trading parties are located in different time zones.

[0163] The series of processes in steps (a) through (e) are linked by an identifier uniquely assigned to each transaction unit, and this identifier is recorded in each record (corresponding to claim 7). This ensures that the records of each transaction are clearly distinguished even when multiple transactions are occurring simultaneously.

[0164] If the deliverable data in step (c) includes multiple inspection items, the processing unit (S1) individually performs a comparison process using only integer arithmetic for each inspection item, and determines that the transaction interval is a match only if all inspection items match (corresponding to claim 9). If even one item is a mismatch, the entire transaction interval is deemed a mismatch, and the negotiation record is automatically fixed.

[0165] The processing device (S1) transmits the determination result in step (c) to the first terminal device (T1) and the second terminal device (T2) related to the transaction interval, respectively (corresponding to claim 10). The transmitted information includes the determination result (match or mismatch), the individual results for each inspection item, and the integer value used for the determination.

[0166] Multi-party multi-dimensional transactions (corresponding to claims 28 to 30)

[0167] In actual transactions, not only a single quality indicator but also multiple different evaluation dimensions (e.g., quality dimension, delivery dimension, cost dimension, compliance dimension, etc.) exist simultaneously. In this invention, when each data in step (a) includes data relating to multiple evaluation dimensions, the processing device (S1) sets a predetermined minimum unit μ individually for each evaluation dimension within the TEE, performs the comparison process in step (c) independently for each evaluation dimension, and records the determination result for each evaluation dimension individually in a record dedicated to appending (corresponding to claim 28). For example, in GPU computing resource procurement, three evaluation dimensions are set and determined independently: quality dimension (TFLOPS: μ1=0.1, T1=50), bandwidth dimension (GB / s: μ2=0.1, T2=10), and uptime dimension (seconds: μ3=1, T3=3600). μ and T for each dimension are set for each dimension based on mutually agreed data.

[0168] There may be predetermined dependencies between multiple evaluation dimensions. For example, a dependency may be such that the determination of the cost dimension is postponed if the quality dimension does not match, or the determination of the compliance dimension is performed only if the delivery date dimension matches. S1 automatically controls the execution or postponement of the comparison process for the second evaluation dimension according to the determination result of the first evaluation dimension, based on the definition of the dependency recorded in an append-only record at the start of the transaction, and records the fact of such control (date and time of control, type of dependency, control result) in an append-only record (corresponding to claim 29). Since the definition of the dependency itself is irreversibly recorded in the append-only record, it cannot be changed afterward.

[0169] S1 aggregates the judgment results for all evaluation dimensions and automatically fixes the delivery record as a match for the multidimensional integrated judgment result for the transaction interval only if all evaluation dimensions match. If there is a mismatch for any evaluation dimension, the negotiation record is automatically fixed as a mismatch for the multidimensional integrated judgment result (corresponding to claim 30). The multidimensional integrated judgment result, the individual judgment results for each dimension, and the control history of dependencies are all recorded in an append-only record, and the terminal device can independently re-verify the integer operations of each dimension based on step (g).

[0170] Multi-party transactions (claims 11 to 17) and multi-dimensional transactions (claims 28 to 30) can be applied in combination. For example, in the global supply chain of Example 3, transaction interval α is evaluated in three dimensions: quality, delivery time, and cost, while transaction interval β is evaluated in two dimensions: quality and compliance. The integrated verification value (claim 17) encompasses all evaluation dimensions across all transaction intervals. This allows for the completeness of the entire multi-party and multi-dimensional transaction process to be verified at once.

[0171] Structural differences from prior art

[0172] Referring to Figure 11, the structural differences between the blockchain architecture and the architecture of the present invention will be explained. Blockchain requires consensus building by multiple nodes, and each node has an interest as a miner or validator. In the present invention, the processing unit (S1) executes all processing as a computational entity technically isolated from all parties within an execution environment with a protection mechanism. Consensus building is not required, and the determination results based solely on integer arithmetic are the same across all processors, so in principle, discrepancies in results between multiple nodes do not occur. Furthermore, terminal devices can verify the legitimacy of S1's processing using proof values, and independent re-verification is possible through input / output records of integer arithmetic.

[0173] Referring to Figure 12, the position of the present invention relative to the prior art will be explained. When classified along two axes, the technical assurance of the independence of the processing entity (vertical axis) and deterministic processing (horizontal axis), the present invention is located in the realm of "technically guaranteed independent processing entity × deterministic (integer arithmetic) × independently reverifiable". Existing technologies such as DAML, Ethereum EVM, and Corda are not located in this realm. Conventional TTPs (notaries, escrow, etc.) only guarantee independence contractually and lack technical assurance. RFC 3161 TSA only provides time proof and does not perform verification. Git only provides version control and does not manage transactions. BTG does not have technically predefined emergency conditions.

[0174] Referring to Figure 13, the difference in reproducibility between floating-point and integer arithmetic is explained. In IEEE 754 floating-point arithmetic, a processor-specific rounding error ε occurs, which can lead to different judgment results for the same data on different processors (x86, ARM, CUDA, etc.). In the comparison process using only integer arithmetic of the present invention, rounding errors do not occur in principle, thus guaranteeing the same judgment result on any processor. This determinism is a prerequisite that enables independent re-verification in step (g).

[0175] Referring to Figure 14, the defense of the present invention against Section 101 of the U.S. Patent Act (Alice two-stage test) is explained. With respect to Step 2A Prong 1 (Applicability of judicial exceptions), the claims of the present invention are not directed to abstract ideas such as "organization of economic activity" or "mathematical relationships," but rather to specific technical solutions to particular technical problems of computer technology itself (non-determinism of floating-point arithmetic, lack of technical independence of processing entities, and tamperability of records) (see Enfish, LLC v. Microsoft Corp., 822 F.3d 1327 (Fed. Cir. 2016)). With respect to Step 2A Prong 2 and Step 2B, the present invention includes (i) a specific hardware configuration such as a processing unit equipped with an execution environment having a protection mechanism, (ii) a specific data conversion process such as setting a predetermined minimum unit based on mutually agreed data and converting it to an integer value, (iii) a technical constraint of comparison processing using only integer operations, (iv) a specific cryptographic operation such as concatenation of cryptographic summary values, (v) a specific data structure such as sequential concatenation of append-only records, (vi) a specific access control mechanism such as read-only release based on predetermined technical conditions, and (vii) a specific verification mechanism such as generation of proof values ​​and independent re-verification of integer operations, all of which are not abstract ideas but constitute specific technical improvements.

[0176] The seven technical improvements described above constitute an inventive concept not only as individual elements but also as an ordered combination. Specifically, the integer value conversion in step (a) enables the deterministic determination in the subsequent step (c), the synchronization proof value in step (b) mathematically guarantees confirmation by both parties to the record in step (a), the chain structure in step (d) enables tamper detection of the entire record from steps (a) to (c), the TEE in step (f) technically guarantees the independent execution of steps (a) to (e), and the input / output record in step (g) allows the processing of the TEE in step (f) itself to be verified externally. This interdependent chain of seven steps is not merely a collection of any individual elements (TEE, hash chain, integer arithmetic, etc.), but constitutes an integrated technical improvement that simultaneously realizes objectivity, reproducibility, and verifiability of the transaction process.

[0177] Furthermore, the incident described in the background technology section above, in which approximately $128 million in assets were lost in 2025 due to an attack exploiting rounding errors in IEEE 754 floating-point arithmetic in the decentralized financial protocol Balancer, demonstrates the existence of a long-standing unresolved problem in transaction decision-making that relies on floating-point arithmetic. For many years, engineers in this field have recognized the non-deterministic nature of floating-point arithmetic, yet have been unable to present a solution that combines integer-only decision-making with independent arithmetic entities within a protected execution environment in the context of transaction process management. This invention provides a concrete technical solution to this long-standing problem, and its technical effectiveness cannot be achieved by any simple combination of prior art. [Industrial applicability]

[0178] This invention is applicable to all industrial sectors where it is necessary to manage the quality of deliverables based on specifications, and is not limited to any particular industry or region. By using comparison processing with only integer arithmetic and abstract variables (V_raw, V_int, μ, Δ, T), it achieves industry-independent, universal transaction process management.

[0179] Specifically, the effects of this invention are particularly evident in transactions with mega AI companies, such as GPU computing resource procurement, AI model licensing, training data procurement, and API SLA compliance; multi-party and multi-continental quality control in global supply chains; pandemic response (vaccine supply chains, government procurement of supplies, crisis response for international organizations); and the preservation of transaction records and dispute resolution under force majeure (war, riots, natural disasters, economic sanctions).

[0180] Furthermore, it operates over any transport, including terrestrial internet, 5G / millimeter wave, LEO satellite communications (Starlink, OneWeb, Kuiper, etc.), submarine cables, and mobile cellular communications, and can be used from any terminal device such as PCs, smartphones, tablets, and IoT sensor devices, making it applicable to any trading environment worldwide, including areas with weak communication infrastructure.

[0181] Furthermore, the present invention is also applicable to protecting user-to-user transactions on internet platforms. In marketplaces, freelance platforms, crowdfunding, and online auctions, the processing unit (S1) performs transaction determination and record management as an independent computing entity from the platform operator, thereby eliminating the "single point of confidence" problem caused by the centralized data management of the platform operator. In addition, cryptographic guarantees of bilateral participation by the transaction parties through synchronous proof values ​​provide technical countermeasures against internet-based criminal fraud such as identity theft, special fraud, impersonation of government agencies, and romance scams.

[0182] Furthermore, the present invention is applicable to the detection of deepfakes (AI-generated fake images, videos, and audio), the verification of the origin of fake news, and the proof of content authenticity. In the proof of authenticity of articles and videos by news organizations, the proof of origin of posted content on social media, the assurance of authenticity of product images in e-commerce, and the detection of tampering with digital content as legal evidence, the content authenticity proof of the present invention provides a permanent guarantee of authenticity based on the mathematical irreversibility of cryptographic computation, even in environments where the reliability of pattern-matching authentication (including biometric authentication) is declining due to the evolution of artificial intelligence technology.

[0183] Furthermore, the present invention is useful in complying with the laws and regulations of various countries and regions. In Japan, the Act on Fair Trade Practices (amendment to the former Subcontracting Act, effective in 2026) strengthens the obligation to digitize and maintain records of subcontracting transactions, and the objective transaction process record using append-only records of the present invention structurally satisfies this obligation. The European Union's AI Regulation (EU AI Act) mandates transparency and audit trails for AI systems, and the recording of cryptographic summaries and version information of the inference model of the present invention in append-only records (corresponding to claim 19) technically realizes this obligation. The European Union's Due Diligence Directive for Business Sustainability (CSDDD) mandates transparency throughout the entire supply chain, and the generation of integrated verification values ​​in multi-party transactions of the present invention (corresponding to claim 17) enables comprehensive verification of the integrity of the entire supply chain. The US Presidential Order on AI Security requires AI security and transparency, and the inference model audit trail of the present invention addresses this requirement. In addition, in relation to the data integrity principle of the European Union's General Data Protection Regulation (GDPR), the chain structure of append-only records of the present invention cryptographically guarantees the integrity of the records. (Quantitative demonstration of the technical improvements of the present invention) Each step of the present invention provides technical effects that are fundamentally impossible to achieve through human mental activity for the following reasons, and therefore does not constitute an abstract idea. Firstly, the generation of the synchronous proof value H(H(d1)||H(d2)) in step (b) is a concatenation operation of cryptographic hash functions, which cannot be generated without input from both terminal devices. This operation requires bitwise operations on a specific processor and is cryptographically impossible to perform in a human mind. Secondly, the use of integer arithmetic only in steps (a) through (c) eliminates floating-point arithmetic, thereby guaranteeing determinism that produces the same output for the same input on any processor architecture. This determinism is a technical improvement that cannot be achieved with the floating-point arithmetic used in the prior art (US Patent Nos. 10,637,967 and 10,326,862). Thirdly, the computation within the protected execution environment (TEE) in step (f) requires hardware-level isolation and cannot be achieved in software alone. The generation of the proof value is a technical effect inherent to the hardware capabilities of the TEE. Fourth, the independent re-verification in step (g) involves re-executing integer calculations based on all input values ​​recorded in the append-only record, which requires thousands or even tens of thousands of calculations when there are many transaction intervals. This far exceeds the processing capacity of human mental activity. [Explanation of Symbols]

[0184] 100 Transaction Process Management System T1 First terminal device (client side) T2 Second terminal device (contracted side) S1 Processing unit (an independent computing entity equipped with an execution environment that has a protection mechanism) N Network G Third-party inspection terminal device V_raw Raw data (real values) V_int integer value (V_raw / μ) μ is a predetermined minimum unit (set based on mutual agreement). Δ The absolute value of the difference (|V_int_actual - V_int|) T is a predetermined integer threshold. K1 Data Verification Freeze (DVF) K2 Sync Proof Generation (SPG) K3 Auto State Transition (AST) K4 Tamper-Detection Chain (TDC) K5 Emergency Read Access (ERA) based on technical requirements K6 Arbiter Independence and Proof of Arbitration (ABI) TEE (Trusted Execution Environment) D1, D2, ..., Dm Evaluation dimensions (quality, delivery time, cost, compliance, etc.)

Claims

1. A transaction process management method that is performed in a system comprising at least one first terminal device (T1), at least one second terminal device (T2), and a processing device (S1) having a processor and a memory device that does not belong to the administrator of any of the terminal devices, wherein each of the terminal devices and the processing device are connected via a network (N), (a) The processing device sets a predetermined minimum unit based on the agreed data received from both the at least one first terminal device and the at least one second terminal device, records the predetermined minimum unit in the storage device, divides each data received from the at least one first terminal device by the predetermined minimum unit to convert it into an integer value, generates a cryptographic summary value for each integer value, and irreversibly fixes the record in the storage device; (b) The processing device calculates a cryptographic summary value of a value obtained by concatenating the cryptographic summary value of each acknowledgment data received from the at least one second terminal device with the corresponding cryptographic summary value from step (a), thereby generating a synchronization proof value that cannot be generated without input from both corresponding terminal devices. (c) The processing device divides the deliverable data received from the at least one second terminal device by the predetermined minimum unit to convert it into an integer value, performs a comparison process using only integer arithmetic to output a determination result of whether it matches or does not match, automatically fixes the delivery record for the transaction section if it matches, and automatically fixes the consultation record for the transaction section if it does not match. (d) The processing device sequentially concatenates the results of steps (a) to (c) as append-only records that include the cryptographic summary value of the immediately preceding record in the record, and records them in the storage device. (e) The processing device, when a predetermined technical condition recorded in advance in the record for appending is satisfied, releases at least a portion of the record in read-only mode, and adds data identifying the fact of the release operation and the satisfied technical condition to the concatenation of the record for appending; (f) The processing unit executes all the processes of steps (a) through (e) and (g) as an independent processing unit from the administrator of any terminal device constituting the system, within an execution environment having a protection mechanism that prevents the administrator of any terminal device from accessing the processing unit's calculation process and storage contents, and generates an attribution value indicating the integrity of the program code executed in the execution environment and provides it to at least one of the terminal devices. (g) The processing device records the input and output values ​​of each integer operation in steps (a) to (c) in the record dedicated to appending, thereby enabling each of the terminal devices to independently re-execute the integer operation based on the input value and verify that it matches the output value. A transaction process management method that includes this.

2. The transaction process management method according to claim 1, wherein the cryptographic summary value is generated by the SHA-256 hash function.

3. The transaction process management method according to claim 1, wherein the comparison process using only integer arithmetic in step (c) is a process of calculating the absolute value of the difference between an integer value obtained by dividing the deliverable data by the predetermined minimum unit and an integer value fixed in step (a), determining that there is a match if the absolute value is less than or equal to a predetermined integer threshold, and determining that there is a mismatch if it exceeds that threshold.

4. The transaction process management method according to claim 1, wherein the synchronization proof value in step (b) is the SHA-256 hash value of the value obtained by concatenating the cryptographic summary value of step (a) and the cryptographic summary value of the acknowledgment data from the at least one second terminal device in that order.

5. The transaction process management method according to claim 1, further comprising the step of the processing unit sequentially recalculating the cryptographic summary values ​​of all records from the beginning in order to verify the integrity of the concatenation of append-only records in step (d), and confirming that they match the recorded cryptographic summary values, wherein each of the terminal devices can independently perform the verification using the input and output values ​​recorded in step (g).

6. The transaction process management method according to claim 1, wherein each record in step (d) includes a Coordinated Universal Time (UTC) timestamp indicating the time of creation of the record.

7. The transaction process management method according to claim 1, wherein the series of processes in steps (a) to (e) are linked by an identifier uniquely assigned to each transaction unit, and the identifier is recorded in each record.

8. The transaction process management method according to claim 1, further comprising the step between step (a) and step (c) of the processing device converting the interim report data received from the at least one second terminal device into the integer value and recording the comparison result with a predetermined warning threshold in the record for appending.

9. The transaction process management method according to claim 1, wherein the output data in step (c) includes multiple inspection items, and the processing device individually performs the comparison process using only integer arithmetic for each inspection item, and determines that the transaction interval is a match only when all inspection items match.

10. The transaction process management method according to claim 1, further comprising the step of the processing device transmitting the determination result in step (c) to each of the first terminal device and the second terminal device relating to the transaction interval.

11. The transaction process management method according to claim 1, wherein the at least one second terminal device functions as a first terminal device in another transaction interval, and the processing device performs steps (a) to (g) for each of a plurality of linked transaction intervals.

12. The transaction process management method according to claim 1, wherein the system further comprises a third terminal device different from both the first terminal device and the second terminal device, the output data in step (c) is received from the third terminal device, and the protection mechanism in step (f) also prevents the administrator of the third terminal device from accessing the processing device.

13. The transaction process management method according to claim 1, wherein in step (a), in addition to the data received from the first terminal device, a cryptographic summary value of the specification data of raw materials supplied from the first terminal device to the second terminal device is additionally generated and fixed.

14. The transaction process management method according to claim 1, wherein step (c) is performed for each of the multiple deliverable data separated by process, and the determination result for each process is individually recorded in the record dedicated to appending.

15. The transaction process management method according to claim 1, wherein the deliverable data in step (c) is received from a third-party inspection terminal device that does not belong to the administrator of either the first terminal device or the second terminal device, and the identification information of the third-party inspection terminal device is recorded in the record for appending.

16. The transaction process management method according to claim 11, further comprising the step of the processing device notifying terminal devices relating to other relevant transaction segments of the fact of the mismatch if a mismatch is determined in step (c).

17. The transaction process management method according to claim 11, further comprising the step of the processing device generating an integrated verification value by concatenating the final cryptographic summary values ​​of each of the append-only records corresponding to a plurality of transaction intervals.

18. The transaction process management method according to claim 1, wherein at least one second terminal device is connected to a sensor device for measuring physical quantities, and the output data is automatically acquired from the sensor device.

19. The transaction process management method according to claim 1, wherein when an inference model is used to generate integer values ​​to be used in the comparison process using only integer arithmetic in step (c), the processing unit converts the output of the inference model into an integer value by dividing it by the predetermined minimum unit and then uses it in the comparison process, records the cryptographic summary value and version information of the inference model in the dedicated append record, and includes the calculation by the inference model and the conversion of its output to an integer value as targets for independent re-execution in step (g).

20. The transaction process management method according to claim 1, wherein each record in step (d) is accompanied by the electronic signature of the terminal device or processing device that was involved in the generation of the record.

21. The transaction process management method according to claim 1, wherein the processing apparatus is composed of a plurality of geographically dispersed processing nodes, each processing node executes the same program code within an execution environment having the protection mechanism, all processing nodes are verified to be executing the same process by matching the proof of integrity values ​​of each processing node, and each processing node maintains an identical copy of the append-only record.

22. The transaction process management method according to claim 1, wherein the scope of read-only access in step (e) is limited to recording the transaction interval relating to the technical conditions.

23. The transaction process management method according to claim 1, wherein the predetermined technical condition in step (e) is the continuation of a state in which no authenticated access is performed from any terminal device included in the system within a predetermined period of time.

24. The transaction process management method according to claim 1, wherein the predetermined technical condition in step (e) is the establishment of threshold authentication using a predetermined number or more emergency keys from a plurality of emergency keys that have been registered in advance.

25. The transaction process management method according to claim 1, wherein the execution environment having the protection mechanism in step (f) is a hardware-isolated trusted execution environment.

26. The transaction process management method according to claim 1, wherein the proof value generated in step (f) is recorded in the record dedicated to appending.

27. The transaction process management method according to claim 1, wherein the predetermined minimum unit in step (a) is set only when the agreement data received from both the at least one first terminal device and the at least one second terminal device match, and the cryptographic summary value of each of the agreement data is recorded in the append-only record.

28. The transaction process management method according to claim 1, wherein each data in step (a) includes data relating to a plurality of evaluation dimensions that are different from each other, the processing device sets the predetermined minimum unit individually for each evaluation dimension, performs the comparison process in step (c) independently for each evaluation dimension, and records the determination result for each evaluation dimension individually in the record for appending.

29. The transaction process management method according to claim 28, wherein, when predetermined dependencies between the plurality of evaluation dimensions are recorded in the append-only record, the processing device automatically controls the execution or suspension of the comparison process for the second evaluation dimension based on the determination result of the first evaluation dimension, and records the fact of said control in the append-only record.

30. The transaction process management method according to claim 28, wherein the processing device aggregates the determination results of each of the plurality of evaluation dimensions, records the multidimensional integrated determination result of the transaction interval as a match only when there is a match for all evaluation dimensions, and records the multidimensional integrated determination result of the transaction interval as a mismatch in the dedicated append record when there is a mismatch for any evaluation dimension.

31. A transaction process management system comprising: at least one first terminal device (T1); at least one second terminal device (T2); a processing unit (S1) that does not belong to the administrator of either terminal device and has an execution environment having a processor, a storage device for storing records for appending only, and a protection mechanism to prevent access to the calculation process and stored contents by any administrator of the terminal device; and a network (N) connecting each terminal device and the processing unit, wherein the processor of the processing unit sets and records a predetermined minimum unit based on agreement data received from both the at least one first terminal device and the at least one second terminal device within the execution environment having the protection mechanism, divides each data received from the at least one first terminal device by the predetermined minimum unit to convert it to an integer value, generates a cryptographic summary value for each integer value and irreversibly fixes the recording in the storage device, and calculates a cryptographic summary value of a value obtained by concatenating the cryptographic summary value of each acknowledgment data received from the at least one second terminal device with the corresponding cryptographic summary value, thereby enabling both terminal devices to A transaction process management system is configured to generate a synchronization proof value that cannot be generated without input, convert the deliverable data received from at least one second terminal device into an integer value by dividing it by the predetermined minimum unit, perform a comparison process using only integer arithmetic and output a determination result of match or mismatch, automatically fix the delivery record for the transaction section if there is a match and automatically fix the consultation record for the transaction section if there is a mismatch, sequentially concatenate each of the above results as an append-only record that includes the cryptographic summary value of the previous record in the record and record it in the storage device, release at least a portion of the record in read-only mode when predetermined technical conditions previously recorded in the append-only record are satisfied, append data that identifies the fact of the release operation and the satisfied technical conditions to the concatenation of the append-only records, generate an attributable proof value that shows the integrity of the program code executed in the execution environment and provide it to at least one of the terminal devices, and record the input and output values ​​of each of the integer arithmetic in the append-only record so that each of the terminal devices can independently re-execute and verify the integer arithmetic.

32. In an execution environment in which the processor of a processing unit (S1) is connected via a network (N) to at least one first terminal device (T1) and at least one second terminal device (T2), and does not belong to the administrator of either of the terminal devices, access to the calculation process and stored contents by any administrator of the terminal devices is prevented by a protection mechanism of the processing unit, (a) a predetermined minimum unit is set and recorded based on agreed data received from both the at least one first terminal device and the at least one second terminal device, and then the at least one first (b) a process of converting each data received from the terminal device into an integer value by dividing it by the predetermined minimum unit, generating a cryptographic summary value for each integer value and irreversibly fixing the record; (b) a process of generating a synchronization proof value that cannot be generated without input from both corresponding terminal devices by calculating a cryptographic summary value of the concatenated value of the cryptographic summary value of each acknowledgment data received from the at least one second terminal device and the corresponding cryptographic summary value of the process (a); (c) a process of dividing the output data received from the at least one second terminal device by the predetermined minimum unit and staging (d) a process that converts the data into a numerical value, performs a comparison process using only integer arithmetic and outputs a result of determining whether the data matches or does not match, automatically fixing the delivery record for the transaction interval if there is a match and automatically fixing the negotiation record for the transaction interval if there is a mismatch; (e) a process that sequentially concatenates the results of each of the above processes (a) to (c) as an append-only record that includes the cryptographic summary value of the immediately preceding record in the append-only record and records (e) if predetermined technical conditions recorded in advance in the append-only record are met, at least a portion of the record is released in read-only mode, and the fact of the release operation and the met technical conditions A transaction process management program that causes (f) a process to append data specifying technical conditions to the concatenation of the append-only records, (g) a process to generate an attribution value indicating the completeness of the program code executed in the execution environment and provide it to at least one of the terminal devices, and (g) a process to record the input and output values ​​of each integer operation in processes (a) to (c) in the append-only records, thereby enabling each of the terminal devices to independently re-execute and verify the integer operation. The program is configured to execute these processes as an independent computation entity, separate from the administrator of any terminal device.

33. A transaction process management method to be performed in a system comprising at least one first terminal device (T1), at least one second terminal device (T2), and a processing device (S1) having a processor and a storage device and not belonging to the administrator of any of the terminal devices, wherein each terminal device and the processing device are connected via a network (N), the method comprising: (a) the processing device setting a predetermined minimum unit based on agreement data received from both the at least one first terminal device and the at least one second terminal device, and recording the predetermined minimum unit in the storage device; (b) The processing device calculates a cryptographic summary value of a value obtained by concatenating the cryptographic summary value of each acknowledgment data received from the at least one first terminal device with the corresponding cryptographic summary value from step (a), thereby generating synchronization proof values ​​that cannot be generated without input from both corresponding terminal devices; (c) The processing device calculates the cryptographic summary value of a value obtained by concatenating the cryptographic summary value of each acknowledgment data received from the at least one second terminal device with the corresponding cryptographic summary value from step (a), thereby generating synchronization proof values ​​that cannot be generated without input from both corresponding terminal devices; (d) The processing device converts the output data received from one second terminal device into an integer value by dividing it by the predetermined minimum unit, performs a comparison process using only integer arithmetic, outputs a determination result of whether it matches or not, automatically fixes the delivery record for the transaction section if it matches, and automatically fixes the negotiation record for the transaction section if it does not match; (d) The processing device sequentially concatenates each result of steps (a) to (c) as an append-only record that includes the cryptographic summary value of the immediately preceding record in the record and records it in the storage device; (e) The processing device executes all the processes of steps (a) to (d) and step (f) as an independent processing entity from the administrator of any terminal device in an execution environment having a protection mechanism that prevents the administrator of any terminal device constituting the system from accessing the processing process and storage contents of the processing device, generates an attribution proof value that shows the integrity of the program code executed in the execution environment and provides it to at least one of the terminal devices; (f) The processing device records the input and output values ​​of each integer arithmetic in steps (a) to (c) in the append-only record.A transaction process management method comprising the step of enabling each of the terminal devices to independently re-execute the integer arithmetic based on the input value and verify its agreement with the output value.

34. A transaction process management system comprising: at least one first terminal device (T1); at least one second terminal device (T2); a processing unit (S1) that does not belong to the administrator of either terminal device and has an execution environment having a processor, a storage device for storing records for appending only, and a protection mechanism that prevents access to the calculation process and stored contents by any administrator of the terminal device; and a network (N) connecting each terminal device and the processing unit, wherein the processor of the processing unit sets and records a predetermined minimum unit based on agreement data received from both the at least one first terminal device and the at least one second terminal device within the execution environment having the protection mechanism, divides each data received from the at least one first terminal device by the predetermined minimum unit to convert it into an integer value, generates a cryptographic summary value for each integer value and irreversibly fixes the record in the storage device; and the cryptographic summary of each acknowledgment data received from the at least one second terminal device A transaction process management system is configured to generate a synchronous proof value that cannot be generated without input from both corresponding terminal devices by calculating a cryptographic summary value of a value obtained by concatenating a summary value and the corresponding cryptographic summary value; to convert the deliverable data received from at least one second terminal device into an integer value by dividing it by the predetermined minimum unit; to perform a comparison process using only integer arithmetic and output a determination result of whether it matches or not; to automatically fix the delivery record for the transaction interval if it matches and to automatically fix the consultation record for the transaction interval if it does not match; to sequentially concatenate each of the above results as an append-only record that includes the cryptographic summary value of the immediately preceding record in the record and record it in the storage device; to generate an attributable proof value that shows the integrity of the program code executed in the execution environment and provide it to at least one of the terminal devices; and to enable each of the terminal devices to independently re-execute and verify the integer arithmetic by recording the input and output values ​​of each integer arithmetic in the append-only record.

35. A method for verifying a transaction process performed by a terminal device used in the transaction process management method described in claim 1, comprising: (a) the terminal device obtaining from the processing device a record specifically for appending, which includes the input and output values ​​of the integer operation recorded in step (g); (b) the terminal device re-executing the integer operation based on the input values ​​of the integer operation recorded in the obtained record specifically for appending, and verifying that the result of the re-execution matches the output values ​​recorded in the record specifically for appending; and (c) the terminal device sequentially recalculating the cryptographic summary values ​​contained in each record of the record specifically for appending, starting from the beginning of the concatenation in step (d), and confirming that they match the recorded cryptographic summary values. A transaction process verification method comprising: (d) verifying the integrity of the chain of append-only records; (f) verifying the attribution value received from the processing device in step (f) to confirm that the processing device executed predetermined program code within the execution environment having the protection mechanism; and (e) verifying the predetermined minimum unit setting data based on the mutual agreement data in step (a) and the synchronization proof value in step (b) recorded in the append-only record, to confirm that the minimum unit was set based on inputs from both terminal devices and that the synchronization proof value could not be generated without inputs from both.

Images