Search

KR-20260064664-A - Security Erasure and Hash Tree Proof Mediation System Based on Textless Input Control and Stochastic State Machine for High-Density Document Environment

KR20260064664AKR 20260064664 AKR20260064664 AKR 20260064664AKR-20260064664-A

Abstract

The present invention is a textless security erasure and hash tree post-verification system optimized for document verification environments. The input control module blocks text inflow through L7 schema verification and accepts only binary judgment (YES/NO) packets for specific coordinates. The state machine-based data management unit performs flawless security erasure by transitioning to storage overwriting or symmetric key destruction based on the threat index. The proof issuance unit guarantees post-verification even without text records by configuring binary judgment events into a hash tree. Furthermore, through a viewing-sensitive payment unit, it systematically defends against payment evasion after information acquisition by synchronizing and settling escrow payments the moment a customer requests a decryption key for detailed coordinate verification, and maximizes resource allocation efficiency by combining page-based non-linear billing logic.

Inventors

  • 신채은

Assignees

  • 신채은

Dates

Publication Date
20260507
Application Date
20260405

Claims (4)

  1. A textless control and state machine-based document verification mediation system comprising: an input control module that controls a textless input path by analyzing a received packet through a data verifier of the application layer (L7) during a verification session with a verifier terminal, dropping the communication if text string data is detected, and accepting only a whitelist structure consisting of a target page identifier, spatial coordinates, and a binary judgment status value (YES or NO); a state machine-based multi-layer data management unit that monitors whether a threat event occurs within the session, transitions to a security erasure mode that performs a physical overwrite on the storage where the target document is stored upon normal termination, and branches to a logical security revocation mode that blocks access to the storage and destroys the encryption symmetric key in memory upon detection of an abnormal threat; and a proof issuance unit that issues a verification receipt object (VRO) proving the binary judgment fact for a specific coordinate without storing the text plaintext, including the root value of a hash tree constructed by unidirectional hashing the judgment events collected through the textless input path.
  2. A document verification brokerage system according to claim 1, further comprising a viewing-sensitive payment unit that first provides statistical data including the total number of error (NO) judgments among the binary judgment status values to the client terminal, and immediately upon the occurrence of a decryption key request event for viewing result data including the error detail page and coordinates at the client terminal, automatically transfers a preset percentage of the escrow amount to the verifier side in synchronization with the logical moment when the decryption key is returned.
  3. A document verification brokerage system according to claim 1, further comprising a non-linear matching engine that analyzes multidimensional parameters including data error suspicion density and behavioral heuristics collected from the verifier terminal based on the page unit of the target document, and dynamically calculates the processing unit cost per page by converting linear weights into non-linear exponential function weights when a preset threshold is exceeded, wherein the behavioral heuristics are calculated by including the mouse dwell time, scroll trajectory, and click frequency of the verifier terminal.
  4. A document verification intermediary system according to claim 1, wherein the verification receipt object (VRO) issued by the evidence issuing unit is calculated through a hash tree formation logic consisting of a combination of leaf nodes that have been unidirectionally hashed, including a timestamp, page identifier, coordinate vector, binary judgment status value, and session identifier, while excluding natural language text data regarding the error (NO) and normal (YES) judgments, thereby enabling post-verification of whether the judgment for a specific coordinate has been tampered with even after the permanent erasure of text data.

Description

Textless Input Control and Stochastic State Machine-Based Security Erasure and Hash Tree Proof Mediation System for High-Density Document Environment The present invention relates to a data processing system and network security control technology for verifying the integrity of a large volume of complex documents. More specifically, the invention relates to a textless input control protocol that inspects incoming packets at the application layer (L7) to block string inflow and accepts only binary state values, a probabilistic state machine architecture that branches physical overwriting of storage and logical destruction of symmetric keys according to threat events, and a hash tree object issuance technology that secures post-hoc proof through a time series of binary determination events without plaintext. In verification systems for high-density complex documents (including human-written and algorithm-generated) requiring security, conventional technology has the following limitations in information security and system resource management. First, there is the security vulnerability caused by free-form text input. Providing a text input interface to validators can serve as a channel for the leakage of original data and expose them to malicious script injection attacks (such as XSS). Second, there is the limitation of batch data deletion logic. A method that simply deletes storage data upon session termination results in the loss of reproducibility, which is necessary to retrospectively prove the legitimacy of the verification action. Third, there is the vulnerability of deterministic responses. In the event of complex threats such as network hijacking, even the physical I/O latency required to overwrite data can become an opportunity for system takeover, yet there is a lack of dynamic bypass mechanisms to address this. Fourth, there are issues of traffic inefficiency and information asymmetry resulting from the uniform provision of data regardless of document processing load. Hereinafter, the architecture and data flow of the system are described in detail so that those skilled in the art to which the present invention pertains can implement it. Specific embodiments are included to clarify the operation of the system. (1) Non-linear page-based billing engine The matching engine divides the page units of the document into the minimum logical units for billing. $$Price_{page} = Base\_Rate + \lambda \cdot \exp(D_{anomaly} \oplus L_{heuristic})$$ (Here, $Base\_Rate$ is the base price of a page, $D_{anomaly}$ is the density of suspected anomaly tokens, $L_{heuristic}$ is behavioral parameters such as mouse dwell time and scroll trajectory collected from the validator terminal, and $\exp$ is a non-linear exponential weight.) (2) L7 schema verification-based 'YES/NO binary' textless protocol The input control module controls backend communication (L7) between the validator terminal and the platform. Incoming payloads pass through a JSON schema validator, and the platform accepts only whitelisted data (page ID, spatial coordinates, enumerated binary state values). If a packet containing variable-length string data is received, the network stack permanently drops the packet. (3) Information asymmetric control and viewing-sensitive escrow interlock Upon completion of verification, the platform initially provides the client terminal with statistics on the 'total number of error (NO) judgments'. When the client attempts to 'view detailed results' to check the detailed page and coordinates of a specific error, the smart contract module transmits the decryption key to the client and simultaneously processes an automatic transfer (Settlement Trigger) of a pre-configured percentage of the funds deposited in the escrow wallet to the validator's account. (4) Probabilistic state machine binary security erasure and hash tree proof Upon normal session termination, the original is destroyed by executing a random-based physical overwrite (Secure Erasure) on the storage layer. If the threat variable threshold is exceeded, storage I/O access is hardware-blocked, and a bypass transition to a logical discard mode is performed to destroy the symmetric key on the Key Management System (KMS). The proof issuance unit hashes the events to form the following Leaf nodes and issues a Validated Receipt Object (VRO). $$Leaf_i = H_{crypto}(\Delta t_i \oplus Page\_ID \oplus V_{coord} \oplus [YES/NO] \oplus ID_{session})$$ [First Embodiment: Textless Data Communication and Billing Control Flow] The client uploads a 500-page technical document on algorithm generation to the system. The matching engine detects that the technical contradiction word (error suspicion density, $D_{anomaly}$) on page 45 exceeds a threshold, automatically increases the unit price of that page from the base 1,000 won to 5,000 won by applying exponential weighting, and then determines the total escrow amount. The validator determines an error (NO) for a specific coordinate