Language data translation method and system based on visual process configuration

Abstract

The invention discloses a language data translation method and system based on visual process configuration, which can realize safe key distribution based on quantum key negotiation and based on an inaccurate measurement principle and an unclonable theorem of quantum mechanics, and radically solve the problem of key leakage. Chaotic mapping has extreme sensitivity and pseudo-randomness to an initial value and a control parameter, and a sequence generated by the chaotic mapping can be used for encryption parameter optimization, so that the anti-cracking capability is improved; and the visual process configuration can meet the customized requirements of the user.

Description

Technical Field

[0001] This invention relates to the fields of natural language processing and data security technology, and more specifically, to a language data translation and system based on a visual process configuration. Background Technology

[0002] With the acceleration of globalization, language data translation (such as document translation, real-time dialogue translation, and business text conversion) is increasingly widely used in government affairs, finance, healthcare, and other fields. Traditional language data translation methods mainly suffer from the following shortcomings:

[0003] Insufficient configuration flexibility: Traditional translation tools mostly have fixed processes, and users cannot customize the translation process according to actual scenarios (such as different language combinations, format requirements, translation accuracy), resulting in poor adaptability.

[0004] Sensitive data security risks are prominent: During the translation process, core sensitive language data (such as classified documents and private conversations) need to go through multiple stages such as translation engines, transmission channels, and storage media. Traditional methods lack end-to-end encryption protection and are easily eavesdropped, tampered with, or leaked during transmission or processing.

[0005] Static encryption mechanism: Existing translation tools with encryption functions mostly use fixed keys and a single encryption algorithm (such as static AES key). Key distribution relies on traditional channels (which are easily eavesdropped on), and there is no dynamic update mechanism for encryption parameters. Attackers can deduce encryption patterns through historical ciphertexts, resulting in weak resistance to cracking.

[0006] Poor coordination between translation and encryption: In existing technologies, translation and encryption functions are independent of each other, and encryption is not differentiated according to the sensitivity level of language data. This results in insufficient protection of core data or redundant encryption of non-sensitive data, and a weak ability to balance security and translation efficiency.

[0007] Lack of identity authentication mechanism: The receiving end cannot verify the legitimacy of the sending end's identity, making it vulnerable to forged translation attacks, which can lead to the spread of erroneous information or the illegal acquisition of sensitive data. Summary of the Invention

[0008] The purpose of this invention is to provide a language data translation and system based on a visual process configuration, in order to solve the above-mentioned problems existing in the prior art.

[0009] In a first aspect, embodiments of the present invention provide a language data translation method based on a visual process configuration, including:

[0010] Step S1: Visual translation workflow configuration stage, including obtaining user-inputted language data to be translated;

[0011] Step S2: Language Data Sensitivity Level Classification Stage: The input language data to be translated is automatically identified for sensitivity level. Based on the preset thresholds set by the user, the sensitivity level is classified into three levels: core sensitive language data, general sensitive language data, and non-sensitive language data. Among them, core sensitive language data includes classified documents, private conversations, and core business documents; general sensitive language data includes routine business texts and descriptions of basic user information; and non-sensitive language data includes publicly available promotional materials and general information texts.

[0012] Step S3: Dynamic encryption key generation stage: Based on the BB84 quantum key negotiation protocol, a secure channel is established between the two communicating parties to generate an initial shared key SK. SK is then hashed using SHA-256 to obtain a hash value H(SK). H(SK) is split into multiple sets of parameter fragments, which are used as the initial values ​​and control parameters for the Logistic chaotic mapping and the Lorenz chaotic mapping, respectively. A two-dimensional chaotic mapping system is constructed to generate a chaotic sequence set C={C1,C2}. Based on the sensitivity level classification results, the corresponding level translation encryption key K={K1,K2,K3} is extracted from C to generate the corresponding level.

[0013] Where C1 is a chaotic sequence generated by the Logistic chaotic mapping; C2 is a chaotic sequence generated by the Lorenz chaotic mapping; K1 represents the encryption key for core sensitive data; K2 represents the encryption key for general sensitive data; and K3 represents the encryption key for non-sensitive data.

[0014] Step S4: Translation Instruction Chaotic Obfuscation Stage: The translation instruction set generated by the visualization process analysis is converted into an intermediate instruction sequence. For intermediate instructions involving sensitive language data processing, a chaotic sequence C1 is used for double obfuscation. Specifically, the instruction line is randomly rearranged based on the numerical sorting result of C1, and then the instruction byte is XORed with the corresponding position value of C1 modulo 256 to obtain the obfuscated intermediate instruction set, in order to prevent the translation instructions from being tampered with or reverse parsed.

[0015] Step S5: Language Data Translation and Segment Encryption Stage

[0016] During the language data translation and segmented encryption stage, the operations performed on core sensitive language data include: performing translation operations based on the obfuscated intermediate instruction set, and then using a dual mechanism of "chaotic sequence diffusion + AES-256 encryption" to obtain the translation. First, C2 is used to perform modulo 256 addition diffusion processing on the translated bytes, and then K1 is used for AES-256 encryption.

[0017] During the language data translation and segmented encryption stage, the operations performed on generally sensitive language data include: after performing the translation operation, using the SM4 encryption mechanism and encrypting the translation with K2;

[0018] In the language data translation and segmented encryption stage, operations on non-sensitive language data include: after performing the translation operation, lightweight chaotic XOR encryption is adopted, and the translated text is XOR processed with K3;

[0019] Step S6: Secure delivery and dynamic update stage: The segmented encrypted translated text is concatenated in the format of "sensitive level identifier + ciphertext length + ciphertext content", and identity field verification is performed based on the hardware fingerprint and RSA-2048 signature to obtain the identity authentication field, and then the identity authentication field is transmitted to the receiving end; after the receiving end verifies the identity legality, the translated text is restored according to the corresponding key and decryption mechanism; after every M translation tasks are completed, the initial shared key SK is updated through the quantum key negotiation protocol, and steps S3 - step S5 are repeated to regenerate the chaotic sequence and encryption key.

[0020] Optionally, the parameter segment division method in step S3 is: the first 64 bits of the hash value H(SK) are converted into the initial value x0 of the Logistic chaotic map, where 0 < x0 < 1; the middle 64 bits of the hash value H(SK) are converted into the control parameter μ, where 3.57 < μ < 4; the 129 - 256 bits of the hash value H(SK) are split and converted into the initial value (x1, y1, z1) and control parameters (σ, ρ, β) of the Lorenz chaotic map.

[0021] Optionally, the encryption key generation rule in step S3 is: K1 is generated as a 256-bit key through SHA-384 hashing after XOR operation of C1 and C2; K2 is generated as a 128-bit key through SM3 hashing after concatenating C1; K3 intercepts the first 64 bits of C2 as the key.

[0022] Optionally, the translation instruction set parsing process in step S4 is: perform syntax analysis on the operator nodes and association relationships of the visualization process to generate a structured translation instruction set including language conversion rules, semantic mapping logic, and format optimization instructions to ensure the accuracy of instruction execution.

[0023] Optionally, the identity authentication field in step S6 includes the sender's hardware fingerprint, user public key, and public key signature information, and the receiving end verifies the signature legality through the user public key to confirm the sender's identity.

[0024] Optionally, the preset thresholds include the sensitive level threshold for language data sensitive level division, the eavesdropping detection bit error rate threshold for quantum key negotiation, and the translation task threshold for key dynamic update.

[0025] Optionally, the sensitive level threshold for language data sensitive level division ≥ 3.

[0026] Optionally, the eavesdropping detection bit error rate threshold for quantum key negotiation is 0.05 - 0.1.

[0027] Optionally, the threshold for translation tasks with dynamically updated keys is between 5 and 50.

[0028] Secondly, embodiments of the present invention also provide a language data translation system based on visual process configuration, including a visual process configuration module, a sensitivity level classification module, a dynamic key generation module, an instruction obfuscation module, a translation and encryption module, and a secure delivery and update module, wherein each module is used to execute the corresponding steps of any of the methods described above.

[0029] Compared with the prior art, the present invention achieves the following beneficial effects:

[0030] Significantly improved configuration flexibility: Through visual drag-and-drop configuration and template reuse, users can customize translation processes according to different scenarios, solving the problems of fixed processes and poor adaptability of traditional translation tools, and adapting to the needs of multiple fields such as government affairs, finance, and healthcare;

[0031] Strong protection capabilities for sensitive data: It integrates quantum key negotiation and two-dimensional chaotic mapping to achieve full-link protection of "key secure distribution → dynamic parameter generation → differentiated encryption". Core sensitive data is double-encrypted, and it has outstanding resistance to quantum attacks, brute-force attacks and statistical attacks, fundamentally solving the risk of leakage of sensitive language data.

[0032] Excellent encryption and translation synergy: Differentiated encryption is achieved based on sensitivity level classification, with high-strength protection for core data and lightweight encryption for non-sensitive data, balancing translation efficiency and security, and avoiding encryption redundancy or insufficient protection;

[0033] Complete security closed loop: Integrated hardware fingerprint and RSA features include a signature authentication mechanism to prevent forged translation attacks; the key is dynamically updated according to the number of translations to reduce the risk of key leakage, forming a complete security closed loop of “configuration → translation → encryption → transmission → verification → update”.

[0034] Highly practical and scalable: Supports translation needs in multiple languages, formats, and scenarios. The generated target encrypted text can be securely deployed on different platforms (terminals, servers, cloud platforms), complying with the requirements of the Personal Information Protection Law and the Data Security Law for sensitive data processing. Attached Figure Description

[0035] Figure 1 This is a flowchart of a language data translation method based on visual process configuration provided in an embodiment of the present invention. Detailed Implementation

[0036] The present invention will now be described in detail with reference to the accompanying drawings.

[0037] Example 1

[0038] like Figure 1 As shown, this embodiment of the invention provides a language data translation method based on visual process configuration, including:

[0039] Step S1: The visual translation workflow configuration stage includes obtaining the language data to be translated input by the user. The language data to be translated includes audio and text. In this embodiment, the visual translation workflow configuration stage further includes providing a predefined multi-scenario translation operator library, which includes operators for language identification, word segmentation and parsing, semantic mapping, syntax optimization, format adaptation, and encryption control. This embodiment supports users to combine operator nodes, configure node relationships and operator parameters by dragging and dropping to build a customized visual translation workflow. Operator parameters include the target language, translation accuracy, and sensitivity level threshold. The workflow supports saving, reuse, and secondary editing.

[0040] Step S2: Language data sensitivity level classification stage, specifically: automatically identify the sensitivity level of the input language data to be translated, and classify it into three levels: core sensitive language data, general sensitive language data, and non-sensitive language data, based on the user's preset threshold.

[0041] Among them, the preset thresholds include the sensitivity level threshold for classifying the sensitivity level of language data, the bit error rate threshold for eavesdropping detection in quantum key negotiation, and the translation task threshold for dynamic key updates.

[0042] In this application, the user's preset threshold is mainly reflected in two scenarios, and the specific values ​​can be as follows:

[0043] 1. Sensitivity level threshold for language data sensitivity classification: This threshold is used to mark sensitive data processing nodes or classify the sensitivity level of language data. The patent does not specify a fixed value, but supports users to customize the configuration according to the actual scenario (such as marking text containing keywords such as "confidential" and "privacy" as core sensitive), or set threshold rules in combination with the scope of leakage impact and compliance requirements (such as the Personal Information Protection Law standard). It is a flexible configuration item and there is no uniform fixed value.

[0044] 2. Bit error rate threshold for eavesdropping detection in quantum key negotiation: In the quantum key negotiation stage of dynamic encryption key generation, the preset bit error rate threshold is 0.05-0.1 (i.e. 5%-10%): If the bit error rate detected by sampling is ≤ this threshold, it is determined that there is no eavesdropping, and the initial shared key SK is generated again; if the bit error rate is > this threshold, it is determined that there is eavesdropping, and negotiation is restarted.

[0045] 3. Translation task threshold (M) for key dynamic update: During the secure delivery and dynamic update phase, the threshold M for the number of translation tasks triggered by key update is a configurable value between 5 and 50. Users can flexibly set it according to business needs (e.g., set to 5 for high-sensitivity scenarios and 50 for ordinary scenarios).

[0046] As an optional implementation, the sensitivity level threshold for language data sensitivity classification is ≥3, the bit error rate threshold for eavesdropping detection in quantum key negotiation is 0.05-0.1, and the threshold for translation tasks with dynamic key updates is between 5 and 50.

[0047] In summary, the user-preset thresholds in this application are mostly configurable ranges or custom rules to adapt to the security needs and business characteristics of different scenarios.

[0048] Among them, core sensitive language data includes classified documents, private conversations, and core business documents; general sensitive language data includes routine business texts and descriptions of basic user information; and non-sensitive language data includes publicly available promotional materials and general information texts.

[0049] Step S3: Dynamic encryption key generation stage, specifically: Based on the BB84 quantum key negotiation protocol, a secure channel is established between the two communicating parties to generate an initial shared key SK; SK is hashed using SHA-256 to obtain a hash value H(SK), H(SK) is split into multiple sets of parameter fragments, and these multiple sets of parameter fragments are used as the initial values ​​and control parameters of the Logistic chaotic mapping and the Lorenz chaotic mapping, respectively, to construct a two-dimensional chaotic mapping system and generate a chaotic sequence set C={C1,C2}. Then, according to the sensitivity level classification result, the corresponding level translation encryption key K={K1,K2,K3} is extracted from C to generate the key.

[0050] Wherein, C1 is a chaotic sequence generated by the Logistic chaotic mapping; C2 is a chaotic sequence generated by the Lorenz chaotic mapping; K1 represents the encryption key for core sensitive data; K2 represents the encryption key for general sensitive data; and K3 represents the encryption key for non-sensitive data.

[0051] In S3, the specific operations for establishing a secure channel and generating the initial shared key SK between the two communicating parties, including the translation initiator and receiver, are as follows:

[0052] 1. Quantum state preparation and transmission (executed at the initiating end):

[0053] The translation initiator (such as the user terminal) randomly generates a set of binary sequences as the raw key bit stream (denoted as K_raw);

[0054] Randomly select two sets of orthogonal basis vectors (Z basis: |0>, |1>; X basis: |+>, |->), and encode each original key bit:

[0055] If the bit is 0 and the Z-basis is selected, it is encoded as a horizontally polarized photon (|0>); if the bit is 1 and the Z-basis is selected, it is encoded as a vertically polarized photon (|1>).

[0056] If the bit is 0 and the X basis is selected, the encoding is a 45° polarized photon (|+>); if the bit is 1 and the X basis is selected, the encoding is a 135° polarized photon (|->).

[0057] The encoded single-photon pulse sequence is transmitted to the receiving end via a quantum channel (such as a fiber optic quantum link).

[0058] 2. Quantum state measurement (performed at the receiver)

[0059] The receiver randomly selects either the Z-based or X-based method to measure each received single photon;

[0060] Record the measurement results (0 or 1) and the corresponding measurement basis vectors (Z basis or X basis) to form the measurement bit stream K_meas and the basis vector selection record B_meas.

[0061] 3. Basis vector alignment (classical channel interaction)

[0062] The initiating end and the receiving end exchange basis selection information (B_send of the initiating end and B_meas of the receiving end) through a classic channel (such as an encrypted TCP / IP link).

[0063] Both parties retain the bits that match the basis selection (bits that do not match the basis selection are discarded), forming the filtered original key fragment K_filtered.

[0064] 4. Eavesdropping detection (security verification)

[0065] Both parties randomly selected approximately 20%-30% of the bits in K_filtered for sampling and comparison:

[0066] Calculate the bit error rate (the proportion of inconsistent bits in the sampled bits);

[0067] If the bit error rate is less than or equal to the preset threshold (0.05-0.1, i.e., 5%-10%), it is determined that there is no eavesdropping and the remaining bits are retained; if the bit error rate is greater than the threshold, it is determined that there is eavesdropping, the negotiation is terminated and restarted.

[0068] 5. Error correction and privacy enhancement (generating the final SK)

[0069] Error correction: Error correction codes (such as LDPC codes or concatenated codes) are used on the remaining bits to eliminate noise errors during transmission, resulting in a consistent key segment K_corrected;

[0070] Privacy Amplification: K_corrected is processed by a hash function (such as SHA-256) to compress and generate a final initial shared key SK of fixed length (≥256 bits), ensuring that the remaining key still has high security even if some information is leaked.

[0071] The entire process described above utilizes the uncertainty principle of quantum mechanics (eavesdropping will change the quantum state) and the no-cloning theorem of quantum mechanics (it is impossible to accurately copy an unknown quantum state) to ensure that the key distribution process is "detectable by eavesdropping," thus guaranteeing the security of the initial shared key SK at the physical level. Furthermore, the generated SK is resistant to quantum computing attacks.

[0072] In step S3, SK is hashed using SHA-256 to obtain the hash value H(SK). H(SK) is then split into multiple sets of parameter fragments, which are used as the initial values ​​and control parameters for the Logistic chaotic mapping and the Lorenz chaotic mapping, respectively, to construct a two-dimensional chaotic mapping system and generate a chaotic sequence set C={C1,C2}. Based on the sensitivity level classification, the corresponding level of translation encryption key K={K1,K2,K3} is extracted from C, specifically including:

[0073] Step 1: Perform a SHA-256 hash on the private key SK to generate a fixed-length hash value H(SK).

[0074] Objective: To convert an arbitrary-length raw private key SK into a fixed-length hash value with an avalanche effect, eliminating the influence of SK format / length on subsequent parameter generation, while ensuring that small changes to SK will lead to drastic changes in the hash value. Core operations:

[0075] Format conversion: Converts the SK string to a UTF-8 encoded byte string to avoid encoding confusion for Chinese characters / special characters.

[0076] SHA-256 calculation: The hash algorithm is called to process the byte string and generate a fixed-length hash value H(SK) (byte string format) of **32 bytes (256 bits)**.

[0077] Feature utilization: Relying on the one-wayness of SHA-256 (SK cannot be derived from H(SK)) and the avalanche effect (if one bit of SK changes, more than half of the bits of H(SK) change), the confidentiality and uniqueness of the hash value are ensured.

[0078] Step 2: Decompose H(SK) into fragments of initial values ​​and control parameters for the chaotic mapping.

[0079] Objective: To split the 32-byte H(SK) into multiple sets of parameter fragments, adapting them to the parameter requirements of Logistic and Lorenz chaotic maps respectively, thus providing input for a two-dimensional chaotic system. Core Operations:

[0080] The splitting rule design is as follows: The 32-byte H(SK) is split into 8 segments (total 4 × 8 = 32 bytes) using 4-byte units as the basic unit. The number of parameters in the chaotic mapping must be matched for the splitting.

[0081] Logistic mapping (one-dimensional): Two segments are assigned, serving as the initial value x0 and the control parameter r, respectively;

[0082] Lorenz mapping (3D): 6 segments are assigned as initial values ​​(x0, y0, z0) and control parameters (σ, ρ, β).

[0083] Parameter range mapping: Converts byte fragments into numerical values ​​and strictly maps them to the chaotic range of the chaotic map (avoiding mapping into a non-chaotic state):

[0084] Logistic: Initial input x0∈(0,1), r∈(3.57,4) (the mapping in this interval exhibits strong chaotic characteristics);

[0085] Lorenz: x0 / y0 / z0∈(0,1), σ∈(9,11), ρ∈(27,29), β∈(2,3) (classical chaotic interval).

[0086] Tool-based conversion: Achieve quantitative mapping from hash fragments to chaotic parameters by converting bytes to unsigned integers and normalizing them to a specified range.

[0087] Step 3: Construct a two-dimensional chaotic mapping system to generate a chaotic sequence set C={C1,C2}

[0088] Objective: Based on the split parameters, run Logistic and Lorenz chaotic mappings respectively to generate chaotic sequences with high randomness, providing raw data for key extraction. Core operations:

[0089] Logistic sequence C1 generation:

[0090] According to the Logistic iteration formula Perform iterative calculations; where For the next iteration value, This is the current iteration value. For control parameters (or growth rate). It must fall strictly within the interval (3.57, 4).

[0091] Discard transient values: discard the first 100 or so iteration values ​​(transient values ​​are greatly affected by the initial values ​​and have poor randomness), and only retain the steady-state sequence.

[0092] Lorenz sequence C2 generation:

[0093] The Lorenz differential equations were numerically solved using the Runge-Kutta method (RK4):

[0094]

[0095] in, It is a system state variable used to describe the three-dimensional state of the Lorenz chaotic system (it has no direct physical meaning and is only a mathematical variable), and it is the core object to be solved numerically. It is the time derivative of the state variable, representing the change in time with the change in state. The rate of change of "virtual time" is the core of the differential equation and determines the direction of the system's evolution. These are control parameters. .

[0096] Extract the values ​​in the x (or y / z) dimension as a sequence, and discard transient values ​​to ensure the pseudo-randomness of the sequence.

[0097] Sequence set encapsulation: The processed Logistic sequence C1 and Lorenz sequence C2 are integrated into a set C={C1,C2}, which serves as the data source for hierarchical key extraction.

[0098] Step 4: Extract the hierarchical encryption key K={K1,K2,K3} from C according to the sensitivity level.

[0099] Objective: Design differentiated extraction rules based on low / medium / high sensitivity levels, generating corresponding keys K1, K2, and K3. Ensure key security increases with sensitivity level to adapt to the encryption requirements of different translated content. Core operations (extraction rule complexity increases progressively):

[0100] Low-level key K1 (for public / normal translation content):

[0101] Convert C1 to a byte array by normalizing it to 0-255.

[0102] Directly extracting the first N bytes (e.g., 32 bytes = 256 bits) as K1 balances efficiency but has the lowest security.

[0103] Medium-level key K2 (suitable for commercial / internal translation content):

[0104] Both C1 and C2 are normalized;

[0105] A fusion sequence is generated by cross-splicing (C1 for even positions and C2 for odd positions);

[0106] Perform a simple XOR operation (such as XORing with a fixed value 0x3F) on the fused sequence to enhance randomness, with moderate security.

[0107] High-level key K3 (for confidential / privacy translations):

[0108] Perform a byte-by-byte XOR operation on the byte arrays of C1 and C2 to generate the basic fusion sequence;

[0109] Using subsequent fragments of the Lorenz sequence as seeds, Fisher-Yates scrambling (disordering) is performed on the base sequence.

[0110] Performing a second SHA-256 hash on the scrambled sequence further enhances randomness by leveraging the hash avalanche effect, resulting in the highest level of security.

[0111] Key encapsulation: Convert the extracted K1, K2, and K3 into byte string format to form the final hierarchical key set K={K1,K2,K3}.

[0112] In summary, by combining the cryptographic properties of SHA-256 and the pseudo-randomness of chaotic mapping, the key possesses the characteristic of "one-time pad," and high-level keys undergo multiple transformations, making them extremely resistant to cracking. The extraction rules become progressively more complex with increasing sensitivity, taking into account both the encryption efficiency and security requirements of different translated content. The fixed length of the hash value and the range restriction of the chaotic parameters avoid sequence failures caused by parameter anomalies, ensuring the stability of the process.

[0113] Step S4: Translation instruction chaos obfuscation stage, specifically: convert the translation instruction set generated by the visualization process analysis into an intermediate instruction sequence. For intermediate instructions involving sensitive language data processing, a chaotic sequence C1 is used for double obfuscation. Specifically: the instruction line is randomly rearranged based on the numerical sorting result of C1, and then the instruction byte is XORed with the corresponding position value of C1 modulo 256 to obtain the obfuscated intermediate instruction set, in order to prevent the translation instructions from being tampered with or reverse parsed.

[0114] In this embodiment of the invention, converting the translation instruction set generated by the visualization process analysis into an intermediate instruction sequence includes the following steps:

[0115] Step 1: Extract core operations and data objects

[0116] First, filter the "operations that must be performed" and "data that needs to be processed" from the visualization instruction set:

[0117] Core operations: file input, text extraction, format cleaning, proper noun recognition, language detection, translation, format restoration, semantic verification, error output, and file output;

[0118] Data objects: Uploaded file (doc / xlsx / pdf), source text, tagged proper nouns, source language type, translated text, final file.

[0119] Step 2: Instruction Atomization Decomposition

[0120] Break down compound operations into basic operations (one instruction does only one thing):

[0121] The original "preprocessing node" has been split into: FORMAT_CLEAN (clean up format symbols) and PROPER_NOUN_DETECT (identify and mark proper nouns);

[0122] The original "translation node" was split into: PROPER_NOUN_RESERVE (preserve proper nouns), LEGAL_TERM_STANDARDIZE (translation of standard legal terms), and SYNTAX_ADAPT (adaptation to Chinese sentence structure).

[0123] The original "post-processing node" has been split into: FORMAT_RESTORE (restore file format) and SEMANTIC_CHECK (semantic ambiguity check).

[0124] Step 3: Standardize instruction format

[0125] The format should be uniformly [operator, input parameter, output parameter, additional configuration], where:

[0126] Operators: All uppercase English letters (for easier machine recognition);

[0127] Input parameters: The data source for the operation (such as file path, text variable);

[0128] Output parameters: The storage key for the operation result (for easy recall by subsequent commands);

[0129] Additional configurations: Optional (such as supported file types, rule IDs, etc.).

[0130] Step 4: Handle branching and flow logic

[0131] The CONDITION instruction explicitly defines the branch decision, and the GOTO instruction implements the flow jump, forming a closed loop, and finally generating an intermediate instruction sequence (standardized result).

[0132] As an example, the steps involved in translating an English contract into Chinese localization are as follows:

[0133] 1. [Input Node] Receives files uploaded by users (supports doc / xlsx / pdf), extracts the plain text content, and stores it as "source text".

[0134] 2. [Preprocessing Node] Performs two operations: ① Remove formatting marks (such as §, ★) from the text; ② Identify and mark proper nouns (such as company names, legal terms).

[0135] 3. [Branch Detection] Detect the language of the source text:

[0136] If the input is English (en): execute the translation process; if the input is another language: output the error message "English input only".

[0137] 4. [Translation Nodes] Translate according to the following rules: ① Proper nouns should be translated using the original words; ② Legal terms should be translated using standard Chinese translations; ③ Sentence structure should be adapted to Chinese expression habits.

[0138] 5. [Post-processing nodes] ① Restore the original file format (such as tables and paragraph structure); ② Verify that the translated text is free of semantic ambiguity;

[0139] 6. [Output Node] Returns the localized Chinese file.

[0140] Step S5: Language data translation and segmented encryption stage, including:

[0141] During the language data translation and segmented encryption stages, the operations performed on core sensitive language data include: performing translation operations based on the obfuscated intermediate instruction set; and then employing a dual mechanism of "chaotic sequence diffusion + AES-256 encryption" to obtain the translated text. First, C2 is used to perform modulo-256 addition diffusion processing on the translated bytes, followed by AES-256 encryption using K1. The AES-256 encryption uses CBC mode, with the initial vector generated from the first 16 bits of C1.

[0142] In this embodiment of the application, after obtaining the translation, a dual mechanism of "chaotic sequence diffusion + AES-256 encryption" is adopted. First, the translated bytes are diffused by modulo 256 addition using C2, and then AES-256 encryption is performed using K1, including:

[0143] Step A1: Prepare input and key parameters

[0144] Input: Translation (string) → First convert to byte stream (such as UTF-8 encoding), denoted as plaintext_bytes (all subsequent operations are based on bytes).

[0145] Chaos parameter C2: The chaotic mapping type needs to be defined (Logistic mapping is recommended, as it is simple and easy to implement). C2 contains two core parameters:

[0146] Initial value x0 (∈(0,1), such as 0.3456789)

[0147] The control parameter r (in the Logistic mapping, when r∈(3.57,4], it enters a chaotic state, such as 3.999)

[0148] AES-256 key K1: Must be 32 bytes (256 bits). If the input K1 is a string, it needs to be converted to 32 bytes using a hash function (such as SHA-256) (to avoid invalid key length).

[0149] Step A2: Generate a chaotic sequence based on C2 (core: consistent with the length of the translated text in bytes).

[0150] The generation of chaotic sequences must satisfy the following:

[0151] Length = Length of the translated byte stream (ensuring that each byte corresponds to a chaos value);

[0152] Each element ∈ [0,255] (for modulo 256 operations, the (0,1) interval values ​​output by the chaotic mapping need to be scaled to integers).

[0153] Taking the commonly used Logistic mapping as an example, the formula is generated as follows:

[0154] Chaotic iteration formula: x(n+1)=r*x(n)*(1-x(n)) (n starts from 0, x0 is the initial value);

[0155] Scaling to [0,255]: chaos_val(n)=int(x(n)*255) (each value after rounding is ∈[0,255]).

[0156] It should be noted that chaotic sequences are extremely sensitive to initial values ​​(a difference of 1e-10 in x0 will result in a completely different sequence). It is necessary to ensure that the C2 (x0, r) of the sender and receiver are completely identical, otherwise decryption is impossible.

[0157] Step A3: Chaotic sequence diffusion (modulo 256 addition)

[0158] For each byte of the translated byte stream, perform a modulo 256 addition with the chaotic value at the corresponding position in the chaotic sequence to obtain the diffused byte stream diffused_bytes.

[0159] Function: To diffuse the statistical characteristics of the original bytes (for example, the high-frequency byte 0x30 (digit 0) in the original text will be offset into different bytes by the chaotic value), thereby increasing the resistance to attacks of subsequent AES encryption (avoiding plaintext attacks).

[0160] Step A4: AES-256 Encryption Preparation (Handling Key and IV)

[0161] AES is a block cipher algorithm (AES-256 has a block length of 128 bits = 16 bytes), and it must meet two requirements:

[0162] The key K1 must be 32 bytes: If K1 is a string (such as "my_secret_key_123"), it needs to be hashed using SHA-256 (the hash result is fixed at 32 bytes).

[0163] A random IV (Initialization Vector) needs to be generated: AES encryption recommends using CBC mode (requires IV) or GCM mode (with authentication, more secure). The IV length is equal to the block length (16 bytes). It needs to be generated randomly (it needs to be transmitted with the ciphertext when sending, and the IV does not need to be kept secret).

[0164] It should be noted that the IV must be randomized (not fixed) for each encryption; otherwise, the same plaintext plus the same key will generate the same ciphertext, leading to a security vulnerability.

[0165] Step A5: AES-256 encryption (after diffusion of byte stream)

[0166] Encrypted diffused_bytes using K1 with AES-256 encryption, the process is as follows:

[0167] Padding is applied to the diffused byte stream (AES requires the plaintext length to be an integer multiple of the block length): PKCS7 padding (default standard);

[0168] Perform AES encryption: Taking GCM mode as an example (with authentication tag, which can prevent tampering), the output is ciphertext (ciphertext) and tag (authentication tag, which needs to be verified during decryption).

[0169] Step A6: Output the final result (for transmission)

[0170] The following three parts must be sent to the recipient together (none are required):

[0171] AES ciphertext;

[0172] AES certification label (tag, required in GCM mode);

[0173] IV (initial vector, 16 bytes);

[0174] Chaos parameter C2 (If not agreed upon by both parties, C2 must be transmitted in encrypted form to prevent it from being stolen).

[0175] During the language data translation and segmented encryption stage, the operations performed on generally sensitive language data include: after performing the translation operation, using the SM4 encryption mechanism and encrypting the translation with K2;

[0176] In the language data translation and segmented encryption stage, the operations on non-sensitive language data include: after performing the translation operation, using lightweight chaotic XOR encryption, and using K3 to perform XOR processing on the translation;

[0177] Step S6: Secure delivery and dynamic update stage, specifically: the segmented encrypted translation is concatenated in the format of "sensitivity level identifier + ciphertext length + ciphertext content", and an identity field based on hardware fingerprint and RSA-2048 signature is added to obtain the identity authentication field (encrypted with K1). The identity authentication field is then transmitted to the receiving end. After verifying the legitimacy of the identity, the receiving end restores the translation according to the corresponding key and decryption mechanism. After completing M translation tasks (M is a configurable value between 5 and 50), the initial shared key SK is updated through the quantum key negotiation protocol, and steps S3-S5 are repeated to regenerate the chaotic sequence and encryption key.

[0178] In step S6, the segmented encrypted translation is concatenated in the format of "sensitivity level identifier + ciphertext length + ciphertext content", and an identity field based on hardware fingerprint and RSA-2048 signature is appended to obtain the identity authentication field (encrypted with K1). The identity authentication field is then transmitted to the receiving end. After verifying the legitimacy of the identity, the receiving end restores the translation according to the corresponding key and decryption mechanism. The specific steps include:

[0179] Step 1: Translated text segmentation and double encryption

[0180] The translation to be transmitted is split into several segments according to preset rules (e.g., long translations are divided into segments of 500 characters each, and short translations are processed as single segments).

[0181] Perform chaotic sequence diffusion + AES-256 encryption on each segment: Use the initial shared key SK to derive the chaotic sequence parameter C2 (or predetermine C2) to generate a chaotic sequence with the same byte length as the segment, and perform modulo 256 addition diffusion on the segmented translated byte stream; then use the AES-256 key K1 derived from SK (e.g., perform SHA-256 hash on SK to obtain 32 bytes K1) to perform AES-256 encryption on the diffused byte stream (GCM mode, generating the IV and Tag of the segment) to obtain the ciphertext of each segment.

[0182] Step 2: Structured splicing of segmented encrypted text

[0183] For each segment of ciphertext, encapsulate it according to a fixed format: sensitivity level identifier + ciphertext length + ciphertext content.

[0184] Sensitivity level identifier: Marked according to the sensitivity of the content in each segment (e.g., 01 for ordinary translation segments), using a fixed length of 2 characters;

[0185] Ciphertext length: Calculate the byte length of the segmented ciphertext, represented as a fixed-length 4-bit number (padded with 0s if necessary, e.g., length 128 is labeled 0128).

[0186] Ciphertext content: Ciphertext encrypted in segments using AES (hexadecimal / binary format).

[0187] By sequentially concatenating the encapsulation results of all segments, a complete concatenated ciphertext is obtained (e.g., 010128a1b2c3...020256d4e5f6...).

[0188] Step 3: Generate authentication fields

[0189] Extract the hardware fingerprint of the sending device and generate a unique DeviceID using SHA-256 hashing;

[0190] Construct the original text to be signed: concatenate DeviceID + current timestamp + random number (to prevent replay and collision), and convert it into a byte stream;

[0191] RSA-2048 signature: Perform an RS256 signature (SHA-256+RSA, PSS padding) on ​​the original text using the locally held RSA private key to generate a signature value;

[0192] Encapsulate the authentication fields: Integrate DeviceID, timestamp, signature value, and validity period (e.g., 5 minutes) into JSON structured data (e.g., {"device_id":"xxx","timestamp":1735689600,"signature":"yyy","expire":300}), which is the original authentication fields.

[0193] Step 4: Encrypt the authentication field

[0194] The original identity authentication field is encrypted using the AES-256 key K1 (using the AES-GCM mode to generate the encrypted IV and Tag), resulting in the encrypted identity authentication field, thus preventing the theft of identity information during plaintext transmission.

[0195] Step 5: Integrate and transmit data

[0196] The complete concatenated ciphertext, the IV / Tag of each segment, the chaotic parameter C2, the encrypted identity authentication field, and the identity-encrypted IV / Tag are integrated into a transmission data packet and sent to the receiving end through a network / quantum channel; at the same time, the sender counts the translation tasks (the count is incremented by 1 for each completed task).

[0197] Step 6: Task count check (key update prerequisite)

[0198] Check if the number of times the current translation task has been completed has reached the configured threshold M (e.g., M=10): If not, repeat steps S1-S5 to process subsequent translations; if it has, trigger the key update process of quantum key negotiation (enter the third module).

[0199] III. Key update process for quantum key negotiation (steps S7-S9)

[0200] When the task count reaches M, the sender and receiver update the initial shared key SK through a quantum key negotiation protocol and regenerate the chaotic sequence and encryption key to ensure dynamic key security. Specific steps are as follows:

[0201] Step 7: Initiate quantum key negotiation

[0202] The sender and receiver execute a pre-defined quantum key negotiation protocol (such as the BB84 protocol) through a quantum channel (such as a fiber optic quantum channel):

[0203] The sender randomly generates a sequence of qubits (such as 0 / 1 encoding based on polarization states) and sends it to the receiver through a quantum channel;

[0204] The receiver randomly selects a measurement basis to measure the qubit and feeds back the measurement basis information;

[0205] The sender and receiver compare and select consistent measurement bases to generate original quantum key material. After information coordination and privacy amplification processing, a new initial shared key SK_new (with the same length as the original SK, such as 256 bits) is obtained.

[0206] Step 8: Regenerate the chaotic sequence and encryption key

[0207] Re-derive the relevant keys and parameters using the updated SK_new:

[0208] Perform a SHA-256 hash on SK_new to generate a new AES-256 key K1_new;

[0209] Adjust the chaotic mapping parameter C2 using a portion of the bits of SK_new (e.g., the first 64 bits) to update the initial value x0 of the Logistic mapping, and generate a new chaotic sequence parameter C2_new.

[0210] Step 9: Reset task count and synchronize parameters

[0211] The sender resets the task count to 0 and sends the relevant synchronization information of SK_new, C2_new, and K1_new (encrypted via quantum channel) to the receiver to ensure that the parameters of both parties are consistent.

[0212] After synchronization is complete, return to steps 3-5 to reprocess the encryption and transmission of subsequent translations based on the new key and parameters.

[0213] As an optional implementation, in step S3, the parameter segment division method is as follows: the first 64 bits of the hash value H(SK) are converted into the initial value x0 of the Logistic chaotic map, where 0 < x0 < 1; the middle 64 bits of the hash value H(SK) are converted into the control parameter μ, where 3.57 < μ < 4; the 129 - 256 bits of the hash value H(SK) are split and converted into the initial values (x1, y1, z1) and control parameters (σ, ρ, β) of the Lorenz chaotic map. The encryption key generation rules are as follows: K1 is generated by performing an exclusive OR operation on C1 and C2 and then performing SHA - 384 hashing to generate a 256 - bit key; K2 is generated by concatenating C1 and then performing SM3 hashing to generate a 128 - bit key; K3 intercepts the first 64 bits of C2 as the key. In step S4, the translation instruction set parsing process is as follows: perform syntax analysis on the operator nodes and their association relationships in the visualization process to generate a structured translation instruction set including language conversion rules, semantic mapping logic, and format optimization instructions to ensure the accuracy of instruction execution. In step S6, the identity authentication fields include the sender's hardware fingerprint (the sender's hardware fingerprint includes the SHA - 256 hash values of the CPU serial number and the network card MAC address), the user's public key, and the public key signature information. The receiver verifies the signature legality through the user's public key to confirm the sender's identity.

[0214] As an optional embodiment, a language data translation method based on visualization process configuration includes a visualization translation process configuration stage, a language data sensitivity level division stage, a dynamic encryption key generation stage, a translation instruction chaotic confusion stage, a language data translation and segmented encryption stage, and a secure delivery and dynamic update stage. The specific details of each stage are as follows:

[0215] 1. Visualization translation process configuration stage

[0216] This stage provides users with flexible customization capabilities to solve the defect of the fixed traditional translation process:

[0217] Build a multi - scenario translation operator library: The predefined operators include a language recognition operator (supporting automatic recognition of more than 50 mainstream languages), a word segmentation and parsing operator (adapting to the grammar rules of different languages), a semantic mapping operator (supporting general translation and customized mapping in professional fields (finance, medical)), a grammar optimization operator (optimizing the fluency of the translation), a format adaptation operator (supporting output in multiple formats such as documents, conversations, tables), and an encryption control operator (associating sensitivity levels with encryption policies);

[0218] Visual configuration interface: Users can add operator nodes to the configuration canvas by dragging and dropping, set the node relationship by connecting the nodes (such as "language recognition → word segmentation and parsing → semantic mapping → syntax optimization → encryption control → output"), and configure the parameters of each operator (such as target language selection, translation accuracy level (normal / professional / high accuracy), sensitivity level threshold (such as marking keywords containing "confidential" or "privacy" as core sensitive)).

[0219] Workflow management function: Supports saving the configured translation workflow as a template (such as "Government Affairs Confidential Translation Template" or "Medical Record Translation Template"), which can be directly reused or edited based on the template to improve configuration efficiency.

[0220] 2. Language Data Sensitivity Level Classification Stage

[0221] This phase achieves precise classification of language data, providing a basis for differentiated encryption:

[0222] The classification criteria are based on a combination of multiple factors, including text keyword matching (such as sensitive words like "confidential", "secret", "privacy", "account", etc.), user-preset thresholds, data application scenarios (such as government confidentiality scenarios with a default high sensitivity level), and the scope of impact of the leak, to automatically identify the sensitivity level.

[0223] Grading rules:

[0224] Core sensitive language data: This includes classified government documents, core financial business documents, medical privacy descriptions, and sensitive personal conversation records, which, if leaked, could lead to serious security incidents, significant economic losses, or privacy violations.

[0225] General sensitive language data includes routine business communication texts, descriptions of basic user information (such as names and contact information in natural language), and non-core business documents. Data that would have some impact if leaked but does not involve core interests.

[0226] Non-sensitive language data: This includes publicly available promotional materials, general information texts (such as weather and news summaries), and shareable general knowledge texts, which pose no substantial harm if leaked.

[0227] Sensitivity level labeling: Add sensitivity level labels (such as "core sensitive", "generally sensitive", "non-sensitive") to the segmented language data and synchronize them to the subsequent translation and encryption stages.

[0228] 3. Dynamic encryption key generation stage

[0229] This stage integrates quantum key negotiation and two-dimensional chaotic mapping to generate highly secure, dynamic encryption keys, solving the problems of insecurity and static encryption parameters in traditional key distribution.

[0230] Quantum Key Agreement: The translation initiating end (such as a user terminal) and the receiving end (such as a target server) adopt the BB84 quantum key agreement protocol. Encoded single-photon pulses are transmitted through a quantum channel, and basis selection information and sampling detection information are transmitted through a classical channel. Both sides randomly select some key segments for eavesdropping detection. If the bit error rate ≤ the preset threshold (0.05 - 0.1), error correction and privacy amplification are performed on the remaining key segments to generate an initial shared key SK of more than 256 bits. If the bit error rate exceeds the standard, it is determined that there is eavesdropping, and the negotiation is restarted.

[0231] Chaotic Mapping Parameter Configuration: Perform SHA-256 hashing on the initial shared key SK to obtain a 256-bit hash value H(SK), ensuring the uniqueness and irreversibility of parameter generation. Split H(SK) into parameter segments and configure a two-dimensional chaotic mapping:

[0232] Logistic Chaotic Mapping: Take the first 64 bits of H(SK), and convert it to an initial value of 0 < x0 < 1 through the formula x0 = (64-bit value) / 2 64 The middle 64 bits are converted to a control parameter of 3.57 < μ < 4 through the formula μ = 3.57 + (64-bit value) / (2 64 ×0.43) to generate a Logistic chaotic sequence C1 (with a length matching the number of bytes of the data to be translated);

[0233] Lorenz Chaotic Mapping: Take the 129th - 256th bits of H(SK), split and convert them into initial values (x1, y1, z1) (range 1 - 10) and control parameters (σ = 10, ρ = 28, β = 8 / 3), and generate a chaotic sequence C2 through the Lorenz equation;

[0234] Hierarchical Encryption Key Generation: Based on the result of sensitivity level division, generate keys of different strengths based on the chaotic sequence C:

[0235] Core Sensitive Key K1: Exclusive-OR C1 and C2 byte by byte (C1[i]^C2[i]), perform SHA-384 hashing on the result, and take the first 256 bits as K1 (AES-256 key);

[0236] General Sensitive Key K2: Concatenate C1 byte by byte and generate a 128-bit key K2 (SM4 key) through the SM3 hash function;

[0237] Non-Sensitive Key K3:截取 the first 64 bits of C2 as K3 and adopt a lightweight key to reduce the computational overhead.

[0238] 4. Translation Instruction Chaotic Confusion Phase

[0239] This stage involves chaotic obfuscation of the translation instructions to prevent tampering or reverse engineering, thus ensuring the security of the translation process.

[0240] Translation instruction set parsing: Perform syntactic analysis on the translation process of the visual configuration, extract the execution logic, parameter configuration and data flow of the operator nodes, and generate a structured translation instruction set (such as "Identify language as English → Segment words according to financial field → Map to Chinese semantics → Optimize syntactic structure → Encryption level: Core sensitive").

[0241] Intermediate instruction sequence conversion: Converting the structured translation instruction set into a machine-executable intermediate instruction sequence to ensure instruction executability;

[0242] Double Chaos Obfuscation: For intermediate instructions involving sensitive data processing (instructions corresponding to core sensitive and general sensitive data), perform double obfuscation operations:

[0243] Scrambling: Sort the C1 sequence by numerical value to obtain a sorted index sequence. Randomly rearrange the rows of the intermediate instruction sequence according to the index, thus disrupting the original execution order of the instructions.

[0244] Byte obfuscation: Perform a modulo 256 XOR operation between the rearranged instruction byte and the value at the corresponding position in C1 (instruction byte[i] = instruction byte[i]^(C1[i] mod 256)) to obtain an obfuscated intermediate instruction set, ensuring that the instruction cannot be directly parsed.

[0245] 5. Language data translation and segmented encryption stage

[0246] This phase achieves deep collaboration between translation and encryption, employing differentiated mechanisms based on sensitivity levels to balance security and efficiency:

[0247] 5.1 Core Sensitive Language Data Processing

[0248] Translation execution: Based on the obfuscated intermediate instruction set, the translation engine is invoked to perform translation operations on core sensitive language data (such as Chinese-English translation of classified documents), ensuring the accuracy and professionalism of the translation;

[0249] Double encryption:

[0250] Chaotic sequence diffusion: The translation is converted into a byte array, and the value of each byte and the corresponding position of C2 are added modulo 256 (byte[i] = (byte[i] + C2[i]) mod 256), which destroys the original byte distribution pattern of the translation;

[0251] AES-256 encryption: Using K1 as the key, the diffused byte array is encrypted with AES-256 in CBC mode. The initialization vector (IV) is generated from the first 16 bits of C1 to further enhance the encryption strength and obtain the core sensitive ciphertext.

[0252] 5.2 Processing of General Sensitive Language Data

[0253] Translation execution: Perform translation operations based on the obfuscated intermediate instruction set (such as bilingual conversion of business text);

[0254] SM4 encryption: Using K2 as the key, the translated byte array is encrypted using SM4 (ECB mode) to obtain general sensitive ciphertext, balancing security and computational efficiency.

[0255] 5.3 Non-sensitive language data processing

[0256] Translation execution: Perform translation operations based on the original intermediate instruction set (without obfuscation) (such as translating publicly available promotional materials);

[0257] Lightweight XOR encryption: Using K3 as the key, an XOR operation is performed on the translated byte array (byte[i] = byte[i]^K3[imod64]) to obtain non-sensitive ciphertext, reducing computational overhead and improving translation and encryption efficiency.

[0258] Finally, the three types of ciphertext are concatenated in the format of "sensitivity level identifier + ciphertext length + ciphertext content" to form the final ciphertext packet to be transmitted.

[0259] 6. Secure Delivery and Dynamic Update Phase

[0260] This stage achieves secure transmission of encrypted data, authentication, and dynamic key updates, forming a secure closed loop:

[0261] 6.1 Generation and Attachment of Identity Authentication Fields

[0262] The sending end generates an authentication field to ensure that the receiving end can verify the sender's legitimacy:

[0263] Hardware fingerprint extraction: Collect the CPU serial number and network card MAC address of the transmitting device, perform SHA-256 hash processing, and obtain the hardware identifier HF;

[0264] RSA Signature: Use the local RSA-2048 private key to sign the HF and obtain the signature information S;

[0265] Encrypted identity field: The user's public key PK, hardware identifier HF, and signature information S are concatenated and encrypted with K1 using AES-256 (ECB mode) to obtain the identity authentication field Auth.

[0266] Encrypted packet assembly: Auth is appended to the header of the final encrypted packet to form complete transmitted data.

[0267] 6.2 Receiver Verification and Decryption

[0268] After receiving the transmitted data, the receiving end performs the following operations:

[0269] Authentication: Decrypt Auth with K1 to obtain PK, HF, and S; verify the validity of S with PK and confirm that HF ​​is the real hardware fingerprint of the sender; if the verification is successful, continue decryption; if the verification fails, discard the data and refuse to process it.

[0270] Segmented decryption: Based on the sensitivity level identifier in the ciphertext packet, the corresponding keys (K1, K2, K3) and decryption mechanisms are used to restore the translation, ensuring the integrity and accuracy of the translation.

[0271] 6.3 Dynamic Key Update

[0272] Set an update threshold M (configurable from 5 to 50). After each M translation task is completed, the sender and receiver automatically trigger the quantum key negotiation protocol to regenerate the initial shared key SK_new, replace the original SK, and repeat steps 3-5 to regenerate the chaotic sequence and encryption key. This realizes a dynamic update mechanism of "one session, multiple keys" to avoid security risks caused by long-term key use.

[0273] The present invention will be further described in detail below with reference to specific embodiments:

[0274] As an example: the translation of classified government documents

[0275] A government department needs to translate classified policy documents (core sensitive), routine business notices (generally sensitive), and public announcements (non-sensitive) from Chinese to English. The method described in this invention is used to address this issue.

[0276] Visualized workflow configuration: Users can select the operators "Language Recognition → Government Domain Word Segmentation → Semantic Mapping → Syntax Optimization → Encryption Control → PDF Output" by dragging and dropping, configure the target language as English, the translation accuracy as "High Accuracy", and the sensitivity level threshold as "Mark keywords containing 'Confidential' or 'Secret' as core sensitive", and save it as "Government Translation Template";

[0277] Sensitivity Level Classification: The system automatically identifies document keywords and classifies policy documents containing the "Confidential" label as core sensitive, business notices as generally sensitive, and public announcements as non-sensitive;

[0278] Dynamic key generation: The government terminal and the target server generate a 256-bit SK through the BB84 protocol. After hash processing, the SK is split and configured with Logistic and Lorenz chaotic mappings to generate C1 and C2 sequences, and then generate hierarchical keys K1, K2 and K3.

[0279] Instruction obfuscation and translation: Translation instructions corresponding to core sensitive and general sensitive instructions are obfuscated by C1 scrambling and XORing before the translation operation is performed to ensure the accuracy of the translation;

[0280] Segmented encryption: Core sensitive translations are encrypted using C2 diffusion + AES-256, general sensitive translations are encrypted using SM4, and non-sensitive translations are encrypted using K3 XOR encryption. The segments are then concatenated to form a ciphertext packet.

[0281] Secure delivery: Transmission is performed with an additional identity authentication field (government terminal hardware fingerprint + RSA signature). The server decrypts the data after verifying the identity. SK is automatically updated every 20 translation tasks completed.

[0282] This embodiment effectively protects the translation security of classified documents and prevents the leakage of sensitive policy information, while also taking into account the translation efficiency of regular documents.

[0283] As another example: multilingual translation of medical records

[0284] A hospital needs to translate patient medical records (core sensitive information), patient basic information descriptions (generally sensitive information), and publicly available hospital information (non-sensitive information) from Chinese into English for international consultations.

[0285] Visual workflow configuration: Select the operator "Language recognition → Medical domain word segmentation → Professional semantic mapping → Syntax optimization → Encryption control → Word format output", configure the target language as English and the translation accuracy as "Professional level", and save it as "Medical translation template";

[0286] Sensitivity level classification: The system identifies private medical records (including medical history and genetic information descriptions) as core sensitive, patient names and contact information descriptions as generally sensitive, and public introductions as non-sensitive;

[0287] Dynamic key generation and encryption: SK is generated through quantum key negotiation, and the key is generated through chaotic mapping. Core sensitive medical records are double-encrypted, general sensitive information is encrypted with SM4, and non-sensitive information is lightly encrypted.

[0288] Secure delivery: The authentication field, consisting of the fingerprint of the hospital equipment hardware and the doctor's public key signature, is transmitted to the international consultation server. After the server verifies the data, it is decrypted. The SK is updated every 10 translation tasks completed.

[0289] This embodiment ensures the security of translating and transmitting patient privacy medical records, and complies with compliance requirements for medical data privacy protection.

[0290] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of this application.

[0291] In summary, the language data translation method based on visual workflow configuration provided in this application significantly improves configuration flexibility: through visual drag-and-drop configuration and template reuse, it supports users in customizing translation workflows according to different scenarios, solving the problems of fixed workflows and poor adaptability of traditional translation tools, and adapting to the needs of multiple fields such as government affairs, finance, and healthcare; at the same time, it has strong sensitive data security protection capabilities: by integrating quantum key negotiation and two-dimensional chaotic mapping, it achieves full-link protection of "secure key distribution → dynamic parameter generation → differentiated encryption", and core sensitive data adopts double encryption, with outstanding resistance to quantum attacks, brute-force attacks, and statistical attacks, fundamentally solving the risk of sensitive language data leakage; and it has excellent encryption and translation synergy: Differentiated encryption is implemented based on sensitivity levels, with high-strength protection for core data and lightweight encryption for non-sensitive data, balancing translation efficiency and security while avoiding encryption redundancy or insufficient protection. A complete security closed loop is achieved: an integrated hardware fingerprint and RSA signature authentication mechanism prevents attacks that forge translations; keys are dynamically updated based on the number of translations, reducing the risk of key leakage and forming a complete security closed loop of "configuration → translation → encryption → transmission → verification → update". It boasts strong practicality and scalability: supporting translation needs across multiple languages, formats, and scenarios; the generated target ciphertext can be securely deployed on different platforms (terminals, servers, cloud platforms), complying with the requirements of the Personal Information Protection Law and the Data Security Law regarding sensitive data processing.

[0292] Based on the above-described language data translation method based on visual process configuration, this application provides a language data translation system based on visual process configuration, including a visual process configuration module, a sensitivity level classification module, a dynamic key generation module, an instruction obfuscation module, a translation and encryption module, and a secure delivery and update module. Each module is used to execute the corresponding step of any of the methods described in the language data translation method based on visual process configuration.

[0293] Specifically, the visual process configuration module, sensitivity level classification module, dynamic key generation module, instruction obfuscation module, translation and encryption module, and secure delivery and update module are used to execute steps S1 to S6, respectively.

[0294] In summary, the language data translation method and system based on visual process configuration provided in this application, based on quantum key negotiation and the uncertainty principle and no-cloning theorem of quantum mechanics, can achieve secure key distribution and fundamentally solve the key leakage problem. Chaotic mapping has extreme sensitivity to initial values ​​and control parameters and pseudo-randomness; the sequences it generates can be used to optimize encryption parameters and improve anti-cracking capabilities. Visual process configuration can meet users' customized needs. Therefore, deeply integrating visual process configuration, quantum key negotiation, chaotic mapping encryption, and language data translation to construct a translation method of "customized configuration + hierarchical encryption + dynamic security" is an effective way to overcome the shortcomings of existing technologies.

[0295] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, this invention is not directed to any particular programming language. It should be understood that the contents of the invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the invention.

[0296] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0297] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components in the apparatus according to embodiments of the present invention. The present invention can also be implemented as a device or apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

Claims

1. A language data translation method based on a visualized process configuration, characterized in that, The method comprises the following steps: Step S1: visual translation process configuration stage, including obtaining user input language data to be translated; Step S2: language data sensitivity level division stage: automatic sensitivity level identification is performed on the input language data to be translated, combined with the preset threshold value set by the user, and sensitivity level division is performed, which is divided into three levels of core sensitive language data, general sensitive language data and non-sensitive language data; wherein the core sensitive language data includes classified documents, private conversations and core business documents; the general sensitive language data includes routine business text and user basic information description; the non-sensitive language data includes public propaganda materials and general information text; Step S3: dynamic encryption key generation stage: based on the BB84 quantum key agreement protocol, an initial shared key SK is generated for the establishment of a secure channel between the two communication parties, the SK is subjected to SHA-256 hash processing to obtain a hash value H(SK), the H(SK) is split into multiple groups of parameter fragments, and the multiple groups of parameter fragments are used as the initial values and control parameters of the Logistic chaotic mapping and Lorenz chaotic mapping respectively, a two-dimensional chaotic mapping system is constructed, a chaotic sequence set C={C1,C2} is generated, and then according to the sensitivity level division result, the translation encryption key K={K1,K2,K3} of the corresponding level is extracted from C; Wherein, C1 is a chaotic sequence generated by Logistic chaotic mapping; C2 is a chaotic sequence generated by Lorenz chaotic mapping; K1 represents a core sensitive data encryption key; K2 represents a general sensitive data encryption key; K3 represents a non-sensitive data encryption key; Step S4: translation instruction chaotic confusion stage: the translation instruction set generated by visual process analysis is converted into an intermediate instruction sequence, and the intermediate instructions related to sensitive language data processing are subjected to double confusion by using the chaotic sequence C1, specifically: based on the numerical ordering result of C1, the instruction line is randomly rearranged, and then the instruction byte and the numerical value at the corresponding position of C1 are subjected to modulo 256 XOR operation to obtain the confused intermediate instruction set, so as to prevent the translation instruction from being tampered with or reverse analyzed; Step S5: language data translation and segmented encryption stage: In the language data translation and segmented encryption stage, the operation on the core sensitive language data includes: based on the confused intermediate instruction set, the translation operation is performed, the text is subjected to "chaotic sequence diffusion + AES-256 encryption" double mechanism after the translation, the text bytes are subjected to modulo 256 addition diffusion processing by using C2, and then AES-256 encryption is performed by using K1; In the language data translation and segmented encryption stage, the operation on the general sensitive language data includes: after the translation operation, SM4 encryption mechanism is adopted, and the text is encrypted by using K2; In the language data translation and segmented encryption stage, the operation on the non-sensitive language data includes: after the translation operation, a lightweight chaotic XOR encryption is adopted, and the text is subjected to XOR processing by using K3; Step S6: secure delivery and dynamic update phase: the segmented encrypted translation is spliced in the format of "sensitivity level identifier + ciphertext length + ciphertext content", and the identity field is attached based on the hardware fingerprint and RSA-2048 signature, and the identity authentication field is obtained, which is transmitted to the receiving end; after verifying the identity legality, the receiving end restores the translation according to the corresponding key and decryption mechanism; after completing M translation tasks, the initial shared key SK is updated through the quantum key agreement protocol, and the chaotic sequence and encryption key are regenerated by repeating steps S3-S5.

2. The language data translation method based on a visualization flow configuration according to claim 1, characterized in that, In step S3, the parameter segment division method is: the first 64 bits of the hash value H(SK) are converted into the initial value x0 of the Logistic chaotic mapping, where 0 3. The method of claim 1, wherein the visual flow configuration-based language data translation method is characterized by, In step S3, the encryption key generation rule is: K1 is generated by XOR operation of C1 and C2 and then processed by SHA-384 hash to generate a 256-bit key; K2 is generated by splicing C1 and then processed by SM3 hash to generate a 128-bit key; K3 is generated by intercepting the first 64 bits of C2 as a key.

4. The method of claim 1, wherein the visual flow configuration-based language data translation method is characterized by, In step S4, the translation instruction set analysis process is: performing syntax analysis on the operator nodes and associated relationships of the visual process to generate a structured translation instruction set containing language conversion rules, semantic mapping logic, and format optimization instructions to ensure the accuracy of instruction execution.

5. The method of claim 1, wherein the language data translation based on the visualization flow configuration is configured to, In step S6, the identity authentication field includes the sending end hardware fingerprint, user public key, and public key signature information, and the receiving end verifies the signature legality through the user public key to confirm the sending end identity.

6. The method of claim 1, wherein the language data translation based on the visualization flow configuration is configured to, The preset threshold includes a sensitivity level threshold of language data sensitivity level division, an eavesdropping detection error rate threshold of quantum key agreement, and a translation task threshold of key dynamic update.

7. The language data translation method based on a visualization flow configuration according to claim 6, characterized by, The sensitivity level threshold of language data sensitivity level division is greater than or equal to 3.

8. The method of claim 6, wherein the language data translation based on the visualization flow configuration is configured to, The eavesdropping detection error rate threshold of quantum key agreement is 0.05-0.

1.

9. The method of claim 6, wherein the language data translation based on the visualization flow configuration is configured to, The translation task threshold of key dynamic update is between 5 and 50.

10. A language data translation system based on a visualized process configuration, characterized by, The system includes a visual process configuration module, a sensitivity level division module, a dynamic key generation module, an instruction obfuscation module, a translation and encryption module, and a secure delivery and update module, which are respectively used to perform the corresponding steps of the method of any one of claims 1-9.

Images