Platform private corpus unvarnished transmission method based on syntactic tree matching and reversible hash encryption

Abstract

The invention discloses a platform private corpus unvarnished transmission method based on syntactic tree matching and reversible hash encryption. The method comprises the following steps: S1, generating a structured private corpus; s2, performing syntactic analysis on the structured private corpus by adopting a natural language processing technology, and constructing a complete syntactic tree model of the private corpus; s3, generating syntactic structure verification information according to the syntactic tree model, forming structure fingerprint data, and constructing an unvarnished transmission data packet; s4, transmitting the unvarnished transmission data packet between heterogeneous platforms; and S5, after the structural verification is consistent, under the condition that a legal authorization condition is met, performing data restoration on the ciphertext private corpus in the unvarnished transmission data packet by adopting a preset reversible Hash decryption operation, and restoring the original structured private corpus. According to the invention, high-strength and low-redundancy private corpus encryption transmission is realized.

Description

Technical Field

[0001] The present invention relates to the field of encryption technology, and in particular to a platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption. Background Art

[0002] With the increasing prevalence of multi-platform collaboration, the demand for secure transmission of private data between different systems continues to grow. In the field of natural language processing, transmission tasks involving the structural integrity, semantic consistency, and controllable access rights of text are showing a significant upward trend. Traditional data encryption and transmission mechanisms are gradually exposing a series of technical limitations when dealing with such structured data.

[0003] Currently, mainstream corpus encryption methods are mostly based on symmetric encryption algorithms, such as AES or DES. These methods use a shared key for encryption and decryption. While they offer advantages such as fast encryption and decryption speeds and mature algorithms, they were originally designed for data content protection and were not designed for "text structure protection" or "semantic consistency verification." Therefore, when corpora are transmitted across multiple heterogeneous platforms, especially when format conversion, intermediate node storage, or content format compatibility processing are involved, problems such as semantic misalignment, grammatical structure disorder, and tag information loss often occur after decryption, seriously affecting the accuracy of downstream processing and semantic understanding.

[0004] Furthermore, symmetric encryption methods typically output data in an unreadable ciphertext format. Their core advantage lies in the irreversibility of the original text, which in turn leads to a lack of flexibility in scenarios requiring data compliance audits, permission rollback, or error recovery. For example, in the event of mistransmission of text data or policy updates, the recipient cannot recover the original content without the key, hindering problem analysis and decision tracking, limiting their adaptability in scenarios requiring both high security and high controllability.

[0005] In summary, while traditional symmetric encryption can provide content encryption, it suffers from inherent flaws in structural fidelity, decryption controllability, and secure reversibility, making it difficult to meet the comprehensive requirements of corpus-level cross-platform transmission. The reversible hash encryption mechanism proposed in this paper, by integrating natural language structure understanding capabilities with a hierarchical key strategy, achieves functional expansion and enhancement in corpus encryption technology. This significantly differs from existing technology approaches and has broad application value and industrial promotion potential. Summary of the Invention

[0006] One purpose of the present invention is to propose a platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption, which realizes high-intensity, low-redundancy private corpus encrypted transmission.

[0007] According to an embodiment of the present invention, a method for transparently transmitting platform private corpus based on syntax tree matching and reversible hash encryption includes the following steps:

[0008] S1. Obtain the private corpus text to be transmitted, and standardize the private corpus text to generate a structured private corpus;

[0009] S2. Use natural language processing technology to perform syntactic analysis on the structured private corpus and construct a complete syntactic tree model of the private corpus. The complete syntactic tree model reflects the grammatical components, hierarchical relationships, and logical structure of the private corpus.

[0010] S3. Generate grammatical structure verification information based on the syntactic tree model to form structural fingerprint data, implement reversible hash encryption on the structured private corpus, scramble the private corpus using a preset reversible hash encryption algorithm, and embed authorization decryption information during the encryption process to form ciphertext private corpus. Encapsulate the structural fingerprint data and the ciphertext private corpus to construct a transparent transmission data packet;

[0011] S4. Transmit transparent data packets between heterogeneous platforms using multi-node links for data transfer. On the target platform, the received transparent data packets are parsed, structural fingerprint data in the transparent data packets is extracted and verified, and the consistency of the syntactic tree model of the private corpus with the original syntactic structure is verified.

[0012] S5. After the structure verification is consistent, and under the conditions of legal authorization, the ciphertext private corpus in the transparent data packet is restored using the preset reversible hash decryption operation to recover the original structured private corpus.

[0013] Optionally, the S1 includes the following steps:

[0014] S11. Obtain the private corpus text to be transmitted and construct the original private corpus set T, wherein the original private corpus set T consists of multiple private corpus texts to be transmitted. Let the i-th private corpus text be t i , the total number of private corpus texts in the private corpus set is N;

[0015] S12. For each private corpus text t in the original private corpus set T i Perform character encoding standardization processing to uniformly convert the original character encoding format into a standard character encoding format to obtain a private corpus set after encoding standardization;

[0016] S13. Performing text format standardization on the encoded and standardized private corpus set, including removing redundant line breaks, spaces, and non-standard punctuation, and unifying paragraph structure and punctuation style, to obtain a unified format private corpus set;

[0017] S14. Perform semantic label standardization on the private corpus set after format unification, extract the named entities, keywords and grammatical markers contained in each private corpus text, and perform semantic label unified encoding to obtain the semantic label set L corresponding to the i-th private corpus text i and the jth semantic tag l in the semantic tag set ij , forming a semantic annotation mapping set T L ;

[0018] S15. Map the semantic annotations to the set T L Constructed as a structured private corpus set T S .

[0019] Optionally, the structured private corpus set T L The i-th structured private corpus in is S i , structured private corpus consists of private corpus text with unified format With a standardized set of semantic tags composition.

[0020] Optionally, S2 includes the following steps:

[0021] S21. For the structured private corpus set T S Each structured private corpus S in i Perform component syntax analysis and build a component syntax tree in, is the component syntax tree Q i The set of grammatical structure component nodes in represents the kth grammatical structure component node, is the component syntax tree Q i The edge set in represents the hierarchical relationship between grammatical structures. If the grammatical structure is composed of nodes Is the parent node, another grammatical structure component node To modify its child nodes, there is a directed edge

[0022] S22. Private Corpus Text Synchronously perform dependency syntax analysis and build a dependency graph Among them, the node set is the set of vocabulary nodes in the dependency graph, Represents private corpus text The kth lexical node in , is the set of dependency edges between vocabulary nodes, Represents a vocabulary node With vocabulary nodes There are dependency relationships between them, including subject and predicate, verb and object, and modification;

[0023] S23. After the component syntax tree and dependency graph structures are built, in order to determine whether to perform the structure fusion operation, the structure coordination scoring function is calculated:

[0024]

[0025] Among them, δ(·) represents the structural consistency function, and Map(·) represents the structural node mapping function;

[0026] The fusion confidence coefficient is constructed based on the value of the collaborative scoring function and the semantic label scale:

[0027]

[0028] Among them, λ1,λ2 are adjustment parameters, σ(·) is the Sigmoid function, if the fusion reliability coefficient η i >0.5, then the component syntax tree Q i With the dependency graph G i Perform structural fusion to generate a composite syntax tree model P i =(V i ,E i ), otherwise, the structural model with high collaborative score is retained as the composite syntax tree model, and the composite node set V of the composite syntax tree model is obtained. i With the composite edge set E i ;

[0029] S24. Based on the composite syntax tree model P i , for each composite node v ij ∈V i Calculate the structure depth d ij and semantic sensitivity markers ij ∈{0,1}, and according to the fusion reliability coefficient η i Constructing node semantic-structural weight function:

[0030]

[0031] Among them, α and β are structural and semantic adjustment coefficients, w ij Represents a composite node v ij The semantic-structural joint weight of

[0032] S25. In the composite syntax tree model P i Perform root-leaf pre-order traversal in the process and construct the path encoding sequence φ ij , and based on the composite node weight w ij Construct the fusion path representation:

[0033] ψij =φ ij ⊕Hash(w ij );

[0034] Among them, ψ ij represents the fusion path representation after the fusion of path encoding and structural weight, Hash(·) is the structural stable mapping function, and the path fusion feature set is recorded as Ψ i ={ψ ij};

[0035] S26. For continuous structured private corpus pairs (S i ,S i+1 ), constructing a contextual bridging edge set between its composite syntax tree models:

[0036]

[0037] Among them, ContextSim(·) represents the contextual semantic similarity function based on word embedding, and γ is the set threshold;

[0038] Construct a set of structural adhesion paths within a corpus segment based on the set of contextual bridging edges:

[0039]

[0040] in, represents the set of structural adhesion paths between the i-th to the k-th private corpus;

[0041] S27. Composite grammar model after context enhancement Node semantic-structural weight set {w ij}、Path fusion feature set Ψ i ={ψ ij} are stored together in the context-enhanced composite syntax tree model

[0042] Optionally, the composite node set and composite edge set of the composite syntax tree model are respectively defined as:

[0043]

[0044] in, Represents a mapping function based on Aligned connecting edges are created between component nodes and dependent nodes.

[0045] Optionally, S3 includes the following steps:

[0046] S31. Context-enhanced composite syntax tree model Each composite node v ij Traverse and represent the corresponding fusion path ψij and node semantic-structural weight w ij After connecting in sequence, input the one-way hash function Hash to obtain the node structure hash value h ij , and collected into a node structure hash value set {h ij};

[0047] S32. Based on the node semantic-structural weight w ij As a weighted coefficient, the node structure hash value set {h ij} perform weighted summation, and then input the summation result into the one-way hash function Hash to obtain the structured private corpus S i One-to-one corresponding structural fingerprint data F i ;

[0048] S33. The structural fingerprint data F i Bridge path collection with context The hash results are sequentially connected and fed into the hash function based on the key K struct The message authentication code function HMAC is used to obtain the syntax structure verification tag Tag i ;

[0049] S34. Use the preset reversible hash encryption algorithm R enc With a one-time symmetric scrambling key K enc , reversible recovery key index value K rev And the corpus number i is input, for the private corpus text Perform reversible scrambling to obtain the ciphertext private corpus c i ;

[0050] S35. Constructing the authorization decryption information metadata M i , the authorized decryption information metadata contains the reversible recovery key index value K rev , Authorization decryption validity period ExpTime i and the authorization subject identifier AuthID i ;

[0051] S36. The ciphertext private corpus c i , grammatical structure check tag i and authorization decryption information metadata M i Combine to form transparent ciphertext package C i .

[0052] Optionally, the S4 includes the following steps:

[0053] S41. The transparent ciphertext packet set C = {C i Data is transferred through multiple network node links between heterogeneous platforms, and the network node link path is defined as an ordered node sequence:

[0054] N i ={n1→n2→…→n m};

[0055] Among them, N i Indicates the transmission of the i-th transparent ciphertext package C i The network node link path traversed, n j represents the jth transmission node, and m is the total number of transmission nodes;

[0056] S42. Receive network node n on the target platform m For transparent ciphertext package C i Parse and separate the ciphertext private corpus c i , grammatical structure check tag i and authorization decryption information metadata M i ;

[0057] S43. Verify the key K based on the structure struct Check the structure tag i Perform message authentication code verification and extract the structural fingerprint data F contained therein i Bridge path collection with context The hash representation of the node is used to perform the structural consistency reconstruction process, from the node semantic-structural weight set {w′ ij}、Fusion path representation set {ψ′ ij} and context-enhanced syntactic models , calculate the reconstructed structural fingerprint data:

[0058]

[0059] S44. Compare the original value of the structural fingerprint data with the reconstructed value of the structural fingerprint data and calculate the structural consistency error term ∈ i :

[0060] ∈ i =Dist(F i ,F′ i );

[0061] Where Dist(·) is the fingerprint distance calculation function, if the structural consistency error term ∈ i ≤θ, then the grammatical structure check is judged to have passed, and θ is the structural consistency threshold;

[0062] S45. If the syntax structure check passes, the i-th transparent ciphertext packet is marked as valid data and sent to the subsequent decryption process; if the structural consistency error term ∈ i >θ, then the structure error alarm is triggered and the network node link path N is recorded. i Abnormal node identification in And perform error notification and isolation processing operations.

[0063] Optionally, the S5 includes the following steps:

[0064] S51. Under the premise of passing the syntax structure verification, the transparent ciphertext package C i Authorization decryption information metadata M i Conduct legality verification;

[0065] S52. Under the premise that the legal authorization conditions are met, the reversible recovery key index value K recorded in the authorization decryption information metadata is used. rev , in the reversible hash decryption keystore Perform index matching in and determine the reversible hash decryption key K corresponding to the index value dec,i ,The reversible hash decryption key is used as one of the core key parameters for the private corpus decryption operation;

[0066] S53. Use the preset reversible hash decryption algorithm R dec (·), and three inputs are used to participate in the decryption calculation, namely: the ciphertext private corpus c i , one-time symmetric scrambling key K enc and the reversible hash decryption key K dec,i , the ciphertext private corpus c is decrypted by reversible hash algorithm i Perform decryption to restore the original private corpus text The original private corpus text is the standard text format of the corpus before encryption;

[0067] S54. Original private corpus text restored based on decryption Combined with the standardized semantic tag set generated in the initial structured processing stage of the private corpus Reconstruct structured private corpus S i , where the structured private corpus consists of the corpus text And the corresponding semantic label set composition.

[0068] Optionally, the legitimacy check is to verify whether the identity identifier of the current receiving subject is equal to the authorization subject identifier AuthID i , and verify whether the current receiving time is not later than the authorization decryption validity period ExpTime i If the identity of the receiving subject is consistent and the current receiving time is less than or equal to the authorization decryption validity period, it is determined to be legal authorization, passes the authorization verification, and continues the ciphertext private corpus decryption process.

[0069] The beneficial effects of the present invention are:

[0070] The present invention introduces a composite syntactic tree modeling mechanism, combines component syntactic analysis with dependency graph fusion to generate a high-fidelity grammatical structure model, effectively improving the ability to ensure the structural consistency of the corpus during the transmission process. The proposed structural collaboration scoring function and fusion reliability coefficient can dynamically determine the fusion priority of components and dependency structures, and embed a contextual semantic bridging mechanism in the syntactic fusion process to realize the construction of logical adhesion paths between corpora, thereby maintaining dual consistency of structure and semantics during cross-platform restoration.

[0071] The present invention proposes a reversible grammatical fingerprint generation method of path fusion representation + structural weight hashing, which significantly enhances the data integrity verification capability of structural encryption. By performing root-leaf traversal on each node in the composite syntax tree and constructing a path encoding sequence, and then combining semantic sensitivity and structural depth to calculate the weight, and generating a stable mapping through a hash function, it finally forms highly unique structural fingerprint data. It not only participates in ciphertext package encapsulation as a ciphertext binding element, but is also used for structural consistency reconstruction and fingerprint difference verification after transmission, and can accurately locate tampered nodes and semantic breakpoints.

[0072] The present invention implements a reversible hash encryption mechanism driven by structural semantic weights. By dynamically coupling syntactic structure weights with scrambling parameters, high-intensity, low-redundancy encrypted transmission of private corpus is achieved. The encryption path encoding and structural weight joint mapping mechanism enables the same semantic text to generate completely different ciphertext representations under different structural expressions, effectively improving the corpus's anti-reconstruction and anti-frequent attack capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0074] Figure 1 This is a flowchart of a platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption proposed by the present invention;

[0075] Figure 2 This is a schematic diagram of the syntactic analysis process of structured private corpus in a platform private corpus transparent transmission method based on syntactic tree matching and reversible hash encryption proposed by the present invention;

[0076] Figure 3 This is a schematic diagram of the process of generating structural fingerprints and constructing structural verification tags in a platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption proposed by the present invention;

[0077] Figure 4 This is a simplified diagram of the algorithm features of the reversible hash algorithm in the platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption proposed by the present invention. DETAILED DESCRIPTION

[0078] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0079] refer to Figures 1-4 A platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption includes the following steps:

[0080] S1. Obtain the private corpus text to be transmitted, and standardize the private corpus text to generate structured private corpus;

[0081] S2. Use natural language processing technology to perform syntactic analysis on the structured private corpus and construct a complete syntactic tree model of the private corpus. The complete syntactic tree model reflects the grammatical components, hierarchical relationships, and logical structure of the private corpus.

[0082] S3. Generate grammatical structure verification information based on the syntactic tree model to form structural fingerprint data. Perform reversible hash encryption on the structured private corpus. Scramble the private corpus using a preset reversible hash encryption algorithm. Simultaneously, embed authorization decryption information during the encryption process to form ciphertext private corpus. Encapsulate the structural fingerprint data and the ciphertext private corpus to construct a transparent transmission data packet.

[0083] S4. Transmit transparent data packets between heterogeneous platforms using multi-node links for data transfer. On the target platform, the received transparent data packets are parsed, structural fingerprint data in the transparent data packets is extracted and verified, and the consistency of the syntactic tree model of the private corpus with the original syntactic structure is verified.

[0084] S5. After the structure verification is consistent, and under the conditions of legal authorization, the ciphertext private corpus in the transparent data packet is restored using the preset reversible hash decryption operation to recover the original structured private corpus.

[0085] This paper proposes a reversible hash encryption method based on a structural fingerprint embedding mechanism for scrambling and structural restoration of private corpora during inter-platform transmission. Compared to traditional symmetric encryption methods (AES and DES), the reversible hash encryption algorithm of this paper offers significant advantages in encryption structure design, decryption authorization control, semantic consistency assurance, and corpus verification traceability, achieving the technical goal of better structural fidelity and authorized restoration.

[0086] 1. Key differences in structural design

[0087] The closed structure of traditional symmetric encryption: Common symmetric encryption methods such as AES-128 and AES-256 are closed bit operation algorithms based on keys and groups. Their encryption operations do not distinguish between plaintexts and do not consider the semantic structure, syntactic hierarchy, or corpus logic of the plaintext. Once the plaintext is encrypted, the original information cannot be restored unless the exact same key is used for decryption. The intermediate process is also undebugable and unverifiable.

[0088] The structural fingerprint driven reversible hash structure of the present invention: The present invention embeds the structural fingerprint data (by the syntax tree path feature ψ ij , node weight w ij , context structure adhesion information Calculate the generated hash value F i ), and uses this fingerprint to participate in the hash perturbation function control process, making the encryption process sensitive to the structural characteristics of the corpus, forming a structure-related perturbation trajectory. This design makes:

[0089] Fine-tuning the structure of the same corpus (such as replacing a verb) will result in a different structural fingerprint, and thus a different scrambling path; even if the same key is used, encrypting corpora with different structures will not produce the same ciphertext.

[0090] 2. Differences in Decryption Controllability and Authorization Mechanisms

[0091] The "decryption is restoration" model of traditional symmetric encryption: Symmetric encryption is usually based on static key pairs, and authorization control is completed through key distribution. Once the key is leaked or the system is out of control, any holder can decrypt it. There is no way to control access time and access subjects, and it is impossible to audit key usage behavior.

[0092] The authorization and decryption metadata mechanism of the present invention: The present invention adopts independent authorization information metadata M i ={K rev ,ExpTime i ,AuthID i} and compare it with the structure fingerprint tag Tag in the ciphertext bag i After receiving the data, platform B will only allow decryption if the following conditions are met:

[0093] Authorization ID is consistent (the current identity matches AuthID i );

[0094] The current timestamp is less than or equal to the validity period ExpTime i ;

[0095] Decryption key K dec,i Can restore index K from rev Correctly detected.

[0096] This mechanism realizes the structural verifiability, identity controllability and time revocability of decryption behavior.

[0097] 3. The difference between encryption reducibility and structural fidelity

[0098] The irreversible structural verification of traditional symmetric encryption: In the AES method, the encrypted content is a fully encrypted text block, and the semantic structure, word segmentation, and syntactic relationships are all lost. Even if decryption is successful, it is impossible to determine whether the content has been replaced, cropped, or misplaced during transmission because the ciphertext itself does not carry structural information.

[0099] The "structural consistency driven decryption" mechanism of the present invention: In the present invention, the structural fingerprint verification must be completed before decryption, that is, the structural fingerprint F' of the local structure is compared. i With the structural fingerprint F provided in the ciphertext package i , only when the error ∈ i =Dist(F i ,F′ i ) is less than the set threshold θ. This design requires that the corpus must undergo a strict structural consistency check before being restored, effectively preventing the ciphertext from being tampered with, inserted, trimmed, or replaced midway.

[0100] 4. Comparison of Traceability and Anomaly Identification Capabilities

[0101] Traditional symmetric encryption lacks contextual awareness: When multiple pieces of data are transmitted continuously, symmetric encryption processes only a single piece of data, and encryption and decryption are independent of each other. If there are issues such as misalignment of data, line breaks across packets, or truncation, these issues cannot be directly identified through encryption.

[0102] The present invention introduces a context structure bridging mechanism: a set of adhesion paths is constructed between each corpus syntax tree model. This allows the system to perceive the continuous structural consistency between corpora. If a ciphertext is interrupted or misaligned, the context bridge path will fail, causing the structural fingerprint verification to fail. The system can proactively block decryption and mark transmission abnormal nodes, forming a traceable and locatable transmission link security mechanism.

[0103] By comparing the differences in structural design and operating mechanisms, it can be seen that the reversible hash encryption method of the present invention has the following technical effects, which is a substantial improvement over the traditional encryption mechanism:

[0104] Structural interpretability: The encryption process incorporates the structural information of private corpus, improving the ability to restore the semantics of the corpus;

[0105] Controllable decryption: Introducing an independent authorization mechanism and structural fingerprint verification to ensure compliance of corpus transmission;

[0106] Verification and traceability: Based on structural fingerprints and context paths, cross-platform integrity verification and exception tracking can be performed;

[0107] Decryption and restoration: This solves the irreversibility problem of traditional hashing and supports original text recovery under the premise of security.

[0108] Therefore, the reversible hash encryption method proposed in the present invention not only makes up for the shortcomings of traditional symmetric encryption in structural expression, controllable authorization and decryption tracing, but also demonstrates the innovative advantages of high security, high consistency and strong adaptability in the actual cross-platform private corpus transmission process.

[0109] In this embodiment, S1 includes the following steps:

[0110] S11. Obtain the private corpus text to be transmitted and construct the original private corpus set T, wherein the original private corpus set T consists of multiple private corpus texts to be transmitted. Let the i-th private corpus text be t i , the total number of private corpus texts in the private corpus set is N;

[0111] S12. For each private corpus text t in the original private corpus set T i Perform character encoding standardization processing to uniformly convert the original character encoding format into a standard character encoding format to obtain a private corpus set after encoding standardization;

[0112] S13. Performing text format standardization on the encoded and standardized private corpus set, including removing redundant line breaks, spaces, and non-standard punctuation, and unifying paragraph structure and punctuation style, to obtain a unified format private corpus set;

[0113] S14. Perform semantic label standardization on the private corpus set after format unification, extract the named entities, keywords and grammatical markers contained in each private corpus text, and perform semantic label unified encoding to obtain the semantic label set L corresponding to the i-th private corpus text i and the jth semantic tag l in the semantic tag set ij , forming a semantic annotation mapping set T L ;

[0114] S15. Map the semantic annotations to the set T L Constructed as a structured private corpus set T S .

[0115] In this embodiment, the structured private corpus set T L The i-th structured private corpus in is S i , structured private corpus consists of private corpus text with unified format With a standardized set of semantic tags composition.

[0116] In this embodiment, S2 includes the following steps:

[0117] S21. For the structured private corpus set T S Each structured private corpus S in i Perform component syntax analysis and build a component syntax tree in, is the component syntax tree Q i The set of grammatical structure component nodes in represents the kth grammatical structure component node, is the component syntax tree Q i The edge set in represents the hierarchical relationship between grammatical structures. If the grammatical structure is composed of nodes Is the parent node, another grammatical structure component node To modify its child nodes, there is a directed edge

[0118] S22. Private Corpus Text Synchronously perform dependency syntax analysis and build a dependency graph Among them, the node set is the set of vocabulary nodes in the dependency graph, Represents private corpus text The kth vocabulary node in , is the set of dependency edges between vocabulary nodes, Represents a vocabulary node With vocabulary nodes There are dependency relationships between them, including subject and predicate, verb and object, and modification;

[0119] S23. After the component syntax tree and dependency graph structures are built, in order to determine whether to perform the structure fusion operation, the structure coordination scoring function is calculated:

[0120]

[0121] Among them, δ(·) represents the structural consistency function, and Map(·) represents the structural node mapping function;

[0122] The fusion confidence coefficient is constructed based on the value of the collaborative scoring function and the semantic label scale:

[0123]

[0124] Among them, λ1,λ2 are adjustment parameters, σ(·) is the Sigmoid function, if the fusion reliability coefficient η i>0.5, then the component syntax tree Q i With the dependency graph G i Perform structural fusion to generate a composite syntax tree model P i =(V i ,E i ), otherwise, the structural model with high collaborative score is retained as the composite syntax tree model, and the composite node set V of the composite syntax tree model is obtained. i With the composite edge set E i ;

[0125] S24. Based on the composite syntax tree model P i , for each composite node v ij ∈V i Calculate the structure depth d ij and semantic sensitivity markers ij ∈{0,1}, and according to the fusion reliability coefficient η i Constructing node semantic-structural weight function:

[0126]

[0127] Among them, α and β are structural and semantic adjustment coefficients, w ij Represents a composite node v ij The semantic-structural joint weight of

[0128] S25. In the composite syntax tree model P i Perform root-leaf pre-order traversal in the process and construct the path encoding sequence φ ij , and based on the composite node weight w ij Construct the fusion path representation:

[0129] ψ ij =φ ij ⊕Hash(w ij );

[0130] Among them, ψ ij represents the fusion path representation after the fusion of path encoding and structural weight, Hash(·) is the structural stable mapping function, and the path fusion feature set is recorded as Ψ i ={ψ ij};

[0131] S26. For continuous structured private corpus pairs (S i ,S i+1 ), constructing a contextual bridging edge set between its composite syntax tree models:

[0132]

[0133] Among them, ContextSim(·) represents the contextual semantic similarity function based on word embedding, and γ is the set threshold;

[0134] Construct a set of structural adhesion paths within a corpus segment based on the set of contextual bridging edges:

[0135]

[0136] in, represents the set of structural adhesion paths between the i-th to the k-th private corpus;

[0137] S27. Composite grammar model after context enhancement Node semantic-structural weight set {w ij}、Path fusion feature set Ψ i ={ψ ij} are stored together in the context-enhanced composite syntax tree model

[0138] In this embodiment, the composite node set and composite edge set of the composite syntax tree model are defined as:

[0139]

[0140] in, Represents a mapping function based on Aligned connecting edges are created between component nodes and dependent nodes.

[0141] In this embodiment, S3 includes the following steps:

[0142] S31. Context-enhanced composite syntax tree model Each composite node v ij Traverse and represent the corresponding fusion path ψ ij and node semantic-structural weight w ij After connecting in sequence, input the one-way hash function Hash to obtain the node structure hash value h ij , and collected into a node structure hash value set {h ij};

[0143] S32. Based on the node semantic-structural weight w ij As a weighted coefficient, the node structure hash value set {h ij} perform weighted summation, and then input the summation result into the one-way hash function Hash to obtain the structured private corpus S i One-to-one corresponding structural fingerprint data F i ;

[0144] S33. The structural fingerprint data F iBridge path collection with context The hash results are sequentially connected and fed into the hash function based on the key K struct The message authentication code function HMAC is used to obtain the syntax structure verification tag Tag i ;

[0145] S34. Use the preset reversible hash encryption algorithm R enc With a one-time symmetric scrambling key K enc , reversible recovery key index value K rev And the corpus number i is input, for the private corpus text Perform reversible scrambling to obtain the ciphertext private corpus c i ;

[0146] S35. Constructing the authorization decryption information metadata M i , the authorized decryption information metadata contains the reversible recovery key index value K rev , Authorization decryption validity period ExpTime i and the authorization subject identifier AuthID i ;

[0147] S36. The ciphertext private corpus c i , grammatical structure check tag i and authorization decryption information metadata M i Combine to form transparent ciphertext package C i .

[0148] In this embodiment, S4 includes the following steps:

[0149] S41. The transparent ciphertext packet set C = {C i Data is transferred through multiple network node links between heterogeneous platforms, and the network node link path is defined as an ordered node sequence:

[0150] N i ={n1→n2→…→n m};

[0151] Among them, N i Indicates the transmission of the i-th transparent ciphertext package C i The network node link path traversed, n j represents the jth transmission node, and m is the total number of transmission nodes;

[0152] S42. Receive network node n on the target platform m For transparent ciphertext package C i Parse and separate the ciphertext private corpus c i , grammatical structure check tag i and authorization decryption information metadata M i;

[0153] S43. Verify the key K based on the structure struct Check the structure tag i Perform message authentication code verification and extract the structural fingerprint data F contained therein i Bridge path collection with context The hash representation of the node is used to perform the structural consistency reconstruction process, from the node semantic-structural weight set {w′ ij}、Fusion path representation set {ψ′ ij} and context-enhanced syntactic models , calculate the reconstructed structural fingerprint data:

[0154]

[0155] S44. Compare the original value of the structural fingerprint data with the reconstructed value of the structural fingerprint data and calculate the structural consistency error term ∈ i :

[0156] ∈ i =Dist(F i ,F′ i );

[0157] Where Dist(·) is the fingerprint distance calculation function, if the structural consistency error term ∈ i ≤θ, then the grammatical structure check is judged to have passed, and θ is the structural consistency threshold;

[0158] S45. If the syntax structure check passes, the i-th transparent ciphertext packet is marked as valid data and sent to the subsequent decryption process; if the structural consistency error term ∈ i >θ, then the structure error alarm is triggered and the network node link path N is recorded. i Abnormal node identification in And perform error notification and isolation processing operations.

[0159] In this embodiment, S5 includes the following steps:

[0160] S51. Under the premise of passing the syntax structure verification, the transparent ciphertext package C i Authorization decryption information metadata M i Conduct legality verification;

[0161] S52. Under the premise that the legal authorization conditions are met, the reversible recovery key index value K recorded in the authorization decryption information metadata is used. rev , in the reversible hash decryption keystore Perform index matching in and determine the reversible hash decryption key K corresponding to the index value dec,i,The reversible hash decryption key is used as one of the core key parameters for the private corpus decryption operation;

[0162] S53. Use the preset reversible hash decryption algorithm R dec (·), and three inputs are used to participate in the decryption calculation, namely: the ciphertext private corpus c i , one-time symmetric scrambling key K enc and the reversible hash decryption key K dec,i , the ciphertext private corpus c is decrypted by reversible hash algorithm i Perform decryption to restore the original private corpus text The original private corpus text is the standard text format of the corpus before encryption;

[0163] S54. Original private corpus text restored based on decryption Combined with the standardized semantic tag set generated in the initial structured processing stage of the private corpus Reconstruct structured private corpus S i , where the structured private corpus consists of the corpus text And the corresponding semantic label set composition.

[0164] In this embodiment, the legitimacy check is to verify whether the identity of the current receiving subject is equal to the authorization subject identifier AuthID i , and verify whether the current receiving time is not later than the authorization decryption validity period ExpTime i If the identity of the receiving subject is consistent and the current receiving time is less than or equal to the authorization decryption validity period, it is determined to be legal authorization, passes the authorization verification, and continues the ciphertext private corpus decryption process.

[0165] Example 1:

[0166] At 10:45 AM on December 16, 2024, the knowledge corpus management system on Platform A initiated a private corpus transmission task to the natural language analysis center on Platform B. The target was a batch of text corpus data containing task dialogues, agreement terms, and the original text of the technical agreement. The original corpus totaled 2,624 items, each averaging 256 bytes in length, for a total data volume of approximately 670KB. The data sensitivity was classified as "Platform Internal Confidentiality."

[0167] After the task is initiated, Platform A first performs standardization on the original corpus using the method described in this invention. The system uniformly converts the source text encoding format to UTF-8, removes abnormal characters (such as "\u200b" and redundant line breaks), adjusts the paragraph separation format, and uniformly labels the terms and named entities in the corpus content. Take one of the corpus as an example:

[0168] Original text: "The initial review of the agreement template V1.3 has been completed, including the access rights control strategy and API interface signature verification process description."

[0169] The standardized corpus after processing is: formatted text (encoding conversion + punctuation standardization + entity annotation):

[0170] "The preliminary review of [Agreement Template] V1.3 has been completed, including [Access Control Policy] and [API Interface Signature Verification Process Instructions]."

[0171] Subsequently, the system performs component syntactic analysis and dependency structure modeling on the standardized corpus set to construct a syntactic tree structure model. In the text sample numbered t0846, the system constructs a complete component syntactic tree Q 0846 With the dependency graph G 0846 , the syntactic fusion score is S 0846 =0.9321, the fusion reliability coefficient η is calculated by the Sigmoid function 0846 =0.8127, entering the structural fusion stage, successfully building a composite syntax tree model P 0846 , containing a total of 46 nodes and a maximum level depth of 5.

[0172] After the structure fusion is completed, the system extracts the path encoding sequence φ according to the path fusion strategy 0846,j , construct the fusion path feature ψ 0846,j , and combined with the node weight w 0846,j Construct a set of structural fingerprint hash values, and calculate the structural fingerprint data hash as: F 0846 =0x9e7b2a6e5b8840f…;

[0173] This fingerprint is used as the basis for verifying the complete syntactic structure and is written into the structure verification tag. At the same time, a reversible hash scrambling encryption operation is performed on the corpus text. The input parameters include: Source corpus text: Scrambling key: K enc =0x80a4f…; Reversible recovery key index: K rev =id-13; Authorization ID: AuthID user =001129-A; Decryption validity period: ExpTime 0846 =2024 / 12 / 18-00:00:00;

[0174] Finally, the transparent ciphertext package C is generated 0846 The data size is 2.2KB. The system packages all transparent ciphertext packets and performs encrypted transmission through the node link {nodeA1→nodeA2→nodeB1→nodeB2}. During the transmission process, the link monitoring module continuously tracks the status of the relay node.

[0175] At 10:52 on December 16, 2024, Platform B received the 38th ciphertext package C 0846 , start to perform structural fingerprint verification. The system parses the structural fingerprint hash F from the ciphertext package 0846 At the same time, based on the local restoration model, the syntactic structure and fusion path features are reconstructed to calculate the reconstructed fingerprint hash F′ 0846 , comparison function Dist(F 0846 ,F′ 0846 )=1.79×10 -8 , which is much smaller than the set threshold θ=10 -6 , the structure verification passed.

[0176] The system continues to read the authorization information and finds that the authorization subject ID is "001129-A", which is consistent with the current access requester ID; the current timestamp is "2024 / 12 / 1610:52:07", which is still within the authorization validity period, and the authorization verification is successful. The system retrieves the reversible hash decryption key K from the local key index pool dec,0846 , with the scrambling key K enc The data was input into the decryption engine and successfully decrypted and restored the ciphertext private corpus. The decryption took 14.7ms, and the output structured corpus text was completely consistent with the semantic tag set.

[0177] For the entire transparent transmission process, the project team simultaneously compared the performance of the traditional AES encryption scheme and the method of the present invention under the same conditions. Some key performance indicators are as follows:

[0178] Table 1 Performance of the traditional AES encryption scheme and the method of the present invention under the same conditions

[0179]

[0180] The above data demonstrates that this invention significantly outperforms traditional encrypted transmission solutions in terms of transmission accuracy, structural consistency, corpus integrity, and anomaly prevention and control. During the actual deployment test cycle, the system processed a total of 16,700 ciphertext data packets without any structural misalignment or irreversible data. The overall operation was stable, the process was closed-loop, and the results were reliable, validating the engineering usability and algorithmic innovation of this invention in private corpus transparent transmission scenarios.

[0181] The present invention introduces a composite syntactic tree modeling mechanism, combines component syntactic analysis with dependency graph fusion to generate a high-fidelity grammatical structure model, effectively improving the ability to ensure the structural consistency of the corpus during the transmission process. The proposed structural collaboration scoring function and fusion reliability coefficient can dynamically determine the fusion priority of components and dependency structures, and embed a contextual semantic bridging mechanism in the syntactic fusion process to realize the construction of logical adhesion paths between corpora, thereby maintaining dual consistency of structure and semantics during cross-platform restoration.

[0182] The present invention proposes a reversible grammatical fingerprint generation method of path fusion representation + structural weight hashing, which significantly enhances the data integrity verification capability of structural encryption. By performing root-leaf traversal on each node in the composite syntax tree and constructing a path encoding sequence, and then combining semantic sensitivity and structural depth to calculate the weight, and generating a stable mapping through a hash function, it finally forms highly unique structural fingerprint data. It not only participates in ciphertext package encapsulation as a ciphertext binding element, but is also used for structural consistency reconstruction and fingerprint difference verification after transmission, and can accurately locate tampered nodes and semantic breakpoints.

[0183] The present invention implements a reversible hash encryption mechanism driven by structural semantic weights. By dynamically coupling syntactic structure weights with scrambling parameters, high-intensity, low-redundancy encrypted transmission of private corpus is achieved. The encryption path encoding and structural weight joint mapping mechanism enables the same semantic text to generate completely different ciphertext representations under different structural expressions, effectively improving the corpus's anti-reconstruction and anti-frequent attack capabilities.

[0184] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption, characterized in that: The steps include: S1. Obtain the private corpus text to be transmitted, and standardize the private corpus text to generate a structured private corpus; S2. Use natural language processing technology to perform syntactic analysis on the structured private corpus and construct a complete syntactic tree model of the private corpus. The complete syntactic tree model reflects the grammatical components, hierarchical relationships, and logical structure of the private corpus. S3. Generate grammatical structure verification information based on the syntactic tree model to form structural fingerprint data, implement reversible hash encryption on the structured private corpus, scramble the private corpus using a preset reversible hash encryption algorithm, and embed authorization decryption information during the encryption process to form ciphertext private corpus. Encapsulate the structural fingerprint data and the ciphertext private corpus to construct a transparent transmission data packet; S4. Transmit transparent data packets between heterogeneous platforms using multi-node links for data transfer. On the target platform, the received transparent data packets are parsed, structural fingerprint data in the transparent data packets is extracted and verified, and the consistency of the syntactic tree model of the private corpus with the original syntactic structure is verified. S5. After the structure verification is consistent, and under the conditions of legal authorization, the ciphertext private corpus in the transparent data packet is restored using the preset reversible hash decryption operation to recover the original structured private corpus.

2. A platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption according to claim 1, characterized in that: Said S1 comprises the following steps: S11. Obtain the private corpus text to be transmitted and construct the original private corpus set T, wherein the original private corpus set T consists of multiple private corpus texts to be transmitted. Let the i-th private corpus text be t i , the total number of private corpus texts in the private corpus set is N; S12. For each private corpus text t in the original private corpus set T i Perform character encoding standardization processing to uniformly convert the original character encoding format into a standard character encoding format to obtain a private corpus set after encoding standardization; S13. Performing text format standardization on the encoded and standardized private corpus set, including removing redundant line breaks, spaces, and non-standard punctuation, and unifying paragraph structure and punctuation style, to obtain a unified format private corpus set; S14. Perform semantic label standardization on the private corpus set after format unification, extract the named entities, keywords and grammatical markers contained in each private corpus text, and perform semantic label unified encoding to obtain the semantic label set L corresponding to the i-th private corpus text i and the jth semantic tag l in the semantic tag set ij , forming a semantic annotation mapping set T L ; S15. Map the semantic annotations to the set T L Constructed as a structured private corpus set T S .

3. A platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption according to claim 2, characterized in that: The structured private corpus set T L The i-th structured private corpus in is S i , structured private corpus consists of private corpus text with unified format With a standardized set of semantic tags composition.

4. The platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption according to claim 2 is characterized in that: The S2 comprises the following steps: S21. For the structured private corpus set T S Each structured private corpus S in i Perform component syntax analysis and build a component syntax tree in, is the component syntax tree Q i The set of grammatical structure component nodes in represents the kth grammatical structure component node, is the component syntax tree Q i The edge set in represents the hierarchical relationship between grammatical structures. If the grammatical structure is composed of nodes Is the parent node, another grammatical structure component node To modify its child nodes, there is a directed edge S22. Private Corpus Text Synchronously perform dependency syntax analysis and build a dependency graph Among them, the node set is the set of vocabulary nodes in the dependency graph, Represents private corpus text The kth lexical node in , is the set of dependency edges between vocabulary nodes, Represents a vocabulary node With vocabulary nodes There are dependency relationships between them, including subject and predicate, verb and object, and modification; S23. After the component syntax tree and dependency graph structures are built, in order to determine whether to perform the structure fusion operation, the structure coordination scoring function is calculated: Among them, δ(·) represents the structural consistency function, and Map(·) represents the structural node mapping function; Construct the fusion confidence coefficient η based on the value of the collaborative scoring function and the semantic label scale i , if the fusion reliability coefficient η i >0.5, then the component syntax tree Q i With the dependency graph G i Perform structural fusion to generate a composite syntax tree model P i =(V i ,E i ), otherwise, the structural model with high collaborative score is retained as the composite syntax tree model, and the composite node set V of the composite syntax tree model is obtained. i With the composite edge set E i ; S24. Based on the composite syntax tree model P i , for each composite node v ij ∈V i Calculate the structure depth d ij and semantic sensitivity markers ij ∈{0,1}, and according to the fusion reliability coefficient η i Construct node semantic-structural weight function w ij ; S25. In the composite syntax tree model P i Perform root-leaf pre-order traversal in the process and construct the path encoding sequence φ ij , and based on the composite node weight w ij Construct the fusion path representation: Among them, ψ ij represents the fusion path representation after the fusion of path encoding and structural weight, Hash(·) is the structural stable mapping function, and the path fusion feature set is recorded as Ψ i ={ψ ij }; S26. For continuous structured private corpus pairs (S i ,S i+1 ), constructing a contextual bridging edge set between its composite syntax tree models Constructing a set of structural adhesion paths within a corpus segment based on a set of contextual bridging edges S27. Composite grammar model after context enhancement Node semantic-structural weight set {w ij }、Path fusion feature set Ψ i ={ψ ij } are stored together in the context-enhanced composite syntax tree model 5. The platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption according to claim 3 is characterized in that: The composite node set and composite edge set of the composite syntax tree model are defined as: in, Represents a mapping function based on Aligned connecting edges are created between component nodes and dependent nodes.

6. The platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption according to claim 4 is characterized in that: The S3 includes the following steps: S31. Context-enhanced composite syntax tree model Each composite node v ij Traverse and represent the corresponding fusion path ψ ij and node semantic-structural weight w ij After connecting in sequence, input the one-way hash function Hash to obtain the node structure hash value h ij , and collected into a node structure hash value set {h ij }; S32. Based on the node semantic-structural weight w ij As a weighted coefficient, the node structure hash value set {h ij } perform weighted summation, and then input the summation result into the one-way hash function Hash to obtain the structured private corpus S i One-to-one corresponding structural fingerprint data F i ; S33. The structural fingerprint data F i Bridge path collection with context The hash results are sequentially connected and fed into the hash function based on the key K struct The message authentication code function HMAC is used to obtain the syntax structure verification tag Tag i ; S34. Use the preset reversible hash encryption algorithm R enc With a one-time symmetric scrambling key K enc , reversible recovery key index value K rev And the corpus number i is input, for the private corpus text Perform reversible scrambling to obtain the ciphertext private corpus c i ; S35. Constructing the authorization decryption information metadata M i , the authorized decryption information metadata contains the reversible recovery key index value K rev , Authorization decryption validity period ExpTime i and the authorization subject identifier AuthID i ; S36. The ciphertext private corpus c i , grammatical structure check tag i and authorization decryption information metadata M i Combine to form transparent ciphertext package C i .

7. The platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption according to claim 5 is characterized in that: The S4 comprises the following steps: S41. The transparent ciphertext packet set C = {C i Data is transferred through multiple network node links between heterogeneous platforms, and the network node link path is defined as an ordered node sequence N i ; S42. Receive network node n on the target platform m For transparent ciphertext package C i Parse and separate the ciphertext private corpus c i , grammatical structure check tag i and authorization decryption information metadata M i ; S43. Verify the key K based on the structure struct Check the structure tag i Perform message authentication code verification and extract the structural fingerprint data F contained therein i Bridge path collection with context The hash representation of the node is used to perform the structural consistency reconstruction process, from the node semantic-structural weight set {w′ ij }、Fusion path representation set {ψ′ ij } and context-enhanced syntactic models , calculate the reconstructed structural fingerprint data: S44. Compare the original value of the structural fingerprint data with the reconstructed value of the structural fingerprint data and calculate the structural consistency error term ∈ i : ∈ i =Dist(F i ,F′ i ); Where Dist(·) is the fingerprint distance calculation function, if the structural consistency error term ∈ i ≤θ, then the grammatical structure check is judged to have passed, and θ is the structural consistency threshold; S45. If the syntax structure check passes, the i-th transparent ciphertext packet is marked as valid data and sent to the subsequent decryption process; if the structural consistency error term ∈ i >θ, then the structure error alarm is triggered and the network node link path N is recorded. i Abnormal node identification in And perform error notification and isolation processing operations.

8. The platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption according to claim 6 is characterized in that: The S5 comprises the following steps: S51. Under the premise of passing the syntax structure verification, the transparent ciphertext package C i Authorization decryption information metadata M i Conduct legality verification; S52. Under the premise that the legal authorization conditions are met, the reversible recovery key index value K recorded in the authorization decryption information metadata is used. rev , in the reversible hash decryption keystore Perform index matching in and determine the reversible hash decryption key K corresponding to the index value dec,i ,The reversible hash decryption key is used as one of the core key parameters for the private corpus decryption operation; S53. Use the preset reversible hash decryption algorithm R dec (·), and three inputs are used to participate in the decryption calculation, namely: the ciphertext private corpus c i , one-time symmetric scrambling key K enc and the reversible hash decryption key K dec,i , the ciphertext private corpus c is decrypted by reversible hash algorithm i Perform decryption to restore the original private corpus text The original private corpus text is the standard text format of the corpus before encryption; S54. Original private corpus text restored based on decryption Combined with the standardized semantic tag set generated in the initial structured processing stage of the private corpus Reconstruct structured private corpus S i , where the structured private corpus consists of the corpus text And the corresponding semantic tag set composition.

9. The platform private corpus transparent transmission method based on syntax tree matching and reversible hash encryption according to claim 8 is characterized in that: The legitimacy check is to verify whether the identity of the current receiving subject is equal to the authorized subject identifier AuthID i , and verify whether the current receiving time is not later than the authorization decryption validity period ExpTime i If the identity of the receiving subject is consistent and the current receiving time is less than or equal to the authorization decryption validity period, it is determined to be legal authorization, passes the authorization verification, and continues the ciphertext private corpus decryption process.

Images