Blockchain digital copyright encryption management method supporting multi-keyword search

Abstract

This invention discloses a blockchain-based digital copyright encryption management method supporting multi-keyword search, relating to the fields of blockchain and cryptography. The method involves receiving digital works, extracting a keyword set from copyright metadata fields, and calculating the hash value of the work content. The method uses the registration time sequence value of the consortium blockchain as the registration time sequence anchor point. It then symmetrically encrypts and randomizes the keyword set to generate ciphertext keywords, and constructs the registration time sequence anchor point as a monotonic time sequence commitment embedded in the ciphertext inverted index to obtain time sequence anchored ciphertext index items. These index items are stored in an off-chain ciphertext index library, and the time sequence commitment root and the work content hash value are stored on-chain. Based on the on-chain time sequence commitment root, a priori-aware hierarchical trapdoor is generated. A single ciphertext concatenation match is performed on the ciphertext index library, returning the copyright registration record and priori proof that meets the multi-keyword combination conditions and has the earliest registration time. The registration time sequence is then reversed into a partial-order skeleton of the ciphertext index, enabling a one-time output of combined queries and priori proof without decrypting unselected records.

Description

Technical Field

[0001] This invention relates to the fields of blockchain and cryptography, specifically to a blockchain-based digital copyright encryption management method that supports multi-keyword search. Background Technology

[0002] With the rapid development of the digital content industry, the number of copyright registrations for digital works such as text, images, audio and video, and software has exploded. Traditional centralized copyright registration databases, while handling massive amounts of registration data, have long faced the risks of single points of failure and tampering with ownership information, resulting in insufficient credibility of registration evidence. Blockchain, with its decentralized, immutable, and fully traceable characteristics, provides a new technological path for the reliable preservation of copyright registration evidence. However, directly uploading the complete content of digital works to the blockchain would cause a dramatic expansion of blockchain storage capacity, making it difficult to handle the massive registration demands. Therefore, the industry generally adopts a layered model where the hash value of the work content is uploaded to the blockchain, while registration metadata is managed off-chain. In this model, registration metadata typically includes sensitive information from multiple dimensions such as the work name, work type, creation time, registration time, and rights holder. To protect the privacy of rights holders, it needs to be encrypted before storage and retrieval. However, traditional symmetric encryption schemes require decryption before querying and re-encryption after querying encrypted data, resulting in extremely low query efficiency and high key management risks. Searchable encryption technology was proposed to resolve this contradiction.

[0003] Chinese invention patent application CN117009988A discloses a blockchain-based encrypted data storage and retrieval method. This method deploys a consortium blockchain and smart contracts among members of the supply chain. Data is divided into three levels based on size and security level, using on-chain plaintext storage, on-chain ciphertext storage, and off-chain ciphertext storage respectively. During the data retrieval phase, searchable encryption technology is employed. Users generate search credentials based on keywords and send them to a data sharing platform. The platform probes and matches against the ciphertext, returning data containing the keywords to the user. The server executing the search remains unaware of the ciphertext content. This scheme combines searchable encryption with consortium blockchain notarization, enabling ciphertext retrieval without decryption. However, the encrypted retrieval of this scheme is an existence match based on a single keyword, and it treats the registration time sequence only as a normal metadata field used for display after decryption. The registration time sequence and the encrypted retrieval are two independent steps. When copyright administrators need to further determine which of the multiple encrypted registration records retrieved by combining multiple dimensions such as work name, work type and registration time is registered earlier and thus has prior rights, they must first decrypt all matching results and then compare the registration time sequence one by one in the plaintext field. This process not only reveals the registration time sequence of the unselected records, but also cannot provide prior evidence that can be verified by a third party in the encrypted field, making the multi-dimensional combination query lose its legal effect in the core scenario of copyright confirmation.

[0004] Chinese invention patent CN108846776B discloses a digital copyright protection method based on blockchain technology. This method constructs a decentralized peer-to-peer network, generates user addresses through an encryption verification module, and uses a transaction module to record transaction timestamps to realize copyright registration and transactions, storing blocks in a self-referencing chain structure. However, the timestamps in this scheme are only used to display the generation time of block transactions after decryption, without establishing any connection with encrypted text retrieval or supporting priori determination without decryption. Chinese invention patent CN107220343B discloses a Chinese multi-keyword fuzzy sorting encrypted text search method based on locality-sensitive hashing. This method constructs a Bloom filter as an encrypted index for each document, uses a secure nearest neighbor algorithm to encrypt the index and trapdoors, and returns several documents sorted by similarity score. However, this scheme returns results sorted by content similarity, neither introducing partial order semantics of registration time nor involving blockchain notarization, and cannot return records registered earlier or their priori proof.

[0005] In summary, existing solutions, regardless of whether they employ single-keyword retrieval, multi-keyword similarity ranking, or on-chain timestamp storage, treat the registration time sequence as a regular field to be displayed afterward. They cannot enable multi-keyword combination queries on encrypted copyright registration metadata to directly return the earlier registered record and generate verifiable priority proof in a single ciphertext match without decryption or disclosure of the registration time sequence of the unselected records. Summary of the Invention

[0006] Addressing the core technical bottleneck in existing blockchain digital copyright encrypted management technologies, where registration time is treated as a post-event display field, preventing multi-keyword combination queries of encrypted registration metadata without decryption and the direct return of prior-registered records and verifiable priority proofs, this invention provides a blockchain digital copyright encrypted management method supporting multi-keyword search. By constructing registration time anchors as monotonic time commitments embedded in a ciphertext inverted index and performing a single ciphertext conjoint matching on the ciphertext index library using a priority-aware hierarchical trapdoor, and without decrypting unselected records, this method reverses the registration time from the searched field to a partial order skeleton of the ciphertext index, enabling a one-time output of multi-keyword combination queries of encrypted copyright registration metadata and priority proofs of prior-registered records.

[0007] The technical solution of this invention is as follows: A blockchain-based digital rights management method that supports multi-keyword search includes the following steps: Step S1: Receive the digital work, extract the keyword set of the copyright metadata field of the digital work according to the standardized template, and calculate the content hash value of the digital work. The copyright metadata field includes the work name field, work type field, creation time field, registration time field and rights holder field. Step S2: Obtain the block time sequence value generated by the registration of the digital work in the consortium blockchain, determine the block time sequence value as the registration time sequence anchor point, use a symmetric encryption algorithm and randomization processing to generate ciphertext keywords for the keyword set of the copyright metadata field, and construct the registration time sequence anchor point as a monotonic time sequence commitment embedded in the ciphertext inverted index organized by the ciphertext keywords to obtain the time sequence anchored ciphertext index item; Step S3: Store the time-anchored ciphertext index item in the off-chain ciphertext index library, and synchronize the time commitment root of the time-anchored ciphertext index item and the hash value of the work content on the consortium chain for evidence storage. Step S4: Receive a copyright retrieval request carrying multiple keyword combination conditions, generate a priori-aware hierarchical trapdoor based on the on-chain time commitment root, perform a single ciphertext conjunction matching on the off-chain ciphertext index library, and return the copyright registration record that satisfies the multiple keyword combination conditions and has the earliest registration time. Step S5: Return the priority proof of the copyright registration record.

[0008] Further, the step of constructing the registration time-series anchor point as a monotonically time-series commitment embedded in a ciphertext inverted index organized by the ciphertext keywords includes: taking the registration time-series anchor point as input, generating a time-series commitment value through monotonically accumulating mapping, such that the time-series commitment value corresponding to the earlier registered copyright registration record has a partial order in the ciphertext field before the time-series commitment value corresponding to the later registered copyright registration record; and organizing multiple time-series anchored ciphertext index entries linked by the same ciphertext keyword in the ciphertext inverted index into a time-ordered inverted chain according to the partial order of the time-series commitment values.

[0009] Furthermore, before determining the block time sequence value as the registration time sequence anchor point, the finality confirmation depth of the registration block corresponding to the digital work by the consortium blockchain is obtained. When the finality confirmation depth reaches a preset confirmation depth threshold, the block time sequence value of the registration block is locked as the registration time sequence anchor point, and the finality confirmation depth is recorded as the stability weight of the time sequence commitment value. When generating the prior proof, the stability weight is used as the confidence level of the prior proof.

[0010] Furthermore, the single ciphertext concatenation matching includes: converting the multi-keyword combination conditions into multiple ciphertext sub-trapdoors, locating the temporally ordered inverted index corresponding to each ciphertext sub-trapdoor in the off-chain ciphertext index library; synchronously advancing the cursor along each temporally ordered inverted index in a partially increasing direction according to the temporal commitment value; determining a concatenation hit when the temporal commitment value aligned to the cursor of each temporally ordered inverted index first points to the same copyright registration record; and terminating the advance at the first concatenation hit.

[0011] Furthermore, the generation of the priority proof includes: recording the cursor partial order position of the copyright registration record corresponding to the conjunction hit in each of the temporally ordered inverted chains during the single scan process, combining each of the cursor partial order positions into a temporally ordered rank witness; and generating the priority proof using the temporally ordered rank witness, each of the ciphertext trapdoors, and the on-chain temporal commitment root.

[0012] Furthermore, the query pattern distribution of each of the time-ordered inverted chains in the off-chain encrypted index library is statistically analyzed in historical copyright retrieval requests. When the leakage metric corresponding to the query pattern distribution exceeds a preset leakage threshold, randomized re-embedding is performed on the encrypted keywords in the corresponding time-ordered inverted chains while maintaining the partial order of the time commitment value, and the off-chain encrypted index library is updated.

[0013] The beneficial effects of this invention are as follows: First, by constructing the registration time sequence anchor points as monotonic time sequence commitments embedded in a ciphertext inverted index organized by the ciphertext keywords, the index entries linked by the same ciphertext keyword in the ciphertext inverted index are organized into a time-ordered inverted chain according to the time sequence commitment value, thus achieving a comparable partial order of registration time sequence within the ciphertext domain. In the copyright field, registration time sequence carries the unique legal semantic of prior rights confirmation; its essence is not a scalar field to be retrieved, but a naturally ordered structure. After reversing it from an encrypted ordinary field into the partial order skeleton of the ciphertext inverted index, the order between ciphertext index entries can be compared without decryption, thereby shifting the determination of who registered first from the plaintext domain to the ciphertext domain. Compared to the approach of displaying registration time sequence as a decrypted field, requiring the decryption of all matching results before time sequence comparison, this scheme completes the partial order arrangement within the ciphertext domain. The registration time sequence of unselected records does not need to be decrypted and is therefore not leaked.

[0014] Second, by performing a single ciphertext conjunctive match on the off-chain ciphertext index using the priori-aware hierarchical trapdoor, and synchronously advancing the cursor along each temporally ordered inverted chain in a partially increasing direction according to the temporal commitment value, terminating at the first conjunctive hit, this achieves the direct return of the copyright registration record that satisfies the multi-keyword combination condition and has the earliest registration time in a single scan. Each temporally ordered inverted chain is arranged in ascending order according to the temporal commitment value, and the ascending direction is consistent with the direction of registration time from earliest to latest. Therefore, the first time each cursor aligns with the same record, it represents the earliest registered record under all combination conditions, eliminating the need to retrieve all intersections and then sort and compare them. Compared to the approach of returning the first few results sorted by similarity and then having the user decrypt and compare them, this scheme directly provides the unique earliest record at the time cost of a single conjunctive scan, avoiding the retrieval and decryption of unselected records.

[0015] Third, by recording the cursor partial order positions of the conjunction hit records in each temporally ordered inverted chain during a single scan and combining them into a temporal rank witness, and then generating a priori proof using the temporal rank witness, each ciphertext trapdoor, and the on-chain temporal commitment root, verifiable proof of the priori registration is achieved without decrypting the missing records. The cursor partial order position is a natural intermediate product in the conjunction matching process, which precisely characterizes the rank of the hit record in each inverted chain partial order. Combined with the global partial order anchored by the on-chain temporal commitment root, a third party can recalculate and compare the commitment root without touching the plaintext of any missing record, making the priori conclusion cryptographically verifiable. Compared to the approach of only providing on-chain transaction timestamps for manual comparison after decryption, which cannot provide verifiable priori evidence in the ciphertext field, this scheme makes the priori proof an endogenous component of the retrieval results.

[0016] Fourth, the above three technical features constitute a collaborative closed loop of partial order construction of encrypted index, partial order-aware conjunctive retrieval, and prior evidence of single matching, which produces a synergistic effect that exceeds the sum of the individual effects of each feature. Specifically: if there is only partial order construction without partial order-aware conjunctive retrieval, the partial order skeleton cannot be utilized in multi-dimensional combined queries; if there is only conjunctive retrieval without partial order construction, the conjunctive result still needs to be decrypted and compared in terms of time sequence; only when partial order construction, partial order-aware conjunctive, and prior evidence are coupled can a single ciphertext match simultaneously satisfy the triple constraints of "earliest in time sequence and verifiable prior". Furthermore, the finality confirmation depth stabilization mechanism, while locking the registration time sequence anchor point, reuses the confirmation depth as the confidence level of prior proof, unifying the ability to resist consortium chain forks and rollbacks with the credibility of prior evidence under the same parameter; the query pattern indistinguishable order-preserving randomized re-embedding mechanism eliminates the query pattern leakage in the retrieval and retrieval stages while maintaining the partial order of the time sequence commitment value, thereby upgrading the partial order index from a structure that only exposes the access pattern to a structure that satisfies forward privacy constraints. Attached Figure Description

[0017] Figure 1 This is an overall flowchart of the blockchain digital copyright encryption management method that supports multi-keyword search according to the present invention; Figure 2 This is a flowchart of the copyright metadata standardization extraction process in step S1 of the present invention; Figure 3 This is a flowchart of the time-anchored ciphertext index construction process in step S2 of the present invention; Figure 4 This is a flowchart of the encrypted index library organization and on-chain evidence storage process in step S3 of the present invention; Figure 5 This is a flowchart of the prior awareness single ciphertext concatenation matching process in step S4 of the present invention. Figure 6 This is a flowchart of the priority proof generation and forward privacy re-embedding process in step S5 of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] This embodiment uses the collaborative deployment of the national copyright management agency's consortium blockchain platform and the business systems of digital copyright service providers as an application scenario to specifically illustrate the implementation process of the blockchain-based digital copyright encryption management method supporting multi-keyword search of the present invention. In this embodiment, the consortium blockchain is jointly maintained by the national copyright management agency, several regional copyright registration centers, and several digital copyright service providers, with 13 nodes. It adopts a practical Byzantine fault-tolerant consensus mechanism, and the average block generation interval is 3 seconds. The ciphertext inverted index is stored in an off-chain ciphertext index library in a key-value structure, while the hash value of the work content and the time-series commitment root are stored on-chain for evidence. Figure 1 As shown, the method as a whole includes steps S1 to S5. Steps S1 to S3 complete the encrypted access of copyright registration data, the construction of time-anchored ciphertext index and on-chain evidence storage. Steps S4 to S5 complete the multi-keyword combination query, single ciphertext concatenation matching and prior proof generation. The output of each step is used as the input of the next step to form a deeply coupled closed-loop collaborative process.

[0020] Step S1: Receive the digital work, extract the keyword set of the copyright metadata field of the digital work according to the normalized template, and calculate the content hash value of the digital work.

[0021] like Figure 2As shown, the author submits their digital work through the registration access interface of the digital copyright service provider and fills in the copyright metadata fields according to the standardized template. In this embodiment, the copyright metadata fields include five dimensions: work name, work type, creation time, registration time, and rights holder. The standardized template constrains the value format of each field. The work type field uses a controlled thesaurus, limiting its values ​​to several preset categories such as literary works, works of art, photographic works, musical works, audiovisual works, and computer software, thereby avoiding retrieval omissions caused by synonyms and variant spellings.

[0022] For submitted digital works, the first step is to calculate the hash value of the work's complete binary content. In this embodiment, a collision-resistant hash function with an output length of 256 bits is used. The resulting hash value serves as the on-chain fingerprint of the work's integrity and ownership, used for on-chain notarization in subsequent step S3. Next, keyword extraction is performed on the copyright metadata fields according to a standardized template: the work name and rights holder fields are segmented into standardized keywords; the work type field undergoes controlled word table normalization, mapping the author's entered category to standard category words in the controlled word table; the creation time and registration time fields are discretized into standardized time keywords at the year, quarter, and month time granularities to support subsequent combined retrieval by time interval. The standardized keywords obtained from the above field segmentation are then combined to obtain the copyright metadata field keyword set for the digital work.

[0023] Taking the registration of an artwork as an example in this embodiment, the author submits the artwork's image file and fills in the copyright metadata fields on the registration access interface. The artwork's title is entered in the "Artwork Name" field, the "Artwork Type" field is selected by the author from a dropdown menu in the controlled thesaurus, the creation time and registration time fields are filled with the corresponding dates, and the rights holder field is filled with the copyright holder's name. The system segments the artwork's title field using lexical methods to obtain several name keywords, segments the rights holder field by the entity's name to obtain rights holder keywords, and normalizes the artwork type field using the controlled thesaurus to obtain the standard category keyword "artwork." The creation time and registration time fields are discretized into corresponding time keywords at two levels of granularity: year and quarter. The purpose of the controlled thesaurus normalization is that if the author enters the category as synonyms such as "painting" or "artwork," the system normalizes it to the standard category keyword "artwork" based on the synonym mapping relationship of the controlled thesaurus, thus avoiding omissions in subsequent combined searches due to differences in filling habits. The multi-granularity discretization of the time field allows copyright administrators to construct combined search conditions based on time dimensions, either by year or by quarter. After collecting all the standardized keywords obtained from the above field segmentation and normalization, duplicates are removed, and the resulting set of keywords for the copyright metadata field of the digital work is obtained.

[0024] It should be noted that the work content hash value and the copyright metadata field keyword set are generated in parallel in this step but serve different purposes: the former represents the work content itself and will be uploaded to the blockchain along with the time-series commitment root in step S3; the latter represents the work's searchable attributes and will be encrypted into ciphertext keywords in step S2. This separate processing allows the complete work content to be stored as integrity evidence using the hash value without being uploaded to the blockchain, while the registration metadata supports ciphertext retrieval after encryption. This balances the needs of content evidence storage and attribute retrieval while controlling on-chain storage overhead. The copyright metadata field keyword set and the work content hash value output in this step serve as inputs to steps S2 and S3, respectively.

[0025] Step S2: Obtain the block time sequence value generated by the registration of digital works in the consortium blockchain, determine the block time sequence value as the registration time sequence anchor point, use a symmetric encryption algorithm and randomization processing to generate ciphertext keywords for the keyword set of copyright metadata fields, and construct the registration time sequence anchor point as a monotonic time sequence commitment embedded in the ciphertext inverted index organized by the ciphertext keywords to obtain the time sequence anchored ciphertext index item.

[0026] like Figure 3 As shown, this step is one of the core components of this method, which reverses the registration time sequence from encrypted ordinary fields to a partially ordered skeleton of a ciphertext inverted index. First, after the digital work submitted in step S1 initiates a registration transaction on the consortium blockchain and is packaged into a block, the block time sequence value of the block where the registration transaction occurs is obtained. This block time sequence value is composed of the block height and the block timestamp. The registration time sequence anchor is constructed as follows: ,in: For the first The registration timeline anchor of a copyright registration record is a scalar, dimensionless quantity with a value range of positive integers. It is calculated by the block height and block timestamp of the registration block according to this formula, and represents the total order registration order of the copyright registration record on the consortium blockchain. For the first The block height of the registration block corresponding to each copyright registration record is a scalar, dimensionless, and positive integer, generated by the consensus of the consortium blockchain. Let be the time-series base, and be a scalar constant, dimensionless. In this embodiment, is taken as . Its value is based on the upper bound of the block timestamp value, so that the low part corresponding to the timestamps of different blocks within the same block height will not overflow to the adjacent block height bits. If it is too small, the registration time sequence anchor points of adjacent blocks may be out of order; if it is too large, the registration time sequence anchor point bit width is redundant. For the first The timestamp of the block corresponding to each copyright registration record is a scalar in seconds and is generated by the consensus of the consortium blockchain. This is the modulo operator. Since the block height strictly increases with the registration order, the registration sequence anchor point... The sequence of registrations strictly monotonically increases, providing a total order basis for subsequent partial-order embeddings. Then, each keyword in the copyright metadata keyword set is encrypted using a symmetric encryption algorithm and randomized to generate ciphertext keywords. To prevent statistical attacks, this embodiment generates a 128-bit random salt value for each keyword. This keyword is concatenated with the random salt value and then encrypted using a symmetric encryption algorithm. This ensures that the ciphertext keywords corresponding to the same plaintext keyword in different time-series anchored ciphertext index entries are different, thus hiding the frequency of keyword occurrences and preventing attackers from inferring the plaintext based on the repetition pattern of ciphertext keywords. The symmetric encryption algorithm used can be a block symmetric encryption algorithm, and its key is uniformly distributed by the key management center.

[0027] The registered time-series anchors are then constructed as monotonic time-series commitments and embedded into a ciphertext inverted index. Specifically, using the registered time-series anchors as input, time-series commitment values ​​are generated through monotonic cumulative mapping, calculated as follows: ,in: For the first The temporal commitment value of a copyright registration record is a scalar, dimensionless quantity, and takes the value of a positive integer. It is calculated by this formula and represents the partial order coordinates of the copyright registration record that can be compared within the ciphertext domain. For the first The registration timeline anchor point of each copyright registration record is defined in the aforementioned registration timeline anchor point formula; Where is the mask bit width, and is a scalar constant, dimensionless. In this embodiment, we take . The value is determined by a trade-off between the strength of resistance to statistical attacks and the index storage overhead. If it is too small, the mask space will be insufficient and it will be easily broken by frequency analysis. If it is too large, the timing commitment value will be redundant and increase the index storage burden. For The pseudo-random function for the key maps the input to a pseudo-random bit string; It is a pseudo-random function key, 128 bits in length, generated by the key management center; For the first The unique identifier of a copyright registration record; This is the mask modulus. This formula places the registration timing anchor point in the high-order bits and the pseudo-random mask in the low-order bits. When the registration time-series anchor points of two records differ by at least 1, the high-order bits of their time-series commitment values ​​differ by at least 1. The low-order mask value does not exceed Therefore, the timing commitment value of the first registered record must be less than the timing commitment value of the later registered record, that is, the timing commitment value maintains a partial order consistent with the registration timing in the ciphertext field; at the same time, the low-bit pseudo-random mask prevents the timing commitment value from directly exposing the precise value of the registration timing anchor point, thus taking into account both the comparability of the partial order and the resistance to inference.

[0028] Multiple time-anchored ciphertext index entries linked to the same ciphertext keyword in each time-anchored ciphertext index entry are organized into a time-ordered inverted index chain according to the partial order of the time commitment values. That is, the ciphertext keyword is used as the index key, and the sequence of time-anchored ciphertext index entries arranged in ascending order of time commitment values ​​is used as the index value to form a ciphertext inverted index. Thus, each ciphertext keyword corresponds to a time-ordered inverted index chain, and the time-anchored ciphertext index entries on the chain are arranged in ascending order of their registration time. The time-anchored ciphertext index entries output in this step and the ciphertext inverted index they form are used as the input for step S3.

[0029] Step S3: Store the time-anchored ciphertext index item in the off-chain ciphertext index library, and synchronize the time commitment root of the time-anchored ciphertext index item and the hash value of the work content on the consortium blockchain for evidence storage.

[0030] like Figure 4 As shown, considering that directly uploading the complete content of a digital work and all its encrypted indexes to the blockchain would cause an explosion in blockchain storage capacity, this step adopts a combined on-chain and off-chain storage strategy: the larger time-series anchored encrypted index items are stored in the off-chain encrypted index library, while only the time-series commitment root and the hash value of the work content are uploaded to the blockchain for notarization. This ensures reliable notarization while controlling on-chain storage overhead. The time-series commitment root is obtained by recursively calculating all time-series commitment values ​​through a hash chain. Its calculation method is as follows: ,in: The root of the timing commitment is a 256-bit bit string with no dimension. It is recursively calculated by this formula and serves as the commitment value for the partial order structure of all timing commitment values ​​on the chain. To accumulate up to the number The intermediate commitment of each timing commitment value is a 256-bit bit string with no dimension. To accumulate up to the number The intermediate commitments of each time-series commitment value have the same meaning as... ; It is a collision-resistant hash function with an output length of 256 bits; This is a bit string concatenation operator; The first time series commitment value is the first time series commitment value after being sorted in partial ascending order. Each time commitment value is defined in the same way as the aforementioned time commitment values. ; is the initial vector constant, which is a 256-bit bit string with no dimension. The total number of time-anchored ciphertext index entries in the time-ordered inverted index is a scalar, dimensionless, and takes the value of a positive integer; the subscript... The recursion index is the value of the index. to Since the root of the temporal commitment is obtained by recursively hashing all temporal commitment values ​​arranged in partial order, it uniquely anchors the partial order relationship of all temporal commitment values. Any change in the temporal commitment value or its order will cause the root of the temporal commitment to change, thus providing an on-chain anchor for the verifiability of the prior proof in step S5.

[0031] Furthermore, to prevent non-monotonic jumps in block time sequence values ​​during forks or block reorganizations in the consortium blockchain, which could disrupt the stability of the partial order of the time sequence commitment value, before determining the block time sequence value as the registration time sequence anchor point, the final confirmation depth of the registered block corresponding to the digital work is first obtained. Only when the final confirmation depth reaches a preset confirmation depth threshold is the block time sequence value of the registered block locked as the registration time sequence anchor point. Simultaneously, the final confirmation depth is recorded as the stability weight of the time sequence commitment value, and this stability weight is used as the confidence level of the prior proof when generating the prior proof in step S5. The stability weight is calculated as follows: ,in: For the first The stability weight of each copyright registration record, i.e., the confidence level of the priori proof, is a scalar, dimensionless, and its value ranges from [value missing]. The value calculated by this formula represents the degree of reliability of the registration timeline anchor point of the copyright registration record in resisting the rollback of the consortium chain fork. For the first The final confirmation depth of the registered block corresponding to each copyright registration record is a scalar, dimensionless quantity, taking the value of a positive integer. Its value is equal to the difference between the current latest block height and the block height of the registered block. ; The height of the latest block in the current consortium blockchain is a scalar, dimensionless, and its value ranges from positive integers. For the first The block height of the registration block corresponding to each copyright registration record is defined in the same way as the aforementioned registration time sequence anchor point formula; The preset confirmation depth threshold is a scalar constant, dimensionless. In this embodiment, it is taken as... The value is determined based on the finality confirmation convention of the consortium blockchain. If it is too small, the risk of the registered block being forked and rolled back is high and the registration time sequence anchor point may not change monotonically. If it is too large, the registration confirmation delay will increase and the registration timeliness will be affected. In is a base constant. As can be seen from this formula, the greater the finality confirmation depth, the closer the stability weight is to 1, and the higher the credibility of the registered time-series anchor point in resisting rollback. This confirmation depth is used simultaneously as the anchor point locking condition and the prior proof confidence level, unifying the ability to resist fork rollback and the credibility of prior proof into the same physical quantity. This step stores the time-series anchor ciphertext index item in an off-chain ciphertext index library, and uploads the time-series commitment root and the work content hash value to the chain. The resulting off-chain ciphertext index library and on-chain time-series commitment root serve as inputs to steps S4 and S5, respectively.

[0032] Step S4: Receive a copyright retrieval request carrying multiple keyword combination conditions, generate a priori-aware hierarchical trapdoor based on the on-chain time commitment root, perform a single ciphertext conjunction matching on the off-chain ciphertext index, and return the copyright registration record that satisfies the multiple keyword combination conditions and has the earliest registration time.

[0033] like Figure 5 As shown, copyright administrators submit copyright search requests with multiple keyword combinations through a query interface. These multiple keyword combinations consist of at least two of the following: work name, work type, and registration time. These items are conjunctive. For example, the system searches for prior registration records of works of art registered in a specific year and with a specific rights holder. The system generates priority-aware hierarchical trapdoors based on the on-chain temporal commitment root. These priority-aware hierarchical trapdoors convert the multiple keyword combinations into multiple ciphertext sub-trapdoors. Each ciphertext sub-trapdoor corresponds to a combination condition item and carries partial order verification information derived from the on-chain temporal commitment root, enabling the trapdoors to both locate the inverted chain and perceive temporal partial order within the ciphertext domain.

[0034] After converting the multi-keyword combination conditions into multiple ciphertext sub-trapdoors, the temporally ordered inverted index corresponding to each ciphertext sub-trapdoor in the off-chain ciphertext index is located, and then a temporally partial order-aware single ciphertext conjunction matching is performed. The matching process synchronously advances the cursor along each temporally ordered inverted index in a partially increasing direction according to the temporal commitment value. The cursor advancement rule is as follows: ,in: Let be the index of the inverted chain containing the smallest temporal commitment value at the current cursor position of each temporally ordered inverted chain. is a scalar, dimensionless, and its value range is 1. It is calculated using this formula; The number of ciphertext trapdoors converted by the multi-keyword combination conditions is a scalar, dimensionless, and its value range is a positive integer greater than or equal to 2; The subscript for the ciphertext trapdoor number has a value range of 1. ; For the first The temporally ordered inverted chain corresponding to each ciphertext trapdoor is defined in the same way as the aforementioned temporally ordered inverted chain. For the first The cursor position of the sequentially ordered inverted list is a scalar, dimensionless, and its value range is positive integers; For the first The vertex in the sequentially ordered inverted chain The time-series anchored ciphertext index item at the location; To obtain the mapping of the timing commitment value of the timing anchored ciphertext index item, its value definition is the same as the aforementioned timing commitment value. ; The operator that takes the argument that minimizes the objective function; arrow This indicates that the cursor position is incremented. At each step, the inverted index cursor with the smallest current temporal commitment value is shifted one position to the right, ensuring that all cursors advance synchronously in the direction of the earliest to latest registration time. The condition for a conjunction hit is: ,in: The common temporal commitment value, which is the value aligned by the sequentially ordered inverted chain cursors at the point of convergence, is a scalar, dimensionless quantity, and its value is defined as the aforementioned temporal commitment value. ; This is a mapping that involves looking up the corresponding copyright registration record from the time-series commitment value. To retrieve the copyright registration record corresponding to the concatenation result; It is an existential quantifier; "Universal quantifier"; the definitions of other symbols are the same as in the aforementioned cursor advancement rule formula. When the time commitment values ​​aligned to the cursors of each temporally ordered inverted index point to the same copyright registration record simultaneously for the first time, it is determined to be a conjunction hit, and advancement terminates at the first conjunction hit. Since each cursor advances synchronously in ascending order of time commitment values, and the ascending direction of time commitment values ​​is strictly consistent with the direction of registration time from earliest to latest, the copyright registration record corresponding to the first conjunction hit must be the earliest registered time among all conjunction hit records. There is no need to sort after obtaining all intersections; that is, a single scan returns the copyright registration record with the earliest registered time that satisfies the multi-keyword combination condition. The conjunction hit record output in this step and its cursor partial order position in each inverted index are used as the input for step S5.

[0035] To more clearly illustrate the execution process of a single ciphertext concatenation match, a specific example is used. Suppose the multi-keyword combination condition of the copyright search request consists of two fields: the work type field being "artwork" and the registration time field being a specific year. This means the number of ciphertext trapdoors is 2, corresponding to time-ordered inverted chains. and . The copyright registration records corresponding to the time-anchored ciphertext index entries, arranged in ascending order of time commitment value, are A, B, C, and D, respectively. The inverted lists are B, C, and E in that order. At the start of the matching, the cursors of both inverted lists point to their respective first items. The cursor points to A, The cursor points to B; since the timing commitment value of A is less than the timing commitment value of B, according to the cursor advancement rule, it will be... The cursor moves to record B. At this point, both cursors point to B, and their time commitment values ​​are equal. This is considered the first conjunctive match, and the matching immediately terminates. The copyright registration record B is returned as the earliest record that meets the conditions of being a work of art, registered in that year, and having the earliest registration time. As can be seen from this example, the matching process does not need to traverse the entire intersection of the two inverted chains (B and C). It only proceeds to the first common record and then terminates. Furthermore, because the cursor always moves in the direction of the earliest to the latest registration time, the returned record B must be the earliest registered record in the intersection. Compared to first finding the intersection {B, C} and then comparing the registration times to take the earliest one, this scheme saves the retrieval and decryption of record C, and the registration time of the unselected record C is not disclosed.

[0036] Step S5: Return proof of priority for the copyright registration record. For example... Figure 6 As shown, during a single scan in step S4, the cursor partial order position of the copyright registration record corresponding to the conjunction hit in each temporally ordered inverted chain is recorded. These cursor partial order positions are combined to form a temporally ordered rank witness. Then, a priori proof is generated using the temporally ordered rank witness, each ciphertext trapdoor, and the on-chain temporal commitment root. The priori proof is constructed as follows: ,in: For the priority proof, is a tuple, dimensionless, constructed by this formula, enabling a third party to verify the priority of the registration time of the conjunctive hit record without decrypting the unhit copyright registration record; It is the common temporal commitment value at the conjunctive hit, and its definition is the same as the aforementioned conjunctive hit determination formula; To obtain the timing commitment value at the hit point In the The authentication path to the time commitment root in the sequentially ordered inverted chain consists of several hash values ​​of 256 bits in length; The index of the sequentially ordered inverted chain is defined, and its value range is [value range missing]. ; The number of ciphertext trapdoors is defined in the same way as the aforementioned vernier advance rule formula; The confidence score for conjunctive hit records is defined in the same way as the stability weights mentioned above. ; superscript This indicates a connection to a conjunction hit record. The third-party verification method for priori proof is as follows: ,in: This is a priori proof verification function; the output is a Boolean value, dimensionless, and its range is [value range missing]. A value of 1 indicates that the verification passed; The priority proof is defined in the same way as the aforementioned priority proof construction formula; This is the on-chain timing commitment root, defined as the aforementioned timing commitment root formula; This is a function that recalculates the root of the timing commitment from the timing commitment value and the authentication path; and The definition is the same as the aforementioned formula for proving priority; Indicates to From 1 to Logical conjunction; This is the logical conjunction operator; This is a non-existent quantifier; For earlier public time-series commitment values ​​of the candidates; It is the intersection of the sets of copyright registration records corresponding to each chronologically ordered inverted index; The definition is the same as the aforementioned cursor advancement rule formula. The verification function includes two decisions: first, for each temporally ordered backward chain, the temporal commitment value at the conjunct hit is... The authentication path is recalculated to determine the timing commitment root, and its equality with the on-chain timing commitment root is verified to confirm that the merge hit record indeed belongs to each inverted chain; secondly, it is verified that there is no difference between the two. A smaller public time commitment value, which also falls within the intersection of all inverted chains, is used to confirm that the conjoint hit record is indeed the earliest registered record. When both criteria pass, the priori proof is verified, thus proving the priori registration of the conjoint hit record without decrypting any unhit copyright registration record.

[0037] Furthermore, to eliminate query pattern leakage during the retrieval and retrieval stages, the distribution of query patterns accessed in historical copyright retrieval requests for each temporally ordered inverted chain in the off-chain encrypted index is statistically analyzed. When the leakage metric corresponding to the query pattern distribution exceeds a preset leakage threshold, randomized re-embedding is performed on the encrypted keywords in the corresponding temporally ordered inverted chain while maintaining the partial order of the temporal commitment value. The leakage metric is calculated as follows: ,in: Let be a leakage metric for query pattern distribution, be a scalar, dimensionless, and take values ​​ranging from 1 to 2. The value is calculated by this formula. The larger the value, the more unbalanced the access of each time-series ordered inverted chain and the more serious the query pattern leakage. For the first The normalized access frequency of a sequentially ordered inverted index is a scalar, dimensionless, and its value range is 1. ; For the first The number of times a sequentially ordered inverted index is accessed within the statistics window is a scalar, dimensionless, and its value range is a non-negative integer. The total number of time-ordered inverted chains in the off-chain ciphertext index library is a scalar, dimensionless, and its value range is positive integers; and All are sequentially ordered inverted index numbers, with values ​​ranging from 1 to 10. ; Let be a logarithmic function with base 2, and by convention... ; Indicates to From 1 to Summation. This formula uses the entropy deficit of the normalized access frequency distribution to characterize the degree of imbalance in the query pattern. When access is completely balanced, the entropy deficit is 0 and the leakage metric is 0. When access is highly concentrated, the entropy deficit approaches 1 and the leakage metric approaches 1. In this embodiment, a preset leakage threshold of 0.3 is used. When the leakage metric exceeds 0.3, randomized re-embedding is triggered.

[0038] The computation method for order-preserving randomized re-embedding is as follows: ,in: For re-embedding the first The temporal commitment value of each copyright registration record is a scalar, dimensionless, and takes the range of positive integers. It is calculated by this formula and updates the partial order coordinates of the ciphertext field while keeping the partial order of the registration time anchor point unchanged. For the first The registration timeline anchor point of each copyright registration record is defined in the same way as the aforementioned registration timeline anchor point formula and remains unchanged during the re-embedding process; The mask bit width is defined and its value is the same as the aforementioned timing commitment value formula; For The pseudo-random function for the key; The new pseudo-random function key, compared with the original pseudo-random function key. Each key is independent, 128 bits in length, and is regenerated by the key management center. For the first The unique identifier for each copyright registration record, as defined above; The mask modulus. Due to the registration timing anchor point. During re-embedding, the sequence remains unchanged and still in a high position. The temporal commitment value after re-embedding still satisfies the partial order of earlier registered records being less than later registered records. The low-order pseudo-random mask is regenerated by an independent new key, making the ciphertext addresses presented by the same temporally anchored ciphertext index item independent between two adjacent copyright retrieval requests. This satisfies the indistinguishability of query patterns and eliminates query pattern leakage in both the retrieval and retrieval stages. When the leakage metric exceeds a preset leakage threshold, re-embedding is triggered and the off-chain ciphertext index library is updated, forming a closed loop of query behavior adjusting index construction parameters in reverse: the feedback content is the query pattern distribution and leakage metric of each temporally ordered inverted chain. The receiver of the feedback is the index re-embedding stage, which regenerates the ciphertext field partial order coordinates for the ciphertext keywords in the corresponding inverted chain and writes them back to the off-chain ciphertext index library.

[0039] Furthermore, this embodiment automatically records each copyright retrieval request through a smart contract deployed on the consortium blockchain. This record includes the requester's identifier, request time sequence value, and verification result of the priority proof corresponding to the copyright retrieval request. The record is then stored on the consortium blockchain as evidence, providing immutable on-chain evidence for the national copyright management agency to audit and trace the entire copyright registration and query process. Since each query and its priority proof verification result are automatically recorded on the blockchain by the smart contract and are tamper-proof, in the event of an ownership dispute, the on-chain record can be used to verify the retrieval request submitted by a certain entity at a specific time and the obtained priority proof. This ensures that the copyright confirmation process has traceable and auditable credibility, avoiding the risk that query records in centralized databases can be tampered with or repudulated.

[0040] It should be noted that the parameter values ​​given in this embodiment are preferred values, and those skilled in the art can adjust them according to the actual deployment scale and security requirements. For example, for consortium blockchains with more nodes and faster block production, the preset confirmation depth threshold can be appropriately increased to further reduce the risk of fork rollback; for scenarios with a larger ciphertext index library and higher query concurrency, the preset leakage threshold can be appropriately decreased to trigger order-preserving randomization re-embedding more frequently, thereby maintaining the indistinguishability of query patterns at a finer granularity; the mask bit width can be selected within a reasonable range based on the total number of registered records and the strength requirements for resisting statistical attacks. The adjustment of the above parameters does not change the core principle of this method, which constructs the registration sequence as a partial order skeleton of the ciphertext index and outputs the earliest registered record and prior proof simultaneously with a single ciphertext concatenation match.

[0041] As can be seen from steps S1 to S5, the keyword set of copyright metadata fields output in step S1 and the hash value of the work content drive the generation of encrypted keywords in step S2 and the on-chain storage in step S3, respectively. Step S2 constructs the registration time-series anchor point as a monotonic time-series commitment and embeds it into the encrypted inverted index. The resulting time-series ordered inverted chain provides a partial-order skeleton for the partial-order-aware conjunction matching in step S4. The finality confirmation depth stabilization in step S3 provides rollback protection for the registration time-series anchor point and provides confidence for the prior proof in step S5. The cursor partial-order position generated by the conjunction matching in step S4 in a single scan is exactly the time-order rank witness for the prior proof generated in step S5. The query pattern statistics in step S5, in turn, trigger the randomization re-embedding of the mask in the partial-order construction in step S2. The output of each step becomes the key input of the next step, and the query pattern statistics of the subsequent steps have a reverse influence on the index construction parameters of the preceding steps, thus forming a deeply coupled closed-loop collaborative process, so that a single ciphertext match can simultaneously satisfy the triple constraints of combined hit, earliest in time sequence, and verifiable first.

[0042] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made under the inventive concept of the present invention using the contents of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A blockchain-based digital copyright encrypted management method supporting multi-keyword search, characterized in that: include: Upon receiving a digital work, the copyright metadata field keyword set of the digital work is extracted according to a standardized template, and the content hash value of the digital work is calculated. The copyright metadata fields include the work name field, work type field, creation time field, registration time field, and rights holder field. Obtain the block time sequence value generated by the registration of the digital work in the consortium blockchain, determine the block time sequence value as the registration time sequence anchor point, use a symmetric encryption algorithm and randomization processing to generate ciphertext keywords for the keyword set of the copyright metadata field, and construct the registration time sequence anchor point as a monotonic time sequence commitment embedded in the ciphertext inverted index organized by the ciphertext keywords to obtain the time sequence anchored ciphertext index item; The time-anchored ciphertext index item is stored in the off-chain ciphertext index library, and the time commitment root of the time-anchored ciphertext index item and the hash value of the work content are synchronously uploaded to the consortium chain for evidence storage. Upon receiving a copyright retrieval request carrying multiple keyword combination conditions, a priori-aware hierarchical trapdoor is generated based on the on-chain time commitment root. A single ciphertext conjoint matching is performed on the off-chain ciphertext index library, and the copyright registration record that satisfies the multiple keyword combination conditions and has the earliest registration time, along with the priori proof of the copyright registration record, is returned.

2. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 1, characterized in that, The step of constructing the registration time-series anchor points as monotonic time-series commitments embedded in a ciphertext inverted index organized from the ciphertext keywords includes: Using the registration timeline anchor as input, a timeline commitment value is generated through monotonically accumulating mapping, such that the timeline commitment value corresponding to the copyright registration record registered earlier has a partial order in the ciphertext field before the timeline commitment value corresponding to the copyright registration record registered later. Multiple time-anchored ciphertext index entries linked to the same ciphertext keyword in the ciphertext inverted index are organized into a time-ordered inverted chain according to the partial order of the time commitment value.

3. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 2, characterized in that, Before determining the block time series value as the registration time series anchor point, the process also includes: Obtain the final confirmation depth of the registration block corresponding to the digital work by the consortium blockchain. When the final confirmation depth reaches a preset confirmation depth threshold, lock the block time sequence value of the registration block as the registration time sequence anchor point. The finality confirmation depth is recorded as a stability weight of the time-series commitment value, and the stability weight is used as the confidence level of the prior proof when generating the prior proof.

4. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 3, characterized in that, The step of performing a single ciphertext join matching on the off-chain ciphertext index includes: The multi-keyword combination conditions are converted into multiple ciphertext sub-trapdoors, and the temporally ordered inverted chain corresponding to each ciphertext sub-trapdoor in the off-chain ciphertext index is located respectively. The cursor is synchronously advanced along each of the temporally ordered inverted chains in a partially increasing direction according to the temporal commitment value. When the temporal commitment value aligned with the cursor of each of the temporally ordered inverted chains first points to the same copyright registration record, it is determined as a conjunction hit. The advancement is terminated at the first conjunction hit, and the copyright registration record with the earliest registration time and satisfying the multi-keyword combination condition is returned in a single scan.

5. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 4, characterized in that, The priori proof of returning the copyright registration record includes: During the single scan, the cursor partial order position of the copyright registration record corresponding to the conjunction hit in each of the temporally ordered inverted chains is recorded, and the cursor partial order positions are combined to form a temporally ordered witness. The priority proof is generated using the chronological rank witness, each of the ciphertext trapdoors, and the on-chain chronological commitment root, so that the priority proof can verify the priority of the registration time of the copyright registration record corresponding to the conjoint hit without decrypting the copyright registration record that was not hit.

6. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 5, characterized in that, Also includes: The distribution of query patterns accessed in historical copyright retrieval requests for each of the time-ordered inverted chains in the off-chain encrypted index library is statistically analyzed. When the leakage metric corresponding to the query pattern distribution exceeds the preset leakage threshold, randomized re-embedding is performed on the ciphertext keywords in the corresponding temporally ordered inverted chain to maintain the partial order of the temporal commitment value, and the off-chain ciphertext index library is updated. The randomized re-embedding makes the temporally anchored ciphertext index item indistinguishable from the query pattern between two adjacent copyright retrieval requests.

7. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 1, characterized in that, Also includes: Each copyright retrieval request is automatically recorded by a smart contract deployed on the consortium blockchain. The record includes the requester's identifier, the request time sequence value, and the verification result of the priority proof. The record is stored on the consortium blockchain for auditing and traceability of the entire copyright registration and query process.

8. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 1, characterized in that, The set of keywords for extracting the copyright metadata fields of the digital work according to the normalized template includes: Based on the standardized template, the work type field is normalized using a controlled thesaurus. The work name field, work type field, creation time field, registration time field, and rights holder field are respectively segmented into standardized keywords, and combined to obtain the keyword set of the copyright metadata field.

9. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 1, characterized in that, The step of generating ciphertext keywords by using a symmetric encryption algorithm and randomization processing on the keyword set of the copyright metadata field includes: A random salt value is generated for each keyword in the keyword set of the copyright metadata field. The keyword is concatenated with the random salt value and then encrypted using the symmetric encryption algorithm to obtain the ciphertext keyword, so that the ciphertext keyword corresponding to the same plaintext keyword in different time-series anchored ciphertext index items is different from each other.

10. The blockchain digital copyright encryption management method supporting multi-keyword search according to claim 1, characterized in that, Receiving a copyright search request carrying multiple keyword combinations includes: The system receives a combination of multiple keywords consisting of at least two of the following fields: work name, work type, and registration time. The combination of multiple keywords is a conjunction relationship between the keywords.

Images