Expert review method based on alliance chain

Abstract

The application discloses a method for expert review based on a consortium chain, and relates to the technical field of academic review, which comprises the following steps: S1: the central organization invites the review experts to join the network; the method for expert review based on the consortium chain can assign a unique identification to academic achievements through a smart contract, the smart contract calls an expert matching algorithm and an expert node random selection algorithm to match k expert nodes for the academic achievements, and sends the academic achievements to the expert nodes to wait for the expert nodes to return the review results; the expert nodes review the academic achievements, score the academic achievements from three aspects of innovation, integrity and writing level, call the smart contract to return the review scores and review opinions, the smart contract collects the expert review results according to the unique identification of the academic achievements, calls an academic achievement review algorithm to obtain the final review result, and automatically distributes the academic achievements to the experts in this field for review, so that the distribution is not needed manually, and thus the expert review process is optimized.

Description

Technical Field

[0001] This invention relates to the field of academic review technology, specifically to an expert review method based on consortium blockchain. Background Technology

[0002] my country's support policies for science and technology projects have guided the implementation of science and technology programs and special projects, while local governments have also established various funding programs to support these projects. This has directly led to an increase in the number of science and technology projects applied for and approved.

[0003] Traditionally, the selection and recommendation of project review experts are mostly done manually by science and technology project managers. This results in many experts being assigned projects completely unrelated to their research areas. Furthermore, relying solely on the subjective judgment of the selectors to determine whether recommended experts meet the review criteria leads to a heavy workload for management personnel and easily results in a mismatch between experts and projects. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an expert review method based on consortium blockchains, which solves the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an expert review method based on a consortium blockchain, comprising the following steps:

[0006] S1: The central organization invites review experts to join the network;

[0007] S2: Authors of academic works register author nodes and join the consortium blockchain network;

[0008] S3: The author node calls the smart contract to upload its academic achievements;

[0009] S4: Smart contracts assign unique identifiers to academic achievements;

[0010] S5: The smart contract calls the expert matching algorithm and the expert node random selection algorithm to match expert nodes for academic achievements;

[0011] S6: Send the academic results to the expert node and wait for the expert node to return the review result;

[0012] S7: Expert nodes review academic achievements, score them, and call the smart contract to return the review score and review comments;

[0013] S8: The smart contract summarizes the expert review results based on the unique identifier of the academic achievement, and calls the academic achievement review algorithm to obtain the final review result;

[0014] S9: The smart contract packages the final review results into a block to be confirmed and sends it to the consensus node to wait for the block to be produced;

[0015] S10: After receiving the block, the block-producing node calls the block verification algorithm to verify the validity of the block;

[0016] S11: The block-producing node returns the block verification result after verifying the block;

[0017] S12: When more than 1 / T of block-producing nodes confirm the block production result, the block is successfully produced.

[0018] Optionally, the step of the S1 central organization inviting review experts to join the network includes S101: registering keys for all experts using an asymmetric encryption algorithm and storing the public keys in the original block; S102: introducing a trust mechanism for experts, with the initial trust level of the expert nodes being C; S103: constructing a domain relevance model using natural language processing technology and using the model to predict the relevance of all experts to the relevant domains.

[0019] Optionally, the step of the S9 smart contract packaging the final review results into a block to be confirmed and sending it to the consensus node to wait for the block to be produced includes S901: setting up n consensus nodes in the consortium blockchain to verify the block containing the expert review results packaged by the smart contract.

[0020] Optionally, in step S102, an expert trust mechanism is introduced, where the initial trust level of all expert nodes is C. Trust level represents the fairness and timeliness of the expert's review of academic achievements. Each expert node receives a reward of λ1 (original trust value) for completing a review task, and also receives a reward of λ2 (original trust value) based on the reciprocal of the review completion time. The initial contribution value of the expert node is updated as shown in Formula 1.

[0021]

[0022] Where Lt represents the original contribution value of the expert node at time t, α represents the forgetting coefficient, which is a positive number less than 1, k1 represents the number of times the expert node correctly reviews, and k2 represents the total time required to complete a correct review. To ensure that the contribution is bounded, the Sigmoid function is used to compress it, resulting in Formula 2:

[0023]

[0024] Where C t+1 This represents the contribution value of the node at time t+1.

[0025] Furthermore, in step S103: constructing a domain relevance model using natural language processing technology and using the model to predict the relevance between all experts and the relevant domains, let's assume there are currently m experts, and let's use the set E = {E1, E2, ..., E...} m Let F = {F1, F2, ..., F} represent the academic achievements and expert set E to be reviewed, which involve L research fields. m The expression `}` represents the construction of a correlation matrix (WEF) between the expert set and the research field set, and the storage of this matrix in the original block. This matrix is ​​used for querying the correlation between expert nodes and each research field during expert matching. Where W... EF As shown in Formula 3

[0026] Optionally, in the step of the S5 smart contract calling the expert matching algorithm and the expert node random selection algorithm to match expert nodes for academic achievements: the expert matching algorithm uses natural language processing technology to construct a domain relevance model, and predicts the academic achievement V through the domain relevance model. i The correlation matrix with related research fields, where Formula 4:

[0027] By querying the correlation matrix W between the expert set and the domain set EF Obtain the correlation matrix between each expert and the relevant research field. Formula 5:

[0028] The academic achievement P was calculated using an expert matching formula. i Matching score with each expert (S) i As shown in Formula 6, the top h experts with the highest matching scores are selected as the expert nodes to be selected.

[0029] The expert random selection algorithm calculates h expert nodes to be selected based on their trust values.

[0030] The probability P that the home node is selected i As shown in Formula 7

[0031]

[0032] Based on the probability P of the expert node i Calculate the cumulative probability Q of the expert node in the roulette wheel algorithm. i As shown in Formula 8, construct the cumulative probability array Q = [Q1, Q2, ..., Q...]. h ];

[0033]

[0034] The hash function is called to return the hash value hashVal1 of the latest block content hash. Based on the maximum hash value maxHash, it is mapped to the interval [0, 1] to obtain the seed value feed for the cumulative probability array. As shown in Formula 9,

[0035]

[0036] The selected expert node Ei is calculated according to the roulette wheel algorithm, and then removed from the list of expert nodes to be selected. Let hashVal = hashVal1.

[0037] Repeat the steps k times to finally determine k expert nodes;

[0038] An evaluation factor set was established, selecting three evaluation factors: the innovativeness, completeness, and writing level of the academic achievement; the weight of each evaluation factor was determined as X = {X1, X2, ..., X3}; and the E of each expert node was calculated based on the review score G of the expert node. i The evaluation score Y for this academic achievement is shown in Formula 10.

[0039]

[0040] Based on the review score of academic achievements by expert Ei, Y i And the trust value C of the expert node i Calculate the final review score of the academic achievement as shown in Formula 11, and determine the review result identifier M of the academic achievement according to Formula 12.

[0041]

[0042]

[0043] Where M=0 indicates acceptance, M=1 indicates pending revision, and M=2 indicates rejection.

[0044] Optionally, in the step of the block-producing node in S10 calling the block verification algorithm to verify the validity of the block after receiving the block, the identity of the expert node is verified by calling the asymmetric encryption algorithm, the expert matching algorithm and the expert node random selection algorithm are called to verify the expert node matching result, and the academic achievement review algorithm is called to verify the academic achievement review result. If the verification is successful, a confirmed block production result is returned; otherwise, a rejected block production result is returned.

[0045] Optionally, when more than 1 / T of block-producing nodes confirm the block production result in step S12, the author node can query the academic achievement review result and perform corresponding operations based on the review result after the successful block production step.

[0046] This invention provides an expert review method based on consortium blockchain, which has the following beneficial effects:

[0047] This expert review method based on consortium blockchain can assign a unique identifier to academic achievements through smart contracts. The smart contract calls an expert matching algorithm and an expert node random selection algorithm to match k expert nodes with the academic achievement, and sends the academic achievement to the expert nodes to wait for the expert nodes to return the review results. The expert nodes review the academic achievement and score it from three aspects: innovation, completeness, and writing level. They also call the smart contract to return the review score and review comments. The smart contract summarizes the expert review results based on the unique identifier of the academic achievement, calls the academic achievement review algorithm to obtain the final review result, and can automatically assign academic achievements to experts in this field for review, thus eliminating the need for manual allocation and optimizing the expert review process. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the steps and structure of the present invention;

[0049] Figure 2 This is a flowchart illustrating the steps of the S1 central organization in this invention to invite review experts to join the network;

[0050] Figure 3 The diagram illustrates the steps of the S9 smart contract described in this invention to package the final review results into a block to be confirmed and send it to the consensus node to wait for the block to be produced. Detailed Implementation

[0051] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0052] Please see Figures 1 to 3 This invention provides a technical solution: an expert review method based on a consortium blockchain, comprising the following steps:

[0053] S1: The central organization invites review experts to join the network;

[0054] S2: Authors of academic works register author nodes and join the consortium blockchain network;

[0055] S3: The author node calls the smart contract to upload its academic achievements;

[0056] S4: Smart contracts assign unique identifiers to academic achievements;

[0057] S5: The smart contract calls the expert matching algorithm and the expert node random selection algorithm to match expert nodes for academic achievements;

[0058] S6: Send the academic results to the expert node and wait for the expert node to return the review result;

[0059] S7: Expert nodes review academic achievements, score them, and call the smart contract to return the review score and review comments;

[0060] S8: The smart contract summarizes the expert review results based on the unique identifier of the academic achievement, and calls the academic achievement review algorithm to obtain the final review result;

[0061] S9: The smart contract packages the final review results into a block to be confirmed and sends it to the consensus node to wait for the block to be produced;

[0062] S10: After receiving the block, the block-producing node calls the block verification algorithm to verify the validity of the block;

[0063] S11: The block-producing node returns the block verification result after verifying the block;

[0064] S12: When more than 1 / T of block-producing nodes confirm the block production result, the block is successfully produced.

[0065] Furthermore, the steps for the S1 central organization to invite review experts to join the network include: S101: registering keys for all experts using an asymmetric encryption algorithm and storing the public keys in the original block; S102: introducing a trust mechanism for experts, with the initial trust level of all expert nodes being C; S103: constructing a domain relevance model using natural language processing technology and using the model to predict the relevance of all experts to the relevant domains, enabling this method to select professional experts for review in the domains corresponding to the academic achievements to be reviewed.

[0066] Furthermore, the steps of the S9 smart contract to package the final review results into a block to be confirmed and send it to the consensus nodes to wait for block production include S901: setting up n consensus nodes in the consortium blockchain, responsible for verifying the block containing the expert review results packaged by the smart contract, used to implement the consortium blockchain consensus algorithm, responsible for verifying the block containing the expert review results packaged by the smart contract, and finally adding the verified block to the consortium blockchain to complete consensus and produce a block.

[0067] Furthermore, in step S102, a trust mechanism is introduced for experts. In this step, where all initially added expert nodes have a trust level of C, the trust level represents the fairness and timeliness of the expert's review of academic achievements. Each time an expert node completes a review task, it receives a reward of λ1 (original trust value) and λ2 (original trust value) based on the reciprocal of the review completion time. The initial contribution value of the expert node is updated as shown in Formula 1.

[0068]

[0069] Where Lt represents the original contribution value of the expert node at time t, α represents the forgetting coefficient, which is a positive number less than 1, k1 represents the number of times the expert node correctly reviews, and k2 represents the total time required to complete a correct review. To ensure that the contribution is bounded, the Sigmoid function is used to compress it, resulting in Formula 2:

[0070]

[0071] Where C t+1 This represents the contribution value of the node at time t+1.

[0072] Furthermore, in step S103: Constructing a domain relevance model using natural language processing techniques and using the model to predict the relevance of all experts to the relevant domains, let's assume there are currently m experts, and let the set E = {E1, E2, ..., E...} m Let F = {F1, F2, ..., F} represent the academic achievements and expert set E to be reviewed, which involve L research fields. m The expression `}` represents the construction of a correlation matrix (WEF) between the expert set and the research field set, and the storage of this matrix in the original block. This matrix is ​​used for querying the correlation between expert nodes and each research field during expert matching. Where W... EF As shown in Formula 3

[0073] Furthermore, in the steps of the S5 smart contract calling the expert matching algorithm and the expert node random selection algorithm to match expert nodes for academic achievements: the expert matching algorithm uses natural language processing technology to construct a domain relevance model, and predicts the academic achievement V through the domain relevance model. i The correlation matrix with related research fields, where Formula 4:

[0074] By querying the correlation matrix W between the expert set and the domain set EF Obtain the correlation matrix between each expert and the relevant research field. Formula 5:

[0075] The academic achievement P was calculated using an expert matching formula. i Matching score with each expert (S) i As shown in Formula 6, the top h experts with the highest matching scores are selected as the expert nodes to be selected.

[0076] Expert random selection algorithm: Calculate the probability P of selecting h candidate expert nodes based on their trust values. i As shown in Formula 7.

[0077]

[0078] Based on the probability P of the expert node i Calculate the cumulative probability Q of the expert node in the roulette wheel algorithm. i As shown in Formula 8, construct the cumulative probability array Q = [Q1, Q2, ..., Q...]. h ];

[0079]

[0080] The hash function is called to return the hash value hashVal1 of the latest block content hash. Based on the maximum hash value maxHash, it is mapped to the interval [0, 1] to obtain the seed value feed for the cumulative probability array. As shown in Formula 9,

[0081]

[0082] The selected expert node Ei is calculated according to the roulette wheel algorithm, and then removed from the list of expert nodes to be selected. Let hashVal = hashVal1.

[0083] Repeat the steps k times to finally determine k expert nodes;

[0084] An evaluation factor set was established, selecting three evaluation factors: the innovativeness, completeness, and writing level of the academic achievement; the weight of each evaluation factor was determined as X = {X1, X2, ..., X3}; and the E of each expert node was calculated based on the review score G of the expert node. i The evaluation score Y for this academic achievement is shown in Formula 10.

[0085]

[0086] Based on the review score of academic achievements by expert Ei, Y i And the trust value C of the expert node i Calculate the final review score of the academic achievement as shown in Formula 11, and determine the review result identifier M of the academic achievement according to Formula 12.

[0087]

[0088]

[0089] Where M=0 indicates acceptance, M=1 indicates pending revision, and M=2 indicates rejection.

[0090] Furthermore, in the step of the S10 block-producing node calling the block verification algorithm to verify the validity of the block after receiving the block, the expert node identity is verified by calling the asymmetric encryption algorithm, the expert matching algorithm and the expert node random selection algorithm are called to verify the expert node matching result, and the academic achievement review algorithm is called to verify the academic achievement review result. If the verification is successful, the block production result is returned as confirmed; otherwise, the block production result is returned as rejected.

[0091] Furthermore, when S12 receives confirmation of block production results from more than 1 / T of block-producing nodes, the author node can query the academic achievement review results and perform corresponding operations based on the review results after the successful block production step.

[0092] In summary, this expert review method based on a consortium blockchain involves the central institution inviting review experts to join the network. An asymmetric encryption algorithm is used to register keys for all experts, and the public keys are stored in the original block for initial registration and identity verification by expert nodes. A trust mechanism is introduced to assign trust values ​​to experts. A domain relevance model is constructed using natural language processing technology, and this model predicts the relevance of experts to their respective domains. Authors of academic works register as author nodes. The consortium blockchain distributes user keys and consortium blockchain certificates to these author nodes. Author nodes join the consortium blockchain network and upload their academic works via a smart contract, awaiting expert review results. The smart contract assigns a unique identifier to each academic work. The smart contract uses an expert matching algorithm and an expert node random selection algorithm to match k expert nodes with the academic work and sends the academic work to these nodes, awaiting their review results. The academic work is reviewed, scored based on three aspects: innovation, completeness, and writing quality. A smart contract is invoked to return the review score and comments. The smart contract summarizes the expert review results based on the unique identifier of the academic work, calls the academic work review algorithm to obtain the final review result, packages the final review result into a block to be confirmed, and sends it to the consensus node to await block production. Upon receiving the block, the block-producing node calls the block verification algorithm to verify the block's validity, the asymmetric encryption algorithm to verify the expert node's identity, the expert matching algorithm and the expert node random selection algorithm to verify the expert node matching result, and the academic work review algorithm to verify the academic work review result. If the verification is successful, a confirmed block production result is returned; otherwise, a rejected block production result is returned. When more than 1 / T of block-producing nodes confirm the block production result, the block is successfully produced. Author nodes can query the academic work review results and perform corresponding operations based on the review results.

[0093] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An expert review method based on consortium blockchain, characterized by: Includes the following steps: S1: The central organization invites review experts to join the network; S101: Use an asymmetric encryption algorithm to register keys for all experts and store the public keys in the original block; S102: Introduce a trust mechanism for experts, with the initial trust level of all expert nodes being C; S103: Construct a domain relevance model using natural language processing technology, and use the model to predict the relevance of all experts to the relevant domains; S2: Authors of academic works register author nodes and join the consortium blockchain network; S3: The author node calls the smart contract to upload its academic achievements; S4: Smart contracts assign unique identifiers to academic achievements; S5: The smart contract calls the expert matching algorithm and the expert node random selection algorithm to match expert nodes for academic achievements; The expert matching algorithm uses natural language processing technology to construct a domain relevance model, and predicts academic achievements through this model. The correlation matrix with related research fields, where Formula 4: Formula 4; By querying the correlation matrix between the expert set and the domain set Obtain the correlation matrix between each expert and the relevant research field. Formula 5: Formula 5; Academic achievements are calculated using an expert matching formula. Matching scores with various experts As shown in Formula 6, the top h experts with the highest matching scores are selected as the expert nodes to be selected. Formula Six; The expert random selection algorithm calculates the probability of selecting h candidate expert nodes based on their trust values. As shown in Formula 7. Formula 7; Based on the probability of expert nodes Calculate the cumulative probability of expert nodes in the roulette wheel algorithm. As shown in Formula 8, construct the cumulative probability array. ; Formula 8; The hash function is called to return the hash value hashVal1 of the latest block content hashVal. Based on the maximum hash value maxHash, it is mapped to the interval [0, 1] to obtain the seed value feed for the cumulative probability array, as shown in Formula 9. Formula Nine; The selected expert node Ei is calculated using the roulette wheel algorithm, and then removed from the list of candidate expert nodes. ; Repeat the steps k times to finally determine k expert nodes; Establish a set of evaluation factors and select three evaluation factors: the innovativeness, completeness, and writing level of academic achievements; Determine the weight of each evaluation factor. Based on the expert node's review score G, calculate the score for each expert node. The evaluation score Y for this academic achievement is shown in Formula 10. Formula 10; Based on the review scores of the academic achievements by the expert Ei And the trust value of expert nodes Calculate the final review score of the academic achievement as shown in Formula 11, and determine the review result identifier M of the academic achievement according to Formula 12. Formula 11; Official Twelve; Where M=0 indicates acceptance, M=1 indicates pending revision, and M=2 indicates rejection; S6: Send the academic results to the expert node and wait for the expert node to return the review result; S7: Expert nodes review academic achievements, score them, and call the smart contract to return the review score and review comments; S8: The smart contract summarizes the expert review results based on the unique identifier of the academic achievement, and calls the academic achievement review algorithm to obtain the final review result; S9: The smart contract packages the final review results into a block to be confirmed and sends it to the consensus node to wait for the block to be produced; S10: After receiving the block, the block-producing node calls the block verification algorithm to verify the validity of the block; S11: The block-producing node returns the block verification result after verifying the block; S12: When more than 1 / T of block-producing nodes confirm the block production result, the block is successfully produced.

2. The expert review method based on consortium blockchain according to claim 1, characterized in that: The steps of the S9 smart contract to package the final review results into a block to be confirmed and send it to the consensus node to wait for the block to be produced include S901: setting up n consensus nodes in the consortium blockchain to verify the block containing the expert review results packaged by the smart contract.

3. The expert review method based on consortium blockchain according to claim 1, characterized in that: In step S102, the trust level is used to represent the fairness and timeliness of experts' reviews of academic achievements; each time an expert node completes a review task, it receives a reward of λ1 original trust level value, and simultaneously receives a reward based on the reciprocal of the review completion time. The initial trust value is awarded; the initial contribution value of expert nodes is updated as shown in Formula 1: Formula 1; Where L t Let represent the original contribution value of the expert node at time t, and α represent the forgetting coefficient, which is a positive number less than 1. This indicates the number of times the expert node correctly reviewed the data. This represents the total time required to complete the correct review. To ensure that the contribution is bounded, the Sigmoid function is used to compress it, resulting in Formula 2: Formula 2; in This represents the contribution value of the node at time t+1.

4. The expert review method based on consortium blockchain according to claim 1, characterized in that: In step S103: constructing a domain relevance model using natural language processing technology and using the model to predict the relevance of all experts to the relevant domain, assuming there are currently m experts, a set is used. It is indicated that the academic achievements and expert set E to be reviewed involve a total of L research fields, represented by the set... This involves constructing a correlation matrix (WEF) between the expert set and the research field set, and storing the correlation matrix in the original block. This matrix is ​​used to query the correlation between expert nodes and each research field during expert matching. As shown in Formula 3 Formula 3.

5. The expert review method based on consortium blockchain according to claim 1, characterized in that: In the step of the block-producing node in S10 receiving a block and calling the block verification algorithm to verify the validity of the block, the identity of the expert node is verified by calling the asymmetric encryption algorithm, the expert matching algorithm and the expert node random selection algorithm are called to verify the expert node matching result, and the academic achievement review algorithm is called to verify the academic achievement review result. If the verification is successful, the block production result is returned as confirmed; otherwise, the block production result is returned as rejected.

6. The expert review method based on consortium blockchain according to claim 1, characterized in that: When S12 receives confirmation of block production from more than 1 / T of block-producing nodes, the author node can query the academic achievement review results and perform corresponding operations based on the review results after the successful block production step.

Images