Electronic mediation protocol tamper-proofing verification method based on double hashing and blockchain

Abstract

The present application relates to the field of blockchain storage, especially to an electronic mediation agreement anti-tamper verification method based on double hash and blockchain, the method comprising: determining a registration mediation agreement based on text similarity, and determining reference keywords of a target mediation agreement based on the registration mediation agreement; determining an audio division strategy based on an associated reference value; when executing a first division strategy, screening a number of relevant keywords based on co-occurrence strength, and determining an effective audio paragraph based on the reference keywords and the relevant keywords; when executing a second division strategy, determining whether to perform clustering segmentation based on keyword distribution representation value; for the effective audio paragraph obtained after paragraph segmentation is completed, determining a processing path according to a quality evaluation state to obtain an auxiliary verification paragraph; when the target mediation agreement is not signed, uploading the target mediation agreement and the auxiliary verification paragraph for storage. The present application improves the anti-tamper verification effect of the electronic mediation agreement.

Description

Technical Field

[0001] This invention relates to the field of blockchain-based evidence storage, and in particular to a tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain. Background Technology

[0002] In dispute resolution mechanisms, the integrity and non-repudiation of electronic mediation agreements are crucial to ensuring their legal validity. Hash value notarization technology, due to its high sensitivity to content modification, is widely used to prevent tampering of electronic mediation agreements. However, existing technologies mostly focus on hash notarization of the mediation agreement documents, that is, only paying attention to whether the content has been changed, while ignoring the impact of the environment at the time of signing on the electronic mediation agreement. Hash notarization of the agreement content alone cannot prove the authenticity of the electronic mediation agreement, resulting in a single dimension of notarization and thus reducing the accuracy of verification of electronic mediation agreements. Therefore, how to break through the limitation of only focusing on the content of the agreement itself, incorporate relevant audio of the signing process as key evidence into the notarization system, and construct a multi-dimensional anti-tampering verification method to improve the accuracy of verification of electronic mediation agreements has become an urgent technical problem to be solved.

[0003] Chinese Patent Publication No. CN112434342A discloses a blockchain-based electronic evidence storage method and system. The method involves the user uploading electronic data to a blockchain electronic evidence storage system; the system encrypts the electronic data and stores it on the blockchain; simultaneously, it generates a corresponding electronic evidence based on the digest information in the electronic data; and it establishes a link between the electronic evidence and the electronic data. The system includes: a data upload module, an encryption and storage module, an electronic evidence generation module, and a linking module. However, while the above technical solution considers using blockchain for electronic data storage to address the issues of easy loss and tampering of electronic data, it does not consider the impact of multi-dimensional evidence storage on the tamper-proofing of electronic data, thus reducing the verification effectiveness of the electronic data. Summary of the Invention

[0004] To address this, the present invention provides an anti-tampering verification method for electronic mediation protocols based on dual hashing and blockchain, which overcomes the shortcomings of existing technologies that do not consider the impact of multi-dimensional evidence storage on the anti-tampering of electronic data, thereby reducing the verification effectiveness of electronic data.

[0005] To achieve the above objectives, this invention provides a tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain, comprising: The registration and mediation protocol is determined based on the comparison results between the text similarity and the preset text similarity, and the reference keywords of the target mediation protocol are determined based on the registration and mediation protocol. For reference keywords, determine the audio segmentation strategy based on the associated reference values; The audio segmentation strategies include a first segmentation strategy that segments the target audio segment based on associated reference values ​​and a second segmentation strategy that performs adaptive segmentation on the target audio segment. When implementing the first segmentation strategy, relevant keywords for the number of seats are selected based on co-occurrence intensity, and audio segments are determined based on reference keywords and relevant keywords. The question is whether to increase the number of seats based on the continuity reference value. When implementing the second partitioning strategy, the decision to perform clustering is based on the comparison between the keyword distribution representation value and the preset keyword distribution representation value. For the valid audio segments obtained after segmentation, the quality assessment status of the valid audio segments is determined based on speech rate stability and conflict density, and the processing path is determined according to the quality assessment status: either perform separation difficulty analysis or normal storage. When performing separation difficulty analysis, the separation difficulty characterization value is determined based on voiceprint similarity and interruption frequency; and the audio storage method is adjusted from on-chain evidence storage to ordinary storage based on the comparison result between the separation difficulty characterization value and the preset separation difficulty characterization value. Before the target mediation agreement is signed, the target mediation agreement and the auxiliary verification paragraphs will be stored on the blockchain for evidence.

[0006] Furthermore, historical mediation agreements with text similarity greater than the preset text similarity are recorded as the registered mediation agreements of the target mediation agreement, and the keywords of each registered mediation agreement are recorded as the reference keywords of the target mediation agreement.

[0007] Furthermore, when the correlation reference value is greater than the preset correlation reference value, paragraph segmentation is determined based on the co-occurrence intensity.

[0008] Furthermore, when executing the first segmentation strategy, the effective audio segments of the reference keywords are obtained in descending order of co-occurrence intensity. For a single reference keyword, a number of related keywords are selected to construct the shortest audio segment that includes the reference keyword and the number of related keywords. The first segmentation strategy is to segment paragraphs based on co-occurrence intensity.

[0009] Furthermore, when the penetration reference value is greater than the preset penetration reference value, it is determined that the capacity should be increased. If the associated reference value is less than or equal to the preset associated reference value or the through reference value is less than or equal to the preset through reference value, then there is no need to adjust the capacity.

[0010] Furthermore, when the correlation reference value is less than or equal to the preset correlation reference value, the audio segmentation strategy is determined to be adaptive segmentation for the target audio segment.

[0011] Furthermore, specifically, when the keyword distribution representation value is greater than the preset keyword distribution representation value, it is determined that keyword clustering is performed based on time intervals to obtain several valid audio segments.

[0012] Furthermore, for valid audio segments whose quality assessment status is that the speech rate stability is greater than the preset speech rate stability and the conflict density is less than or equal to the preset conflict density, the processing path is determined to be to perform separation difficulty analysis. For valid audio segments whose quality assessment status is that the speech rate stability is less than or equal to the preset speech rate stability or the conflict density is greater than the preset conflict density, the analysis strategy is to use normal storage.

[0013] Furthermore, when performing separation difficulty analysis, the separation difficulty characterization value is determined based on voiceprint similarity and interruption frequency; When the separation difficulty representation value is less than or equal to the preset separation difficulty representation value, the valid audio segments are stored on the blockchain for evidence. The separation difficulty characterization value is positively correlated with voiceprint similarity and interruption frequency.

[0014] Furthermore, before the target mediation agreement is signed, the auxiliary verification paragraphs are hashed to generate an audio fingerprint, which is then recorded on the blockchain. The audio fingerprint, the document fingerprint of the target mediation agreement before signing, and the document fingerprint after signing are compared and verified to output a verification report.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention determines the registration and mediation protocol based on the comparison results of text similarity and preset text similarity, and determines the reference keywords of the target mediation protocol based on the registration and mediation protocol. The text similarity characterizes the degree of similarity between the target mediation protocol and the historical mediation protocol in terms of text, and then introduces the manually labeled keywords in the historical mediation protocol as the reference keywords of the target mediation protocol, thereby improving the extraction efficiency and accuracy of reference keywords, avoiding keyword deviations caused by special wording or brief expression of the target protocol, thereby improving the segmentation accuracy of the subsequent target audio segments, and thus improving the anti-tampering verification effect of the electronic mediation protocol.

[0016] Furthermore, the audio segmentation strategy based on correlation reference values ​​in this invention involves segmenting the target audio segment based on the correlation reference values ​​or performing adaptive segmentation on the target audio segment. The correlation reference values ​​characterize the degree of correlation between reference keywords. If the correlation between reference keywords is strong, the target audio segment is segmented based on the correlation reference values ​​to avoid splitting highly correlated keywords into different audio segments, which would result in poor segmentation of the target audio segment. If the correlation between reference keywords is weak, adaptive segmentation is performed on the target audio segment. The adaptive segmentation is based on the distribution of reference keywords, which avoids the difficulty of a single segmentation method meeting the needs of actual scenarios, thereby improving the segmentation accuracy of the target audio segment and thus improving the anti-tampering verification effect of the electronic mediation protocol. Attached Figure Description

[0017] Figure 1 The flowchart shows the tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to the present invention. Figure 2 This is a flowchart illustrating the audio segmentation strategy determined by the present invention based on associated reference values; Figure 3 This is a flowchart illustrating the process of determining whether to perform clustering segmentation based on keyword distribution representation values ​​according to the present invention. Figure 4 This is a flowchart illustrating the process of determining whether a valid paragraph is an auxiliary verification paragraph based on the separation difficulty characterization value in this invention. Detailed Implementation

[0018] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0019] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0020] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0021] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0022] Please see Figures 1 to 4 As shown, this invention provides a tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain. The method includes: The registration and mediation protocol is determined based on the comparison results between the text similarity and the preset text similarity, and the reference keywords of the target mediation protocol are determined based on the registration and mediation protocol. For reference keywords, determine the audio segmentation strategy based on the associated reference values; The audio segmentation strategies include a first segmentation strategy that segments the target audio segment based on associated reference values ​​and a second segmentation strategy that performs adaptive segmentation on the target audio segment. When implementing the first segmentation strategy, relevant keywords for the number of seats are selected based on co-occurrence intensity, and audio segments are determined based on reference keywords and relevant keywords. The question is whether to increase the number of seats based on the continuity reference value. When implementing the second partitioning strategy, the decision on whether to perform clustering is based on the comparison between the keyword distribution representation value and the preset keyword distribution representation value. For the valid audio segments obtained after segmentation, the quality assessment status of the valid audio segments is determined based on speech rate stability and conflict density, and the processing path is determined according to the quality assessment status: either perform separation difficulty analysis or normal storage. When performing separation difficulty analysis, the separation difficulty characterization value is determined based on voiceprint similarity and interruption frequency; and the audio storage method is adjusted from on-chain evidence storage to ordinary storage based on the comparison result between the separation difficulty characterization value and the preset separation difficulty characterization value. Before the target mediation agreement is signed, the target mediation agreement and the auxiliary verification paragraphs will be stored on the blockchain for evidence.

[0023] This invention is applied to the tamper-proof verification of mediation agreements. The target audio segment in this invention is an audio recording of the communication process prior to the signing of the mediation agreement, using a recording device; the type of recording device is not limited. The mediation agreement requiring tamper-proof verification analysis is denoted as the target mediation agreement. This invention includes a mediation agreement database, which provides downloadable mediation agreement data for model training.

[0024] The reference test records several instances of anti-tampering verification that meet user requirements, including text similarity, correlation reference value, continuity reference value, time interval, keyword distribution characterization value, speech rate stability, conflict density, separation difficulty characterization value, voiceprint similarity, and interruption frequency. Whether the anti-tampering verification effect meets user requirements can be determined based on self-defined indicators (e.g., the accuracy of anti-tampering verification). For example, if the accuracy of anti-tampering verification is greater than the accuracy of the anti-tampering verification required by the user, then the user requirements are met. This is content already known to those skilled in the art and is not limited here. Taking the preset continuity reference value as an example, a preset value selection method is provided: extract the continuity reference value of the reference test, remove outliers, and record the average value of the continuity reference value after removing outliers as the preset continuity reference value. Methods for removing outliers include, but are not limited to, the 3σ method.

[0025] The preset values ​​in this invention include: preset text similarity, preset association reference value, preset time interval, preset connection reference value, preset keyword distribution characterization value, preset speech rate stability, preset conflict density, fixed duration window, preset separation difficulty characterization value, preset voiceprint similarity, preset interruption frequency, and baseline capacity.

[0026] Specifically, historical mediation agreements with text similarity greater than a preset text similarity are recorded as the registered mediation agreements of the target mediation agreement, and the keywords of each registered mediation agreement are recorded as the reference keywords of the target mediation agreement.

[0027] This invention performs text transcription on a target audio segment to generate target text, and performs similarity analysis on the target text and the historical text corresponding to the historical mediation agreement. The similarity analysis process includes: performing vector mapping on the mediation agreement to generate several word vectors, and recording the average value of the word vectors of the mediation agreement as the semantic vector of the mediation agreement. The mediation agreement can be a target mediation agreement or several historical mediation agreements. The semantic vector of the target mediation agreement is recorded as the target semantic vector A, and the semantic vector of the historical mediation agreement is recorded as the historical semantic vector.

[0028] in, For the first The semantic vector of a historical mediation agreement, p, where p is the number of historical mediation agreements, and each historical mediation agreement corresponds to a set of keywords obtained through manual labeling. , Set C contains reference keywords. To match the number of mediation agreements, it is not difficult to understand that... < In this embodiment of the invention, the preset text similarity value is 0.86. It is easy to understand that the higher the user's requirements for the anti-tampering verification effect of the protocol, the larger the preset text similarity value. In the actual mediation process, the final generated mediation agreement and the recording of the entire mediation process are two heterogeneous data sets. Considering the characteristics of audio being lengthy, colloquial, and containing a large amount of irrelevant content, directly processing the entire audio recording, which is several hours long, would be computationally intensive and noisy. Therefore, the registration of the mediation agreement is determined by calculating text similarity. The key points of contention and the discussion paragraphs of rights and obligations in the registered mediation agreement often have a highly similar semantic distribution to the audio paragraphs of the target mediation agreement. These keywords from the registered mediation agreement are used as reference keywords to improve the analysis efficiency of the target audio paragraphs.

[0029] Specifically, when the correlation reference value is greater than the preset correlation reference value, the segmentation is determined based on the co-occurrence intensity.

[0030] For a single reference keyword, obtain the time interval between the reference keyword and other reference keywords. Reference keywords with time intervals less than or equal to a preset time interval are recorded as related keywords, and the number of related keywords is recorded as the co-occurrence strength. The association reference value = maximum co-occurrence strength / number of reference keywords.

[0031] In this invention, the preset association reference value is 0.6 and the preset time interval is 3, with the unit being min. It is easy to understand that the association reference value represents the degree of association between reference keywords, and the time interval represents the time distance between the occurrence times of two reference keywords. The higher the user's requirements for the anti-tampering verification effect of the protocol, the larger the preset association reference value and the smaller the preset time interval.

[0032] Specifically, when executing the first segmentation strategy, the effective audio segments of the reference keywords are obtained in descending order of co-occurrence intensity. For a single reference keyword, a number of related keywords are selected to construct the shortest audio segment that includes the reference keyword and the number of related keywords. The first segmentation strategy is to segment paragraphs based on co-occurrence intensity.

[0033] For a single reference keyword, select a number of related keywords in ascending order of their time interval with that reference keyword.

[0034] Specifically, when the penetration reference value is greater than the preset penetration reference value, it is determined that the capacity should be increased. Adjusted capacity = Unadjusted capacity × (Penetrating reference value / Preset penetrating reference value); The penetrating reference value is determined by the following method: Penetrating reference value... ;in, For the first The co-occurrence strength of each reference keyword, for The average co-occurrence intensity of the reference keywords in the text. for The number of reference keywords in the text.

[0035] Characterize the set by connecting reference values. The degree of balance in the co-occurrence intensity of reference keywords in the set The co-occurrence strength of the reference keywords is relatively balanced, indicating that the reference keywords have a high degree of consistency in semantic association. They all have strong and similar co-occurrence relationships and together form a tight semantic cluster. At this time, by increasing the number of adjustable inclusions, more related keywords with similar association strengths can be included in the paragraph construction scope to ensure the integrity and continuity of the audio paragraph in semantic coverage and avoid missing key semantic information due to truncation. In this invention, the preset continuity reference value is 10. It is easy to understand that the higher the user's requirements for the anti-tampering verification effect of the protocol, the larger the preset continuity reference value will be.

[0036] If the associated reference value is less than or equal to the preset associated reference value or the through reference value is less than or equal to the preset through reference value, then there is no need to adjust the capacity.

[0037] When there is no need to adjust the capacity, a baseline capacity is used. In this invention, the baseline capacity is 5. It is easy to understand from the baseline capacity that the higher the user's requirements for the anti-tampering verification effect of the protocol, the larger the value of the baseline capacity.

[0038] Specifically, when the correlation reference value is less than or equal to the preset correlation reference value, the audio segmentation strategy is determined to be adaptive segmentation for the target audio segment.

[0039] Specifically, when the keyword distribution representation value is greater than the preset keyword distribution representation value, it is determined that keyword clustering is performed based on time intervals to obtain several valid audio segments.

[0040] When the keyword distribution representation value is less than or equal to the preset keyword distribution representation value, the target audio segment is evenly divided to obtain several effective audio segments.

[0041] The association duration of the reference keyword is obtained, and the minimum duration covering the reference keyword and its related keywords is recorded as the association duration. The sub-distribution representation value = number of associated keywords / association duration. The average value of the sub-distribution representation value is recorded as the distribution representation mean. The keyword distribution representation value = (maximum sub-distribution representation value - distribution representation mean) / (distribution representation mean - minimum sub-distribution representation value). In this invention, the preset value of the keyword distribution representation value is 4. It is easy to understand that the higher the user's requirements for the anti-tampering verification effect of the protocol, the larger the preset value of the keyword distribution representation value will be. The keyword distribution representation value represents the density of keyword distribution within the target audio segment. Therefore, when the density of keyword distribution is large, cluster analysis is considered for keywords to obtain effective audio segments.

[0042] The earliest appearing reference keyword in the target audio segment is used as the starting detection point. The sub-audio segment corresponding to this starting detection point and the associated duration of the related keywords is recorded as a valid audio segment. The above steps are repeated, starting with the earliest appearing reference keyword in the remaining target audio segments, until no reference keyword in the target audio segment meets the segmentation criteria. The segmentation criteria are that the reference keyword does not contain any related keywords.

[0043] The sub-audio segments obtained after dividing the target audio segment are recorded as valid audio segments.

[0044] Specifically, for valid audio segments whose quality assessment status is that the speech rate stability is greater than the preset speech rate stability and the conflict density is less than or equal to the preset conflict density, the processing path is to perform separation difficulty analysis. For valid audio segments whose quality assessment status is that the speech rate stability is less than or equal to the preset speech rate stability or the conflict density is greater than the preset conflict density, the processing path is determined to be normal storage.

[0045] The method for confirming speech rate stability is: Speech rate stability = 1 - coefficient of variation; Coefficient of variation = ;in, For the first The speech rate of a fixed-duration window. The total number of fixed-duration windows, The average speech rate of the valid audio segment is calculated as follows: For a single valid audio segment, the segment is divided evenly into fixed-duration windows, and the number of syllables within a single fixed-duration window is recorded as the speech rate of that fixed-duration window.

[0046] It is easy to understand that the higher the user's requirements for the anti-tampering verification effect of the protocol, the smaller the value of the fixed duration window. A smaller time window can capture more subtle audio changes, thereby improving the verification effect. This invention provides a fixed duration window value of 2s.

[0047] For a single valid audio segment, the conflict density is calculated as: (Duration of overlapping segments) / (Duration of the valid audio segment). The duration of the overlapping segments is determined by obtaining the number of speakers within the valid audio segment and analyzing the speech log for that segment. It's easy to understand that the overlapping segments are audio segments corresponding to time intervals where two or more speakers speak simultaneously. If speaker A's speaking time is 2 to 7 seconds, and speaker B's speaking time is 4 to 9 seconds, then the overlapping segment is 4 to 7 seconds, and the duration of the overlapping segment is 3 seconds. How to generate the speech log for valid audio segments is easily understood by those skilled in the art and will not be elaborated upon here.

[0048] In this invention, the preset speech rate stability value is 0.837, and the preset conflict density value is 0.3. It is easy to understand that speech rate stability represents that the speaker's voiceprint features are relatively stable, and conflict density represents that there are fewer instances of simultaneous speaking among speakers. When the voiceprint features are relatively stable and there are fewer instances of simultaneous speaking among speakers, the quality of the represented effective audio segments is high. When the voiceprint features are unstable or there are many instances of simultaneous speaking among speakers, the quality of the represented effective audio segments is poor, and the analytical value in subsequent analysis is poor. Therefore, effective audio segments with poor quality are stored normally. The higher the user's requirements for the protocol's anti-tampering verification effect, the larger the preset speech rate stability value and the smaller the preset conflict density value.

[0049] Specifically, when performing separation difficulty analysis, the separation difficulty characterization value is determined based on voiceprint similarity and interruption frequency; When the separation difficulty representation value is less than or equal to the preset separation difficulty representation value, the valid audio segments are stored on the blockchain for evidence. The separation difficulty characterization value is positively correlated with voiceprint similarity and interruption frequency.

[0050] The separation difficulty representation value is calculated as follows: α1 × voiceprint similarity / preset voiceprint similarity + α2 × interruption frequency / preset interruption frequency. Here, α1 and α2 are weighting coefficients, α1 + α2 = 1. It is easy to understand that the greater the influence of voiceprint similarity on the separation difficulty representation value, the larger the value of α1; conversely, the greater the influence of interruption frequency on the separation difficulty representation value, the larger the value of α2. This invention provides a set of values ​​for α1 and α2, where α1 is 0.5 and α2 is 0.5.

[0051] Obtain the voiceprint feature vector of the speaker within the overlapping segment. For a single overlapping segment, if there are only two speakers, the cosine similarity of the voiceprint feature vector is recorded as the voiceprint similarity within the overlapping segment. If there are three or more speakers, the cosine similarity of two speakers is obtained to generate a cosine similarity set. For a single speaker, the cosine similarity of that speaker with all other speakers is obtained, and the maximum value in the cosine similarity set is recorded as the voiceprint similarity of the overlapping segment. For a single valid audio segment, the number of overlapping segments is recorded as the interruption frequency of the valid audio segment.

[0052] In this invention, the preset voiceprint similarity value is 0.6, and the preset interruption frequency value is 4, with the unit being "number of interruptions". It is easy to understand that voiceprint similarity is used to characterize the degree of similarity in voiceprint features among different speakers. If the voiceprint similarity is higher, it indicates that the timbre of different speakers is closer, and the separation difficulty is greater. The interruption frequency characterizes the frequency of switching during the speaker's speech. The higher the interruption frequency, the higher the switching frequency, and the greater the separation difficulty. Therefore, the greater the user's tolerance for the influence of voiceprint similarity on the separation difficulty characterization value, the greater the preset voiceprint similarity value. Similarly, the greater the user's tolerance for the influence of interruption frequency on the separation difficulty characterization value, the greater the preset interruption frequency value.

[0053] When the separation difficulty representation value is greater than the preset separation difficulty representation value, the valid audio segments are stored normally, and the valid audio segments with separation difficulty representation values ​​less than or equal to the preset separation difficulty representation value are recorded as auxiliary verification segments, and the auxiliary verification segments are stored on the blockchain for evidence.

[0054] In this invention, ordinary storage refers to storing storage data in a distributed storage system, which includes, but is not limited to, HDFS.

[0055] Specifically, when the target mediation agreement is not signed, the auxiliary verification paragraph is hashed to generate an audio fingerprint and recorded on the blockchain. The audio fingerprint, the document fingerprint of the target mediation agreement before signing, and the document fingerprint after signing are compared and verified to output a verification report.

[0056] Before the parties sign, the target mediation agreement undergoes an initial hash calculation to generate document fingerprint A, which is recorded on the blockchain. Simultaneously, the signing information is also stored on the blockchain as evidence, including the signing time, signatory identity, signing IP address, and device fingerprint. For the signed target mediation agreement, the final document's hash value is calculated again to generate document fingerprint B, which is then compared with the initial hash value to verify whether the document content has changed during the signing process. For verified target mediation agreements, an electronic certificate with the verification conclusion is generated, facilitating evidence submission in subsequent legal proceedings.

[0057] Hash calculations are performed on the selected auxiliary verification segments to generate audio fingerprint C. Audio fingerprint C is then verified against document fingerprint A and document fingerprint B to improve the verification accuracy of the target mediation protocol.

[0058] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain, characterized in that, include: The registration and mediation protocol is determined based on the comparison results between the text similarity and the preset text similarity, and the reference keywords of the target mediation protocol are determined based on the registration and mediation protocol. For reference keywords, determine the audio segmentation strategy based on the associated reference values; The audio segmentation strategies include a first segmentation strategy that segments the target audio segment based on associated reference values ​​and a second segmentation strategy that performs adaptive segmentation on the target audio segment. When implementing the first segmentation strategy, relevant keywords for the number of seats are selected based on co-occurrence intensity, and audio segments are determined based on reference keywords and relevant keywords. The question is whether to increase the number of seats based on the continuity reference value. When implementing the second partitioning strategy, the decision on whether to perform clustering is based on the comparison between the keyword distribution representation value and the preset keyword distribution representation value. For the valid audio segments obtained after segmentation, the quality assessment status of the valid audio segments is determined based on speech rate stability and conflict density, and the processing path is determined according to the quality assessment status: either perform separation difficulty analysis or normal storage. When performing separation difficulty analysis, the separation difficulty characterization value is determined based on voiceprint similarity and interruption frequency. And based on the comparison between the separation difficulty characterization value and the preset separation difficulty characterization value, it is determined whether to change the audio storage method from on-chain evidence storage to ordinary storage; Before the target mediation agreement is signed, the target mediation agreement and the auxiliary verification paragraphs will be stored on the blockchain for evidence. Association reference value = maximum co-occurrence strength / number of reference keywords; For a single valid audio segment, the conflict density = duration of overlapping segments / duration of the valid audio segment; Valid audio segments whose separation difficulty characterization value is less than or equal to the preset separation difficulty characterization value are recorded as auxiliary verification segments; Before the parties sign, the target mediation agreement is hashed for the first time to generate document fingerprint A and recorded on the blockchain. At the same time, the signing information is stored on the blockchain as evidence. For the signed target mediation agreement, the hash value of the final signed document is recalculated to generate document fingerprint B, and compared with the initial hash value to verify whether the document content has changed during the signing process; Hash the selected auxiliary verification segments to generate an audio fingerprint C, and then verify the audio fingerprint C with the document fingerprint A and the document fingerprint B.

2. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 1, characterized in that, Historical mediation agreements with text similarity greater than the preset text similarity are recorded as the registered mediation agreements of the target mediation agreement, and the keywords of each registered mediation agreement are recorded as the reference keywords of the target mediation agreement.

3. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 2, characterized in that, When the correlation reference value is greater than the preset correlation reference value, the paragraph segmentation is determined based on the co-occurrence intensity.

4. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 3, characterized in that, When executing the first segmentation strategy, the effective audio segments of the reference keywords are obtained in descending order of co-occurrence intensity. For a single reference keyword, a number of related keywords are selected to construct the shortest audio segment that includes the reference keyword and the number of related keywords. The first segmentation strategy is to segment paragraphs based on co-occurrence intensity.

5. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 4, characterized in that, When the penetration reference value is greater than the preset penetration reference value, it is determined that the capacity should be increased. If the associated reference value is less than or equal to the preset associated reference value or the through reference value is less than or equal to the preset through reference value, then there is no need to adjust the capacity.

6. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 2, characterized in that, When the associated reference value is less than or equal to the preset associated reference value, the audio segmentation strategy is determined to be adaptive segmentation for the target audio segment.

7. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 6, characterized in that, Specifically, when the keyword distribution representation value is greater than the preset keyword distribution representation value, it is determined that keyword clustering is performed based on time intervals to obtain several valid audio segments.

8. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 3 or 6, characterized in that, For valid audio segments whose quality assessment status is that the speech rate stability is greater than the preset speech rate stability and the conflict density is less than or equal to the preset conflict density, the processing path is to perform separation difficulty analysis. For valid audio segments whose quality assessment status is that the speech rate stability is less than or equal to the preset speech rate stability or the conflict density is greater than the preset conflict density, the analysis strategy is to use normal storage.

9. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 8, characterized in that, When performing separation difficulty analysis, the separation difficulty characterization value is determined based on voiceprint similarity and interruption frequency. When the separation difficulty representation value is less than or equal to the preset separation difficulty representation value, the valid audio segments are stored on the blockchain for evidence. The separation difficulty characterization value is positively correlated with voiceprint similarity and interruption frequency.

10. The tamper-proof verification method for electronic mediation protocols based on double hashing and blockchain according to claim 1, characterized in that, Before the target mediation agreement is signed, the auxiliary verification paragraphs are hashed to generate an audio fingerprint, which is then recorded on the blockchain. The audio fingerprint, the document fingerprint of the target mediation agreement before signing, and the document fingerprint after signing are compared and verified to output a verification report.

Images