Artificial intelligence cloud service platform based on block chain big data

Abstract

The invention discloses an artificial intelligence cloud service platform based on block chain big data, and relates to the technical field of data processing, and the platform comprises four modules: a big data storage module processes multi-source data into data blocks; the block chain security construction module stores the data blocks into security data blocks in a chained manner; the optimal solution data retrieval module obtains a user problem, constructs a multi-dimensional coordinate when a single problem exists, and screens security data blocks through distance calculation and correlation degree verification; the similarity is analyzed when there are multiple problems, and screening is performed after verification precision is dynamically adjusted; and the artificial intelligence service platform receives the data and generates answers through a large model. Related data are accurately screened through random number assignment, coordinate construction and the like through the optimal solution data retrieval module, answer redundancy is avoided, a foundation is laid for accurate answers, and when multiple questions exist, the platform dynamically adjusts verification precision and adapts question logic according to similarity, answer deviation is reduced, and accuracy is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of data processing, and particularly relates to an artificial intelligence cloud service platform based on block chain big data. BACKGROUND

[0002] The artificial intelligence cloud service platform based on block chain big data integrates the distributed accounting, tamper-proof characteristics of the block chain and the massive information processing capacity of big data, deeply analyzes data by means of an artificial intelligence algorithm, and can provide safe and reliable cloud services for users, such as guaranteeing privacy and traceability during data sharing, optimizing resource allocation and predicting trends through intelligent analysis, and is widely applied to fields such as financial risk control and supply chain management, so that efficient, safe and intelligent data analysis and service support are realized.

[0003] For example, the artificial intelligence reasoning system based on the block chain provided in Chinese Patent Publication No. CN115049061A receives a second reasoning request sent by a consensus node cluster through a target accounting node during intelligent reasoning, the second reasoning request is generated according to item characteristic data carried in a first reasoning request by the consensus node cluster when the first reasoning request is confirmed by block, and the first reasoning request is generated by a reasoning application end according to the item characteristic data and a model identifier of a target reasoning model, and the second reasoning request includes the item characteristic data; the item characteristic data is subjected to reasoning operation by the target reasoning model to generate a reasoning result; and the reasoning result is sent to a client through the cloud, which not only makes up for the singleness of node data in the traditional scheme, reduces the interaction amount between the edge node and the cloud, and improves the processing efficiency and accuracy of the reasoning task.

[0004] However, the above scheme does not consider that when a user inputs a question in the artificial intelligence service platform, the artificial intelligence service platform needs to traverse all data corresponding to the question from the cloud until relevant data is traversed, and then the relevant data is sent to the service platform and processed by an artificial intelligence large model to output an answer, but the answer content is expanded too much, even deviating from the user's question, so that the accuracy of the answer is reduced, and the service quality of the artificial intelligence service platform is reduced. SUMMARY

[0005] Technical problems solved

[0006] In view of the deficiencies in the prior art, the application provides an artificial intelligence cloud service platform based on block chain big data, which solves the problem that after a user asks a question in the artificial intelligence service platform, the platform traverses relevant data from the cloud, processes the data by a large model to output an answer, but the answer content is expanded too much or even deviates from the question, resulting in a decrease in accuracy and a decrease in service quality of the platform.

[0007] Technical scheme

[0008] To achieve the above object, the present application is implemented by the following technical solutions: A blockchain big data-based artificial intelligence cloud service platform comprises the following specific modules: a big data storage module: multiple source data are constructed into a unified storage architecture through big data technology, and after format conversion and cleaning, the data are stored into a cloud database to form data blocks; a blockchain security construction module: through distributed node consensus accounting, each data block is associated with a previous block hash value to form a chain structure, and a secure data block is constructed; an optimal solution data retrieval module: within a preset time, user problem data are obtained and the number is counted, if the number is one, a multi-dimensional problem coordinate and a multi-dimensional answer coordinate are established, and the shortest path is calculated to obtain the best multi-dimensional answer coordinate, then the correlation degree is verified, if the verification is successful, based on the best multi-dimensional answer coordinate, the shortest new coordinate is continued to be calculated, and the process is repeated until the verification fails, if the verification fails, the secure data blocks corresponding to the best multi-dimensional answer coordinates that have passed the verification are all sent to an artificial intelligence service platform, if the number is greater than one, problem similarity analysis is performed, if the similarity is high, the correlation degree verification accuracy is improved, and re-verification is performed until there is no new user problem data within the preset time, if the similarity is low, the number is forced to be equal to one, and the cycle is continued; an artificial intelligence service platform: receiving the secure data blocks, generating answer output through a large model, and returning to the optimal solution data retrieval module.

[0009] Further, the specific steps of establishing the multi-dimensional question coordinate and the multi-dimensional answer coordinate are as follows: random numbers are assigned to the letters or Chinese characters of the user question data and the security data block, the random numbers of different letters or Chinese characters are all unique, the random numbers of the user question data are combined to obtain a question array, and the random numbers of the security data block are combined to obtain an allocation array; the random numbers in the question array and the allocation array are counted to obtain the random number quantity of the question array and the random number quantity of the allocation array; the random number quantity of the question array and the random number quantity of the allocation array are compared, if the random number quantity of the question array is equal to the random number quantity of the allocation array, the random number quantity of the question array or the random number quantity of the allocation array is not supplemented, if the random number quantity of the question array is not equal to the random number quantity of the allocation array, the random number quantity of the question array is supplemented so that the random number quantity of the question array is equal to the random number quantity of the allocation array, or the random number quantity of the question array is greater than the random number quantity of the allocation array, then the random number quantity of the allocation array is supplemented so that the random number quantity of the allocation array is equal to the random number quantity of the question array, and the supplemented value is zero; the multi-dimensional question coordinate is established according to the random number quantity of the question array and the random numbers of the question array, the dimension of the multi-dimensional question coordinate is equal to the random number quantity of the question array, and the value of each dimension of the multi-dimensional question coordinate is equal to one of the random numbers of the question array; and the multi-dimensional answer coordinate is established according to the random number quantity of the allocation array and the random numbers of the allocation array, the dimension of the multi-dimensional answer coordinate is equal to the random number quantity of the allocation array, and the value of each dimension of the multi-dimensional answer coordinate is equal to one of the random numbers of the allocation array.

[0010] Further, the specific acquisition method of the optimal multi-dimensional answer coordinate is as follows: the multi-dimensional question coordinate is set as (v1, v2,..., v n ), the multi-dimensional answer coordinate is set as (p1, p2,..., p n ), and the quantity of the multi-dimensional answer coordinate is counted; the distance calculation is performed on the multi-dimensional question coordinate and the multi-dimensional answer coordinate to obtain the distance of the multi-dimensional question coordinate and the multi-dimensional answer coordinate; the distance of the multi-dimensional question coordinate and the multi-dimensional answer coordinate is counted according to the quantity of the multi-dimensional answer coordinate to obtain a distance array of the multi-dimensional question coordinate and the multi-dimensional answer coordinate; the distance of the multi-dimensional question coordinate and the multi-dimensional answer coordinate in the distance set of the multi-dimensional question coordinate and the multi-dimensional answer coordinate is arranged in descending order through a sorting algorithm to obtain the shortest distance of the multi-dimensional question coordinate and the multi-dimensional answer coordinate, and the multi-dimensional answer coordinate corresponding to the distance of the multi-dimensional question coordinate and the multi-dimensional answer coordinate is recorded as the optimal multi-dimensional answer coordinate.

[0011] Further, the specific method of the distance calculation on the multi-dimensional question coordinate and the multi-dimensional answer coordinate is as follows: the distance calculation is performed on the multi-dimensional question coordinate and a certain multi-dimensional answer coordinate through the Euclidean distance formula to obtain the distance of the multi-dimensional question coordinate and the multi-dimensional answer coordinate. wherein, JL represents the distance between the multi-dimensional question coordinate and the multi-dimensional answer coordinate, v1 represents the value of the first dimension of the multi-dimensional question coordinate, v2 represents the value of the second dimension of the multi-dimensional question coordinate, v n represents the value of the n-th dimension of the multi-dimensional question coordinate, p1 represents the value of the first dimension of the multi-dimensional answer coordinate, p2 represents the value of the second dimension of the multi-dimensional answer coordinate, p n represents the value of the n-th dimension of the multi-dimensional answer coordinate.

[0012] Further, the specific steps of the correlation verification are as follows: constructing vectors for the multi-dimensional question coordinate and the optimal multi-dimensional answer coordinate to obtain a multi-dimensional question vector and an optimal multi-dimensional answer vector; calculating the multi-dimensional question vector and the optimal multi-dimensional answer vector by using the cosine formula to obtain the correlation; setting a correlation threshold, comparing the correlation with the correlation threshold, if the correlation is within the correlation threshold, it indicates that the verification is successful, if the correlation is outside the correlation threshold, it indicates that the verification fails.

[0013] Further, the specific steps of constructing vectors for the multi-dimensional question coordinate and the optimal multi-dimensional answer coordinate are as follows: respectively counting the number of values of the same size in the multi-dimensional question coordinate and the optimal multi-dimensional answer coordinate to obtain the number of values of the same size in the multi-dimensional question coordinate and the number of values of the same size in the optimal multi-dimensional answer coordinate; respectively counting the number of values of different sizes in the multi-dimensional question coordinate and the optimal multi-dimensional answer coordinate to obtain the number of values of different sizes in the multi-dimensional question coordinate and the number of values of different sizes in the optimal multi-dimensional answer coordinate; constructing an A coordinate, the dimension of the A coordinate is the number of values of different sizes in the multi-dimensional question coordinate, the value in each dimension of the A coordinate is the number of values of the same size in the multi-dimensional question coordinate, constructing a B coordinate, the dimension of the B coordinate is the number of values of different sizes in the optimal multi-dimensional answer coordinate, the value in each dimension of the B coordinate is the number of values of the same size in the optimal multi-dimensional answer coordinate; arranging the values of the same size in the multi-dimensional question coordinate and the optimal multi-dimensional answer coordinate in the same dimension of the A coordinate and the B coordinate; comparing the dimensions of the A coordinate and the B coordinate, if the dimensions of the A coordinate and the B coordinate are the same, do not supplement zero values to the dimensions of the A coordinate or the B coordinate, if the dimension of the A coordinate is greater than the dimension of the B coordinate, supplement zero values to the dimensions of the B coordinate to make the dimensions of the A coordinate and the B coordinate the same, if the dimension of the A coordinate is less than the dimension of the B coordinate, supplement zero values to the dimensions of the A coordinate; calculating the difference between the A coordinate and the zero coordinate to obtain the multi-dimensional question vector, calculating the difference between the B coordinate and the zero coordinate to obtain the optimal multi-dimensional answer vector.

[0014] Further, the specific method of obtaining the correlation is as follows: wherein, cosβ represents the correlation, represents a multi-dimensional problem vector, represents an optimal multi-dimensional answer vector, represents a module of the multi-dimensional problem vector, represents a module of the optimal multi-dimensional answer vector.

[0015] Further, the specific way of setting the correlation degree threshold is to set the correlation degree threshold in the open interval zero to the closed interval one to represent verification success, and in the closed interval negative one to the closed interval zero to represent verification failure.

[0016] Further, the specific steps of performing the question similarity analysis are: assigning a random number to each letter or Chinese character of the user question data, and the random numbers of different letters or Chinese characters are all unique; taking the previous user question data as J and the next user question data as K, and sequentially calculating the difference between the random numbers in J and the random numbers in K, if the difference between the random numbers in J and the random numbers in K is zero, it means that the random numbers in J and the random numbers in K are equal, if the difference between the random numbers in J and the random numbers in K is not zero, continue to sequentially calculate the difference between the random numbers in J and the random numbers in K until the random numbers in J and the random numbers in K are all calculated; counting the number of times that the difference between the random numbers in J and the random numbers in K is zero to obtain the same calculation times, counting the number of times of calculating the difference between the random numbers in J and the random numbers in K to obtain the total calculation times, and calculating the quotient of the same calculation times and the total calculation times to obtain the similarity; comparing the similarity with one-half, if the similarity is greater than or equal to one-half, it means that the similarity is high, if the similarity is less than one-half, it means that the similarity is low.

[0017] Further, the specific way of improving the correlation degree verification accuracy is: increasing a user question data with high similarity within a preset time, and then removing the new optimal multi-dimensional answer coordinate with the farthest distance from the multi-dimensional problem coordinate.

[0018] Beneficial effects

[0019] Compared with the prior art, the embodiments of the present application have at least the following advantages or beneficial effects:

[0020] 1. The optimal solution data retrieval module assigns a unique random number to the letters or Chinese characters of the user question and the security data block, constructs the multi-dimensional problem coordinate and the multi-dimensional answer coordinate, calculates and selects the optimal multi-dimensional answer coordinate corresponding to the shortest path by using the Euclidean distance formula, and then performs correlation degree verification by using the cosine angle formula. This process accurately selects from the data matching source, ensures that the selected security data block is highly related to the user question, effectively avoids the problem of excessive expansion of answers caused by the introduction of a large amount of irrelevant information during traditional data traversal, and lays a data foundation for the artificial intelligence large model to generate accurate answers.

[0021] 2. When the number of user questions is greater than one, the platform analyzes the similarity of the questions, and for consecutive questions with high similarity, the accuracy of the correlation verification is improved, and for questions with low similarity, they are treated as new questions and processed again. This dynamic adjustment mechanism can adapt to the user's questioning logic, optimize the data retrieval accuracy for possible supplementary or similar questions of the user, further reduce the situation of the answer deviating from the question, and improve the accuracy and pertinence of the output answer of the artificial intelligence service platform.

[0022] Of course, implementing any product of the present application does not necessarily require all the advantages described above to be achieved at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0023] Fig. 1 For the present application: a flow chart of an artificial intelligence cloud service platform based on blockchain big data.

[0024] Fig. 2 For the present application: a structure diagram of an artificial intelligence cloud service platform based on blockchain big data. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0026] It should be noted that in this paper, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0027] Embodiment 1:

[0028] As shown in the figure, the embodiment of the present application provides an artificial intelligence cloud service platform based on blockchain big data: comprising the following specific modules: Figs. 1-2

[0029] 1. An artificial intelligence cloud service platform based on blockchain big data, characterized by comprising the following specific modules:

[0030] ​Big data storage module: using big data technology to build a unified data storage architecture for the multi-source data obtained, realizing centralized management of multi-source data, improving data access efficiency, facilitating quick calling and integrated analysis, and performing format conversion and data cleaning to eliminate format differences, redundancies and error information of multi-source data, ensuring the accuracy of multi-source data, storing to the cloud database to obtain data blocks.

[0031] Blockchain security construction module: the data blocks are recorded through distributed node consensus, avoiding single node tampering with data, realizing decentralized trust of data storage, ensuring data authenticity, each data block is associated with the hash value of the previous block to form a chain structure, and the data source and changes are traced through the hash value association, once the data is tampered with, it can be quickly found, ensuring the data tamper-proof and integrity, and constructing a secure data block.

[0032] Optimal solution data retrieval module: obtaining user question data within a preset time, and counting the number of user question data, when the number of user question data is equal to one, that is, the first question asked by the user within the preset time, establishing multi-dimensional question coordinates and multi-dimensional answer coordinates for user question data and secure data blocks, and calculating the shortest path multi-dimensional answer coordinates, which are recorded as the best multi-dimensional answer coordinates. The best multi-dimensional answer coordinates are verified for correlation degree, if the verification is successful, the shortest new best multi-dimensional answer coordinates are calculated from the best multi-dimensional answer coordinates. Since the best multi-dimensional answer coordinates are the shortest path relative to the multi-dimensional question coordinates, if the multi-dimensional question coordinates are analyzed in breadth from other multi-dimensional answer coordinates, the accuracy is reduced, so the shortest new best multi-dimensional answer coordinates are calculated from the best multi-dimensional answer coordinates, which is to further improve the accuracy based on the high accuracy, that is, to improve the accuracy by depth analysis, and so on until the verification fails. If the verification fails, the secure data block corresponding to the best multi-dimensional answer coordinates that have been verified successfully is sent to the artificial intelligence service platform. When the number of user question data is greater than one, that is, the user asks a new question after asking the first question within the preset time, the problem similarity analysis is performed, if the similarity is high, the accuracy of the correlation degree verification is improved, that is, if the similarity of the first question and the new question is high, it means that the user is not satisfied with the answer to the first question, and then asks a similar question to the first question, indicating that the accuracy of the correlation degree verification needs to be improved, so the correlation degree verification is performed again. Until no user question data is obtained within the preset time, that is, the user gets a satisfactory answer and no longer asks questions, if the similarity is low, the number of user question data is forced to be equal to one, indicating that the user asks a new question and has nothing to do with the first question, then the cycle continues.

[0033] Artificial intelligence service platform: receiving secure data blocks and generating answer output through artificial intelligence big model, and returning to optimal solution data retrieval module.

[0034] Example 2 differs from example 1 in that:

[0035] The specific steps of establishing multi-dimensional question coordinates and multi-dimensional answer coordinates for user question data and security data blocks are as follows:

[0036] Random numbers are assigned to the letters or Chinese characters of the user question data and security data blocks by a random number generator, and the random numbers of different letters or Chinese characters are unique. The random numbers of the user question data are combined to obtain a question array, and the random numbers of the security data blocks are combined to obtain an allocation array;

[0037] The number of random numbers in the question array and the number of random numbers in the allocation array are counted;

[0038] The number of random numbers in the question array and the number of random numbers in the allocation array are compared. If the number of random numbers in the question array is equal to the number of random numbers in the allocation array, no supplement is made to the number of random numbers in the question array or the number of random numbers in the allocation array. If the number of random numbers in the question array is not equal to the number of random numbers in the allocation array, i.e., the number of random numbers in the question array is less than the number of random numbers in the allocation array, the number of random numbers in the question array is supplemented so that the number of random numbers in the question array is equal to the number of random numbers in the allocation array, facilitating the uniformity of dimensions in subsequent coordinate construction. Or the number of random numbers in the question array is greater than the number of random numbers in the allocation array, the number of random numbers in the allocation array is supplemented so that the number of random numbers in the allocation array is equal to the number of random numbers in the question array, facilitating the uniformity of dimensions in subsequent coordinate construction. The supplemented value is zero, so that the random numbers in the question array or the random numbers in the allocation array will not be disturbed;

[0039] Multi-dimensional question coordinates are established according to the number of random numbers in the question array and the random numbers in the question array. The dimensions of the multi-dimensional question coordinates are equal to the number of random numbers in the question array, and the values of each dimension of the multi-dimensional question coordinates are equal to one of the random numbers in the question array. Multi-dimensional answer coordinates are established according to the number of random numbers in the allocation array and the random numbers in the allocation array. The dimensions of the multi-dimensional answer coordinates are equal to the number of random numbers in the allocation array, and the values of each dimension of the multi-dimensional answer coordinates are equal to one of the random numbers in the allocation array.

[0040] The specific way to obtain the optimal multi-dimensional answer coordinates is as follows:

[0041] Set the multi-dimensional question coordinates to (v1, v2,..., v n ), the multi-dimensional answer coordinates to (p1, p2,..., p n ), and count the number of multi-dimensional answer coordinates;

[0042] The distance between the multi-dimensional question coordinates and the multi-dimensional answer coordinates is calculated.

[0043] Based on the number of multidimensional answer coordinates, the distance between the multidimensional question coordinates and the multidimensional answer coordinates is statistically analyzed to obtain a distance array between the multidimensional question coordinates and the multidimensional answer coordinates;

[0044] Bubble sort is used to arrange the distances between multidimensional problem coordinates and multidimensional answer coordinates in the set of distances from multidimensional problem coordinates to multidimensional answer coordinates in descending order. That is, by comparing and swapping the distances of adjacent multidimensional problem coordinates and multidimensional answer coordinates, the smallest distance between the multidimensional problem coordinates and multidimensional answer coordinates is gradually moved to the end of the array, and finally the sorting is achieved from largest to smallest. The multidimensional answer coordinate corresponding to this distance is recorded as the optimal multidimensional answer coordinate.

[0045] The specific method for calculating the distance between the coordinates of the multidimensional question and the coordinates of the multidimensional answer is as follows:

[0046] The Euclidean distance formula is used to calculate the distance between the coordinates of a multidimensional problem and the coordinates of a multidimensional answer. This method is applicable to calculating the straight-line distance between two coordinates in a multidimensional space, and thus obtains the distance between the coordinates of the multidimensional problem and the coordinates of the multidimensional answer.

[0047]

[0048] Where JL represents the distance between the coordinates of the multidimensional question and the coordinates of the multidimensional answer, v1 represents the value of the first dimension of the multidimensional question coordinates, v2 represents the value of the second dimension of the multidimensional question coordinates, and v... n Let p1 represent the value of the nth dimension of the multidimensional question coordinates, p2 represent the value of the first dimension of the multidimensional answer coordinates, and p3 represent the value of the second dimension of the multidimensional answer coordinates. n This represents the value of the nth dimension of the multidimensional answer coordinates, when v i equals p j When, then (v i -p j ) 2 A value of zero means that the security data block contains letters or Chinese characters that are identical to those in the user question data. Assuming the proportion of identical letters or Chinese characters in the security data block to those in the user question data is higher, then (v i -p j ) 2 The greater the proportion of zero values, the shorter the distance between the multidimensional question coordinates and the multidimensional answer coordinates.

[0049] The specific steps for verifying the correlation of the best multidimensional answer coordinates are as follows:

[0050] A vector is constructed for the multi-dimensional question coordinates and the optimal multi-dimensional answer coordinates respectively to obtain a multi-dimensional question vector and an optimal multi-dimensional answer vector;

[0051] The multi-dimensional question vector and the optimal multi-dimensional answer vector are calculated by a cosine formula to obtain a correlation degree;

[0052] A correlation degree threshold is set, and the correlation degree is compared with the correlation degree threshold. If the correlation degree is within the correlation degree threshold, it indicates that the verification is successful, that is, the multi-dimensional question vector and the optimal multi-dimensional answer vector have a high correlation degree. If the correlation degree is outside the correlation degree threshold, it indicates that the verification fails, that is, the multi-dimensional question vector and the optimal multi-dimensional answer vector have a low correlation degree.

[0053] The specific steps of constructing the vector for the multi-dimensional question coordinates and the optimal multi-dimensional answer coordinates are as follows:

[0054] The same size values in the multi-dimensional question coordinates and the optimal multi-dimensional answer coordinates are counted respectively to obtain the number of the same size values in the multi-dimensional question coordinates and the number of the same size values in the optimal multi-dimensional answer coordinates;

[0055] The different size values in the multi-dimensional question coordinates and the optimal multi-dimensional answer coordinates are counted respectively to obtain the number of the different size values in the multi-dimensional question coordinates and the number of the different size values in the optimal multi-dimensional answer coordinates;

[0056] An A coordinate is constructed, the dimension of the A coordinate is the number of the different size values in the multi-dimensional question coordinates, and the value in each dimension of the A coordinate is the number of the same size values in the multi-dimensional question coordinates. A B coordinate is constructed, the dimension of the B coordinate is the number of the different size values in the optimal multi-dimensional answer coordinates, and the value in each dimension of the B coordinate is the number of the same size values in the optimal multi-dimensional answer coordinates;

[0057] The same size values in the multi-dimensional question coordinates and the optimal multi-dimensional answer coordinates are arranged in the same dimension of the A coordinate and the B coordinate, thereby reflecting the proportional relationship of the value M in the multi-dimensional question coordinates and the value M in the optimal multi-dimensional answer coordinates in the same dimension;

[0058] The dimensions of the A coordinate and the B coordinate are compared. If the dimensions of the A coordinate and the B coordinate are the same, no zero value is added to the dimension of the A coordinate or the B coordinate. If the dimension of the A coordinate is greater than the dimension of the B coordinate, a zero value is added to the dimension of the B coordinate to make the dimensions of the A coordinate and the B coordinate the same. If the dimension of the A coordinate is less than the dimension of the B coordinate, a zero value is added to the dimension of the A coordinate;

[0059] The A coordinate and a zero coordinate are calculated by difference to obtain a multi-dimensional question vector, and the B coordinate and the zero coordinate are calculated by difference to obtain an optimal multi-dimensional answer vector.

[0060] The specific acquisition method of the correlation degree is as follows:

[0061]

[0062] wherein cosβ represents the correlation degree, represents a multi-dimensional problem vector, represents an optimal multi-dimensional answer vector, represents a modulus of the multi-dimensional problem vector, represents a modulus of the optimal multi-dimensional answer vector, The denominator is used for normalization processing, so that the range of the correlation degree is between negative one and positive one. When the proportional relationship of the values of each dimension of the A coordinate and the B coordinate is more consistent, the direction of the multi-dimensional problem vector and the optimal multi-dimensional answer vector is more consistent, and the correlation degree is higher.

[0063] The specific way of setting the correlation degree threshold is as follows:

[0064] Since the correlation degree is between negative one and positive one, the correlation degree threshold is set between the open interval zero and the closed interval one to represent a verification success, and between the closed interval negative one and the closed interval zero to represent a verification failure.

[0065] The specific steps of the question similarity analysis are as follows:

[0066] Since the number of user question data is greater than one, and a random number is assigned to each letter or Chinese character of the user question data, the random numbers of different letters or Chinese characters are all unique;

[0067] The last user question data is recorded as J, and the next user question data is recorded as K. The random numbers in J and K are sequentially calculated by difference. If the difference between the random numbers in J and K is zero, it means that the random numbers in J and K are equal, that is, there is the same letter or Chinese character in the last user question data and the next user question data. If the difference between the random numbers in J and K is not zero, the random numbers in J and K are sequentially calculated by difference until the random numbers in J and K are all calculated.

[0068] The number of times that the difference between the random numbers in J and K is zero is counted to obtain the same calculation times. The number of times that the random numbers in J and K are calculated by difference is counted to obtain the total calculation times. The same calculation times and the total calculation times are calculated by quotient to obtain the similarity.

[0069] The similarity is compared with one-half. If the similarity is greater than or equal to one-half, it means that the similarity is high. If the similarity is less than one-half, it means that the similarity is low.

[0070] The specific way to improve the accuracy of the correlation verification is:

[0071] Although the new optimal multi-dimensional answer coordinate is associated with the multi-dimensional question coordinate, the more times the optimal multi-dimensional answer coordinate is obtained, the farther the new optimal multi-dimensional answer coordinate is from the multi-dimensional question coordinate, that is, the more other data not associated with the multi-dimensional question coordinate is mixed in the new optimal multi-dimensional answer coordinate. Therefore, in order to reduce the new optimal multi-dimensional answer coordinate mixed with other data not associated with the multi-dimensional question coordinate, the number of times the optimal multi-dimensional answer coordinate is obtained is reduced, that is, a user question data with high similarity is added within a preset time, and the new optimal multi-dimensional answer coordinate farthest from the multi-dimensional question coordinate is removed.

[0072] The preferred embodiments of the application disclosed above are only used to help explain the application. The preferred embodiments do not describe all the details and limit the application to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of the specification. The specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the application, so that those skilled in the art can well understand and utilize the application. The application is limited by the claims and their entire scope and equivalents.

Claims

1.A blockchain big data based artificial intelligence cloud service platform, characterized in that: The specific modules include the following: A big data storage module: a unified storage architecture is constructed for multi-source data through big data technology, and after format conversion and cleaning, the data is stored in a cloud database to form data blocks; A blockchain security construction module: through distributed node consensus accounting, each data block is associated with the hash value of the previous block to form a chain structure, and a secure data block is constructed; An optimal solution data retrieval module: within a preset time, user problem data is obtained and the number is counted, if the number is one, multi-dimensional problem coordinates and multi-dimensional answer coordinates are established, and the shortest path is calculated to obtain the best multi-dimensional answer coordinates, then the correlation verification is performed, if the verification is successful, the best multi-dimensional answer coordinates are used to continue to calculate the shortest new coordinates, and the process is repeated until the verification fails, if the verification fails, the secure data blocks corresponding to the best multi-dimensional answer coordinates that have passed the verification are sent to an artificial intelligence service platform, if the number is greater than one, the problem similarity is analyzed, if the similarity is high, the correlation verification accuracy is improved, and the verification is repeated until there is no new user problem data within the preset time, if the similarity is low, the number is forced to be equal to one, and the cycle is continued; An artificial intelligence service platform: receiving the secure data blocks, generating answer output through a large model, and returning to the optimal solution data retrieval module. 2.The blockchain big data-based artificial intelligence cloud service platform of claim 1, wherein: The specific steps of establishing multi-dimensional problem coordinates and multi-dimensional answer coordinates are as follows: Random numbers are assigned to letters or Chinese characters of user problem data and secure data blocks, and the random numbers of different letters or Chinese characters are unique, the random numbers of the user problem data are combined to obtain a problem array, and the random numbers of the secure data blocks are combined to obtain an allocation array; The random numbers in the problem array and the allocation array are counted to obtain the number of random numbers in the problem array and the number of random numbers in the allocation array; The number of random numbers in the problem array and the number of random numbers in the allocation array are compared, if the number of random numbers in the problem array is equal to the number of random numbers in the allocation array, the number of random numbers in the problem array or the number of random numbers in the allocation array is not supplemented, if the number of random numbers in the problem array is not equal to the number of random numbers in the allocation array, the number of random numbers in the problem array is supplemented to make the number of random numbers in the problem array equal to the number of random numbers in the allocation array, or the number of random numbers in the allocation array is supplemented to make the number of random numbers in the allocation array equal to the number of random numbers in the problem array, and the supplemented value is zero; Multi-dimensional problem coordinates are established according to the number of random numbers in the problem array and the random numbers in the problem array, the dimension of the multi-dimensional problem coordinates is equal to the number of random numbers in the problem array, and the value of each dimension of the multi-dimensional problem coordinates is equal to one of the random numbers in the problem array, multi-dimensional answer coordinates are established according to the number of random numbers in the allocation array and the random numbers in the allocation array, the dimension of the multi-dimensional answer coordinates is equal to the number of random numbers in the allocation array, and the value of each dimension of the multi-dimensional answer coordinates is equal to one of the random numbers in the allocation array. 3.The blockchain big data-based artificial intelligence cloud service platform of claim 1, wherein: The specific acquisition method of the best multi-dimensional answer coordinates is as follows: Set the multi-dimensional question coordinates as (v1, v2, …, v n ), the multi-dimensional answer coordinates as (p1, p2, …, p n ), and count the number of multi-dimensional answer coordinates; The distance between the multi-dimensional problem coordinates and the multi-dimensional answer coordinates is calculated to obtain the distance between the multi-dimensional problem coordinates and the multi-dimensional answer coordinates. According to the number of multi-dimensional answer coordinates, the distance between the multi-dimensional question coordinates and the multi-dimensional answer coordinates is counted to obtain a distance array of the multi-dimensional question coordinates and the multi-dimensional answer coordinates; The distance between the multi-dimensional question coordinates and the multi-dimensional answer coordinates in the distance set is arranged in descending order by using a sorting algorithm to obtain the shortest distance between the multi-dimensional question coordinates and the multi-dimensional answer coordinates, and the multi-dimensional answer coordinates corresponding to the distance are recorded as the best multi-dimensional answer coordinates. 4.The blockchain big data-based artificial intelligence cloud service platform of claim 3, wherein: The specific manner of calculating the distance between the multi-dimensional question coordinates and the multi-dimensional answer coordinates is as follows: The distance between the multi-dimensional question coordinates and the multi-dimensional answer coordinates is calculated by using the Euclidean distance formula to obtain the distance between the multi-dimensional question coordinates and the multi-dimensional answer coordinates. wherein JL represents the distance between the multi-dimensional question coordinate and the multi-dimensional answer coordinate, vi represents the value of the first dimension of the multi-dimensional question coordinate, v2represents the value of the second dimension of the multi-dimensional question coordinate, vnrepresents the value of the n-th dimension of the multi-dimensional question coordinate, pi represents the value of the first dimension of the multi-dimensional answer coordinate, p2represents the value of the second dimension of the multi-dimensional answer coordinate, pnrepresents the value of the n-th dimension of the multi-dimensional answer coordinate. n wherein JL represents the distance between the multi-dimensional question coordinate and the multi-dimensional answer coordinate, vi represents the value of the first dimension of the multi-dimensional question coordinate, v2represents the value of the second dimension of the multi-dimensional question coordinate, vnrepresents the value of the n-th dimension of the multi-dimensional question coordinate, pi represents the value of the first dimension of the multi-dimensional answer coordinate, p2represents the value of the second dimension of the multi-dimensional answer coordinate, pnrepresents the value of the n-th dimension of the multi-dimensional answer coordinate. n wherein JL represents the distance 5.The blockchain big data-based artificial intelligence cloud service platform of claim 1, wherein: The specific steps of the correlation verification are as follows: Vectors are respectively constructed for the multi-dimensional question coordinates and the best multi-dimensional answer coordinates to obtain a multi-dimensional question vector and a best multi-dimensional answer vector; The correlation is calculated by using the cosine formula to obtain the correlation. The correlation threshold is set, and the correlation is compared with the correlation threshold. If the correlation is within the correlation threshold, the verification is successful, and if the correlation is outside the correlation threshold, the verification fails. 6.The blockchain big data-based artificial intelligence cloud service platform of claim 5, wherein: The specific steps of constructing vectors for the multi-dimensional question coordinates and the best multi-dimensional answer coordinates are as follows: The number of values of the same size in the multi-dimensional question coordinates and the best multi-dimensional answer coordinates is counted to obtain the number of values of the same size in the multi-dimensional question coordinates and the number of values of the same size in the best multi-dimensional answer coordinates. The number of values of different sizes in the multi-dimensional question coordinates and the best multi-dimensional answer coordinates is counted to obtain the number of values of different sizes in the multi-dimensional question coordinates and the number of values of different sizes in the best multi-dimensional answer coordinates. An A coordinate is constructed, the dimension of the A coordinate is the number of values of different sizes in the multi-dimensional question coordinates, and the value in each dimension of the A coordinate is the number of values of the same size in the multi-dimensional question coordinates. A B coordinate is constructed, the dimension of the B coordinate is the number of values of different sizes in the best multi-dimensional answer coordinates, and the value in each dimension of the B coordinate is the number of values of the same size in the best multi-dimensional answer coordinates. The values of the same size in the multi-dimensional question coordinates and the best multi-dimensional answer coordinates are arranged in the same dimension of the A coordinate and the B coordinate. The dimensions of the A coordinate and the B coordinate are compared. If the dimensions of the A coordinate and the B coordinate are the same, zero values are not supplemented to the dimensions of the A coordinate or the B coordinate. If the dimension of the A coordinate is greater than the dimension of the B coordinate, zero values are supplemented to the dimensions of the B coordinate to make the dimensions of the A coordinate and the B coordinate the same. If the dimension of the A coordinate is less than the dimension of the B coordinate, zero values are supplemented to the dimensions of the A coordinate. The A coordinate and the zero coordinate are subtracted to obtain the multi-dimensional question vector, and the B coordinate and the zero coordinate are subtracted to obtain the best multi-dimensional answer vector. 7.The blockchain big data-based artificial intelligence cloud service platform of claim 5, wherein: The specific manner of setting the correlation threshold is as follows: where cos β represents the degree of association, represents the multi-dimensional problem vector, represents the optimal multi-dimensional answer vector, represents the norm of the multi-dimensional problem vector, represents the norm of the optimal multi-dimensional answer vector. 8.The blockchain big data-based artificial intelligence cloud service platform of claim 5, wherein: The correlation threshold is set in the open interval zero to the closed interval one to represent successful verification, and in the closed interval negative one to the closed interval zero to represent failed verification. ​ 9.The blockchain big data-based artificial intelligence cloud service platform of claim 1, wherein: The specific steps of the question similarity analysis are as follows: Random numbers are assigned to letters or Chinese characters of each user question data, and the random numbers of different letters or Chinese characters are unique; The last user question data is recorded as J, and the next user question data is recorded as K, and the random numbers in J and K are sequentially calculated by difference, if the difference between the random numbers in J and K is zero, it means that the random numbers in J and K are equal, if the difference between the random numbers in J and K is not zero, the random numbers in J and K are sequentially calculated by difference, until the random numbers in J and K are all calculated; The number of times that the difference between the random numbers in J and K is zero is counted to obtain the same calculation times, the number of times that the random numbers in J and K are calculated by difference is counted to obtain the total calculation times, and the same calculation times and the total calculation times are calculated by quotient to obtain the similarity; The similarity is compared with one-half, if the similarity is greater than or equal to one-half, it means that the similarity is high, if the similarity is less than one-half, it means that the similarity is low. 10.The blockchain big data-based artificial intelligence cloud service platform of claim 1, wherein: The specific way of improving the association degree verification accuracy is as follows: Within a predetermined time, a user question data with high similarity is added, and the new best multi-dimensional answer coordinate with the farthest multi-dimensional question coordinate distance is removed.

Images