A method for constructing and analyzing a multidimensional topological structure of distributed data

Abstract

The present invention discloses a method for constructing and deconstructing a multidimensional topological structure of distributed data. Different data sequences of a data structure are arranged in parallel, and each data sequence forms a fixed length or an increasing length; each secondary sub-data system generates its own data sequence as the business develops; the time series data of each data sequence are anchored in sequence using the data structure I of the digital characteristic value of the data to generate data blocks, and the data structure II is used to provide more information about the preceding data of the data, and different data sequences respectively form the data structure III given by the present invention, and this data structure III transmits the characteristic values ​​between different data sequences for overall anchoring. When a trigger condition occurs, the data sequences mutually obtain the data structure IV, and the data sequences process the received data IV into the data structure V and save it. In the case of obtaining partial data, the process of verification and collection using various characteristic values ​​and data bodies of the data is iterated to complete the deconstruction.

Description

Technical Field

[0001] The present invention belongs to the field of computer technology, and in particular relates to a method for constructing and deconstructing a logical topological relationship of distributed data in an Internet environment. Background Art

[0002] Distributed applications in the computer Internet environment are both data producers and data users. The logical topology of data is directly related to the efficiency of data transmission and storage. Traditionally, predefined data structures are used to plan the logical topology of the entire data, and specific management units are used to ensure the consistency of data construction and distribution. Before rebuilding the data, data users need to obtain the key characteristic parameters of the data from the above management units to rebuild the data structure. In order to balance efficiency and versatility, predefined data structures are usually designed as fixed one-dimensional mapping relationships, which are difficult to expand and are only suitable for conventional applications with low evolution frequency.

[0003] Data structure is a way for computers to store and organize various types of data in a targeted manner according to different design requirements and application purposes. Data structure refers to a collection of data elements that have one or more specific relationships with each other, or a collection of element collections, etc. In general, a well-designed or selected data structure can bring higher operation, storage efficiency, and security. Data structures are often related to efficient retrieval algorithms and indexing technologies.

[0004] In distributed applications, data users evolve at different frequencies, and strictly predefined data topology mapping limits the scalability of distributed systems. Strictly predefined data has strong global dependencies, and the storage efficiency is not high when the data is highly local.

[0005] To resolve the above contradictions, the industry has adopted a relatively loose logical topology and verification method to construct data, such as hash linked lists, Merkle tree structures and other techniques, but these one-dimensional processing methods have the problem of insufficient dimensions. Different methods lead to different costs or efficiencies. In many cases, they will trigger recursive requests for data on the network, and the cost of data reconstruction is too high or the time is too long, which makes it unfeasible in large-scale distributed applications.

[0006] Therefore, the present invention proposes a method for constructing and deconstructing a multi-dimensional topological structure of distributed data, thereby improving the efficiency of data reconstruction and reducing the cost of data deconstruction. Summary of the invention

[0007] The purpose of the present invention is to provide a method for constructing and deconstructing a multidimensional topological structure of distributed data, which is suitable for applications where data is stored on distributed network nodes and then analyzed and reconstructed as a whole, i.e., deconstructed. This type of data structure can provide the possibility of fast and convenient construction and deconstruction of distributed data storage, especially in the form of time series databases, blockchains, etc.

[0008] In a given computer data system, fixed data or growing data can be divided into a two-dimensional data structure consisting of multiple interrelated sequences (taking two dimensions as an example, multiple dimensions can be expanded in this way). The topological nodes of different data in the two-dimensional data structure (F x , P x,n , H x,n ) is identified by the sequence where the data is located F x 、The position mark P in the sequence where the data is located x,n and the eigenvalue H of the data x,n Determine where x represents the number of the sequence in which the data is located, and n represents the number of the data in the sequence.

[0009] Different data sequences of the data structure are arranged in parallel, and each data sequence has a fixed length or increases in length; each secondary sub-data system generates its own data sequence B as the business develops. x , x = 1 ~ N, the time series data of each data sequence are anchored in sequence using the data structure I of the digital feature value of the data to generate data blocks, and the data structure II is used to provide more information about the preceding data of the data. Different data sequences respectively form the data structure III given by the present invention, and this data structure III transfers the feature values ​​between different data sequences for overall anchoring. When the trigger condition occurs (for example, the trigger conditions of each sequence may be different at intervals), the data sequences obtain the data structure IV from each other, and the data sequences process the received data IV into the data structure V and save it. The data structure III given by the present invention is generated by the interactive preservation of feature values ​​between the data sequences, such as Figure 3 The data structure III is distributed and stored in a network computer system. When obtaining partial data, the various characteristic values ​​and data bodies of the data are used to perform verification and collection process iterations to complete the deconstruction of the data structure. The specific steps of the method for constructing a multi-dimensional topological structure of distributed data of the present invention include:

[0010] 1) The data sequences B1, B2, B3, B4 in the data structure are N Select the corresponding packaging algorithm rules according to different business scenarios or other set criteria, which can be zip format compression, encryption algorithm or other custom algorithm rules. xThe data blocks in are anchored by digital feature values ​​in sequence, x = 1 to N. The data sequences B1, B2, B3, B4 N Finally, they are encapsulated into their own data dependency structures according to their own business characteristics, such as Figure 3 Among them, the data block C that constitutes the data sequence Block Include DATA x,n 、F x 、T x,n , H ComR , H x,(n-1) , H x,Co , R x,n There are 7 parts in total, including DATA x,n Refers to the nth business-related data block in the xth column, F x is the identifier of the data sequence Bx (Fx can be Bx, or a code or ordinal number), T x,n It is the business data block DATA x,n Packaging time, H ComR is the characteristic value of the data group (calculated by converting the digital features of multiple data blocks), H x,(n-1) is the digital eigenvalue of the preceding data block (i.e., the n-1th) of the nth data in the xth column, H x,Co is the identification digital feature of the preceding data combination of the nth data block in the xth column, R x,n The dimension information contained in the nth data block in the xth column, such as Figure 1 As shown. Among them, containing the same H ComR The data block of the value is a data group, H x,Co The digital eigenvalues ​​H of several preceding data blocks x,m+0 ,H x,m+1 ,…,H x,m+X The corresponding transformation operation is performed, such as Figure 2 As shown, the number of preceding data blocks is triggered by conditions (such as time intervals, etc.), wherein the H of the data block x,m+0 ,H x,m+1 ,…,H x,m+X Combined into digital eigenvalue combination H comseq , and H comseq Then obtain the digital feature H x,Co , which is also called the identifier of the preceding data combination.

[0011] 2) After the trigger condition is met (such as after a set time interval), the above B1~B N F of the data block of the data sequence x ,P x,n ,T x,n ,H x,n Combined into [F x ,P x,n ,Tx,n ,H x,n ], that is, C equ , prepare one of the sequences B stored in the data sequence X ,like Figure 4 Among them, C equ The identifier F of the data sequence Bx to which the nth data block in the xth column belongs x , the position P of the data block in the data sequence x,n , data encapsulation time T x,n And the data digital eigenvalue H x,n The identification combination data at least includes a combination of digital features of the data. The nodes of different data sequences may be the same or different, and are all used for processing the data sequence.

[0012] 3) Among them, data sequence B X Get all data sequence identifier combination data C equ After that, the corresponding identification combination data C is calculated equ The numerical eigenvalue H Cx ; The obtained identification combination-feature value pair [C equ ,H Cx ] or logo combination C equ or its characteristic value H Cx , write data sequence B X The newly generated data block; the obtained digital eigenvalue H C1 ,H C2 ,~,H CN Composition of eigenvalues ​​[H C1 ,H C2 ,~,H CN ], that is, H combin ,like Figure 5 , and calculate the eigenvalue combination H combin The numerical eigenvalue H comR , and call it the group eigenvalue; and [H combin , H comR ] or [H C1 ,H C2 ,~,H CN ]Stored in DATA x,n The characteristic values ​​of data sequence data are stored in a combined form, which can be an array, a serial of linked lists or other forms.

[0013] 4) The dimension information R contained in each data block x,n R is identified by the dimensions of different dimensions t (t=1~Z) packaged, see Figure 6 . R1~R ZThe data of different dimensions are marked according to the specific business scenarios, or the data characteristic values ​​after the characteristic value transformation, and the value of the parameter Z is set according to the application scenario, so as to facilitate the rapid comparison and verification of the data structure analysis process. Among them, different dimensions refer to the different classification rules of specific data in the data structure in the business application scenario, such as classification according to different business types, that is, dimensionalization by business type, or classification according to the different protection difficulties of different data, that is, dimensionalization by difficulty. The preceding data combination identifier can be a combination of several preceding data characteristic values, and the combination storage method can be an array, a serial linked list or other forms. The "digital characteristic value" can be a "digital summary" or other characteristic value that uniquely represents the data. If it is a "digital summary", it can be given by a hash calculation method.

[0014] When destructuring after obtaining partial data, it can be expanded based on the information contained in the existing data.

[0015] 1) For the acquired data block [DATA x,n ,F x ,T x,n ,H ComR ,H x,(n-1) ,H x,Co , R x,n ] to analyze and obtain the digital characteristic value H of the preceding data block contained in the data x,(n-1) and group identification digital eigenvalue H ComR , and obtain the corresponding previous data blocks in turn according to such values, and combine all the obtained data blocks with the corresponding previous data to identify H x,Co Perform a comparison and verification, and then according to F x The same value is classified as the same sequence data block, according to H ComR The same values ​​are grouped into the same group of data blocks;

[0016] 2) After obtaining the group identification data characteristic value H through step 1) ComR , and obtain H combin , and this query gives H C1 ~H CN sequence, and then according to H Cy (y≠x, y=1~N) data sequence to obtain the data C equ , thus obtaining B y H of the data block in the data sequence y,n ; At the same time, get all the internal H ComR The data block with the value is used to obtain the data with H in other group data sequences. ComR Data blocks of values;

[0017] 3) Based on step 2), the data block contains H y,(n-1) , and then use this value Hy,(n-1) Get the previous data block of the corresponding sequence data, and according to H Cy Classify into the sequence defined in step 1);

[0018] 4) Obtain all data blocks according to the above steps, and then comR ,F x ,H x,n ) According to the first data group H comR -Re-data sequence F x - Post data block H x,n The data are organized in order so that a data user can restore the connection of some data blocks of the data structure or the connection of all data blocks of the entire data structure.

[0019] 5) During the above steps 1), 2) and 3), distributed data users can query and compare the R stored in different data blocks at any time. t (t=1~Z) are the same to ensure the consistency of any reconstructed data in the data structure destruction process.

[0020] 6) If the business scenario requires specific dimensional data deconstruction, compare R according to business needs t , the data based on Rt of a specific dimension is restored in step 4).

[0021] The technical advantages of the present invention are:

[0022] In order to solve the problem of partial reconstruction of multidimensional data structures after distributed storage in computer systems, the encapsulation time and feature identifiers are used to perform multi-level locking in the process of building the multidimensional data structure, and then the encapsulation time and various feature identifiers are used to perform repeated iterative verification and aggregation to deconstruct the entire data structure. It can achieve:

[0023] Use the same method to construct and deconstruct unstructured data with different business logics;

[0024] Data construction can be implemented in a decentralized manner, and consistency can be guaranteed without a central control node.

[0025] Data can be reconstructed according to specific dimensions, greatly reducing the reconstruction of irrelevant data;

[0026] Ability to cluster and store data of specific dimensions, thus improving the space / time efficiency of distributed storage;

[0027] Destruction information can be limited to a specific dimension, providing a more efficient way to insert / de-register data later. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1The data structure I of the present invention is illustrated;

[0029] Figure 2 The data structure II of the present invention is illustrated;

[0030] Figure 3 The data structure III of the present invention is illustrated;

[0031] Figure 4 The data structure IV of the present invention is illustrated;

[0032] Figure 5 The data structure V of the present invention is illustrated;

[0033] Figure 6 The data structure VI of the present invention is illustrated;

[0034] Figure 7 This is the data recovery process of the present invention. DETAILED DESCRIPTION

[0035] In a given computer data system, fixed data or growing data is divided into a two-dimensional data structure consisting of multiple interrelated sequences. The topological nodes of different data in the two-dimensional data structure (F x , P x,n , H x,n ) is identified by the sequence where the data is located F x 、The position mark P in the sequence where the data is located x,n And the sequence group identifier H where the data is located xn Common instructions.

[0036] Different data sequences of the data structure are parallel, and each data sequence has a fixed length; each secondary sub-data system generates its own data sequence B as the business develops. x , (x=1~N), the time series data of each data sequence are anchored in sequence using the data structure I of the digital feature value of the data to generate data blocks, and different data sequences generate the data structure III given by the present invention respectively. This data structure III transmits the feature values ​​between different data sequences for overall anchoring, and the data structure II is used to provide more information about the preceding data of the data. Assuming that the data sequences obtain related data IV from each other under the trigger condition every 10 minutes, the data sequences process the received data IV into the data structure V and save it. The data sequences generate the data structure III given by the present invention by interactively saving the feature values, such as Figure 3 The data structure III is stored in the computer system, and the data structure is deconstructed by using various characteristic values ​​and data of the data while obtaining part of the data.

[0037] Two cases are given for the construction process of distributed data.

[0038] Case 1 The specific steps include:

[0039] 1) The data sequences B1, B2, B3, B4 in the data structure are N Generate a field based on N business scenarios, use MD5 encryption method to encapsulate data blocks, and use hash values ​​to anchor data blocks in sequence (e.g., B 1,(i+K+1) ={DATA,f H (B 1,(i+K) )}, where: Data is B 1,(i+K+1) Data, f H is the MD5 hash calculation function), x = 1 to N, data sequence B1, ~, B N Finally, they are encapsulated into data block dependency structures with different data difficulty, such as Figure 3 As shown. Among them, the data block C that constitutes the data sequence Block By DATA x,n 、F x 、T x,n , H ComR , H x,(n-1) , H x,Co , R x,n It consists of 7 parts, among which DATA x,n Refers to the nth business-related data in the xth column, F x is the identifier of the data sequence x column, T x,n It is business data DATA x,n Packaging time, H ComR is the eigenvalue of the data group, H x,(n-1) is the hash value of the preceding data (i.e., the n-1th) of the nth data in the xth column, H x,Co is the hash value identifier of the preceding data combination of the nth data block of the x sequence, R x,n The dimension information contained in the nth data in the xth column, such as Figure 1 As shown. Among them, containing the same H ComR The data of the value is a data group, H x,Co From 10 pre-order data, H x,m+0 ,H x,m+1 ,…,H x,m+10 The corresponding hash transformation operation is performed, such as Figure 2 As shown, specifically, the condition of the preceding data is triggered by every 10 data blocks, where the data H x,m+0 ,H x,m+1 ,…,H x,m+10 Combined into a hash value combination H of several previous data comseq , and H comseq Then perform the Merkle tree transformation to generate the root H x,Co , which is also called the identifier of the preceding data combination.

[0040] 2) After the trigger condition is reached every 10 minutes, the above B1~B N F of the data block of the data sequence x ,P x,n ,T x,n ,H x,n Combined into [F x ,P x,n ,T x,n ,H x,n ], that is, C equ , prepare to store the sequence B in the data sequence X In the data block, such as Figure 4 Among them, C equ Includes the identifier F of the nth data sequence x in the xth column x , the position P of the data block in the data sequence x,n , data encapsulation time T x,n And the hash value H of the data x,n .

[0041] 3) Among them, data sequence B X Get other data sequence identifier combination data C equ After that, the corresponding identification combination data C is calculated equ The hash value H Cx ; The obtained identification combination-feature value pair [C equ ,H Cx ], write the sequence data; obtain the identification combination feature value H C1 ,H C2 ,~,H CN Composition of eigenvalues ​​[H C1 ,H C2 ,~,H CN ], that is, H combin ,like Figure 5 , and calculate the hash value H of the digital feature value combination comR as the group eigenvalue; and [[H C1 ,H C2 ,~,H CN ], H comR ]Stored in DATA x,n The characteristic values ​​of the data sequence data are stored in a combined form, specifically in an array format.

[0042] 4) Include the dimension information R x,n Identified by dimension R t For packaging, such as Figure 6 As shown in the figure, each business has one dimension, that is, N dimensions, and different businesses in the business category dimension corresponding to the business scenario generate corresponding R1~R N , different Rt The hash value of the business type data Type is used to make the data block comparison and verification faster during the data decomposition process. N Data is stored in array form for quick reading.

[0043] Case 2 The specific steps include:

[0044] 1) The data sequences B1, B2, B3, B4 in the data structure are N Generate a field based on N business scenarios, use MD5 encryption method to encapsulate data, and use hash values ​​to anchor data in sequence (e.g., B 1,(i+K+1) ={DATA,f H (B 1,(i+K) )}, where: Data is B 1,(i+K+1) Data, f H is the MD5 hash calculation function), x = 1 to N, data sequence B1, ~, B N Finally, they are encapsulated into data dependency structures with different data difficulties, such as Figure 3 As shown. Among them, the data block C that constitutes the data sequence Block By DATA x,n 、F x 、T x,n , H ComR , H x,(n-1) , H x,Co , R x,n It consists of 7 parts, among which DATA x,n Refers to the nth business-related data in the xth column, F x is the identifier of the data sequence x column, T x,n It is business data DATA x,n Packaging time, H ComR is the eigenvalue of the data group, H x,(n-1) is the hash value of the nth preceding data block in the xth column, H x,Co is the hash value identifier of the preceding data combination of the nth data block of the x sequence, R x,n The dimension information contained in the nth data in the xth column, such as Figure 1 As shown. Among them, H x,Co It consists of 10 preceding data blocks, H x,m+0 ,H x,m+1 ,…,H x,m+10 The corresponding hash operation is performed, such as Figure 2 As shown, specifically, the condition of the preceding data is triggered by every 10 data blocks, where the data H x,m+0 ,H x,m+1 ,…,H x,m+10 Combined into a hash value combination H of several previous data comseq , and Hcomseq Then perform the Merkle tree transformation to generate the root H x,Co , which is also called the identifier of the preceding data combination.

[0045] 2) After the trigger condition is reached every 10 minutes, the above B1~B N The hash value H of the data block of the data sequence x,n As C equ , sequence B stored in the data sequence X In the data block, such as Figure 4 .

[0046] 3) Among them, data sequence B X Get other data sequence identifier combination data C equ After that, the corresponding identification combination data C is calculated equ The hash value H Cx ; The obtained identification combination-feature value pair [C equ ,H Cx ]Write the sequence data; get the identification combination feature value H C1 ,H C2 ,~,H CN Composition of eigenvalues ​​[H C1 ,H C2 ,~,H CN ], that is, H combin ,like Figure 5 , and calculate the digital eigenvalue H of the digital eigenvalue combination comR , and call it the group eigenvalue; and [[H C1 ,H C2 ,~,H CN ], H comR ] is stored in the data. The characteristic values ​​of the data sequence data are stored in a combined form, specifically in a linked list serial form.

[0047] 4) Include the dimension information R x,n By dimension R t For packaging, such as Figure 7 As shown. The identification of data difficulty dimension R t It is generated by the hash calculation of the business setting difficulty data and is divided into q dimensions. Among them, the combination of the previous data and R1~R q Data is stored in linked list format for fast storage.

[0048] When partial data is obtained and then deconstructed, it can be expanded based on the information contained in the existing data.

[0049] Two cases are also given for the data deconstruction process.

[0050] Case 1 requires the entire data structure to be restored. The specific steps include: 1) recovering the acquired data [DATA x,n ,F x ,T x,n ,H ComR ,H x,(n-1) ,H x,Co , R x,n ] and dimension symbol R t Analyze and obtain the characteristic value H of the previous data contained in the data x,(n-1) and group identification data eigenvalue H ComR , and obtain the corresponding previous data in turn according to this value, and combine all the obtained data with several previous data to mark H x,Co Perform a comparison and verification, and then according to F x The same value is classified as the same sequence data block, according to H ComR The same values ​​are grouped into the same group of data blocks;

[0051] 2) After obtaining the group identification data characteristic value H through step 1) ComR , and obtain H combin Combination, and this query gives H C1 ~H CN sequence, and then according to H Cy (y≠x, y=1~N) data sequence to obtain the group B y Data sequence data H y,(n) ; At the same time, all the included H ComR Value, and obtain data in other group data sequences based on it;

[0052] 3) Based on step 2), the data contains H y,(n) , and then obtain the previous data of each sequence according to this value, and according to different H Cy Classify into the sequence class defined in step 1);

[0053] 4) Obtain all data according to the above steps, and then use the (H ComR ,F x ,H x,(n) ) According to the first data group H comR -Re-data sequence F x - Post data block H x,n The data are organized in order so that data users can restore the connections of all data blocks of the entire data structure.

[0054] 5) During the above steps 1), 2) and 3), distributed data users can query and compare the R stored in different data blocks at any time. t(t=1~Z), ensuring the consistency of any data obtained for reconstruction during the entire destruction process.

[0055] Case 2 requires the recovery of data of a specific dimension, such as Figure 7 As shown, the specific steps include:

[0056] 1) For the acquired data [DATA x,n ,F x ,T x,n ,H ComR ,H x,(n-1) ,H x,Co o , R x,n ] and dimension symbol R t Analyze and obtain the characteristic value H of the previous data contained in the data x,(n-1) and group identification data eigenvalue H ComR , a specific dimension R given according to business needs Zero , and obtain the corresponding previous data in turn, and combine all the obtained data with several previous data to mark H x,Co Perform a comparison and verification, and then according to the specific dimension R Zero Classification;

[0057] 2) After obtaining the group identification data characteristic value H through step 1) ComR , and obtain H combin Combination, and this query gives H C1 ~H CN sequence, and then according to H Cy (y≠x, y=1~N) data sequence to obtain the data H of the group data By sequence y,(n) ; At the same time, all the included H ComR Value, and obtain data in other group data sequences based on it;

[0058] 3) Based on step 2), the data contains H y,(n) , and then obtain the previous data of each sequence according to this value, and according to different H Cy Classify into the sequence class defined in step 1);

[0059] 4) Obtain all data according to the above steps, and compare the dimension identifier R contained in each data item by item t and business requirements dimension identifier R Zero Obtain the matching data, and then use the (H ComR ,F x ,H x,(n) ,R t ) According to the first data group H comR -Re-data sequence F x- Post data block H x,n The data is organized in order to restore the connections of all data blocks of data of a specific dimension.

[0060] 5) During the above steps 1), 2) and 3), distributed data users can query and compare the R stored in different data blocks at any time. t (t=1~Z), ensuring the consistency of any data obtained for reconstruction during the entire destruction process.

[0061] The above description is only a preferred embodiment of the present invention, which is only used to illustrate the method of the present invention, and is not used to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for constructing a multidimensional topological structure of distributed data, the steps comprising: 1) Select the corresponding encapsulation algorithm rules for each data sequence in the data structure, and use the digital feature values ​​to anchor the data blocks in the same data sequence in sequence; the data block C that constitutes the data sequence Block Include DATA x,n 、F x , H ComR and H x,(n-1) , DATA x,n is the nth business-related data block in the xth column of the data sequence, F x is the identifier of the data sequence, H ComR is the eigenvalue of the data group, H x,(n-1) It is the digital characteristic value of the combination of the preceding data blocks of the nth data block in the xth column of the data sequence; ComR The data of the value is a data group; 2) When the set time interval is reached, each data sequence B1 to B N Perform steps a) to c) respectively: a) All N data sequences B1 to B in the data structure N The data block H x,n Combined into C equ , X = 1 to N; Among them, H x,n is the digital characteristic value of the data block; b) Current data sequence B X Calculate the C equ The numerical eigenvalue H Cx , and send it to other data sequences or a set data sequence; then [C equ ,H Cx ]、C equ or H Cx Write current data sequence B X In a newly generated data block; c) Data sequence B i Get the digital characteristic value H of each data sequence C1 ,H C2 ,~,H CN Then form the eigenvalue combination H combin , and calculate the eigenvalue combination H combin The numerical eigenvalue H comR , as the group eigenvalue; Then [H combin , H comR ] or H combin The data sequence B is stored in i DATA x,n In the data; d) Dimension identifiers R for different dimensions of data blocks t Encapsulate to obtain the dimension information R contained in the data block x,n ; Among them, t = 1 ~ Z, R1 ~ R Z According to the data sequence B X Dimension identifier for corresponding business scenario design.

2. The method according to claim 1, characterized in that The data structure is a two-dimensional data structure used to store fixed data or growing data.

3. The method according to claim 1, characterized in that The data block C Block Also includes T x,n , H x,Co and R x,n ; Among them, T x,n It is the business data block DATA x,n Packaging time, H x,Co is the identification digital feature of the preceding data combination of the nth data block in the xth column of the data sequence, R x,n The dimension information contained in the nth data block in the xth column.

4. The method according to claim 1, characterized in that The C equ Also includes F x ,P x,n ,T x,n , X = 1 to N; Among them, P x,n is the position of the data block in the data sequence, T x,n is the data encapsulation time, H x,n is the digital feature value of the data block.

5. The method according to claim 3, characterized in that H x,Co From X+1 preceding data H x,m+0 ,H x,m+1 ,…,H x,m+X The corresponding transformation operation is performed to obtain the value of X, which is triggered by the set conditions; wherein the transformation operation method is: transform the data H x,m+0 ,H x,m+1 ,…,H x,m+X Combined into data eigenvalue combination H comseq , H comseq Then transform the eigenvalue to generate H x,Co .

6. The method according to claim 1, characterized in that In step 1), each data sequence in the data structure is encapsulated according to different business scenarios or setting criteria, and the corresponding encapsulation algorithm rules are selected. N They are encapsulated into their own data dependency structures according to their own business characteristics.

7. A method for deconstructing the multi-dimensional topological structure of distributed data constructed by the method of claim 1, the steps comprising: 11) [DATA x,n ,F x ,H ComR ,H x,(n-1) ] to analyze the data according to the digital characteristic value H of the previous data block contained in the data x,(n-1) and group identification digital eigenvalue H ComR , obtain the corresponding preceding data blocks, and combine the obtained data blocks with the corresponding preceding data to identify H x,Co Perform comparison and verification, and then classify the data blocks. x Data blocks with the same value are classified into the same data sequence, and H ComR The same data blocks are grouped into the same data group; 12) The group identification data characteristic value H obtained according to step 11) ComR Get H combin , and the sequence H is obtained by querying C1 ~H CN ; Then according to H Cy Sequence of data to obtain data C equ , thus obtaining the data sequence B y H of the data block in y,n , where y≠x, y=1~N; at the same time, obtain all the internal H ComR The data block with the value is used to obtain the data with H in other group data sequences. ComR Data blocks of values; 13) Based on step 12), obtain the H of the data block y,(n-1) , then according to H y,(n-1) Get the preceding data block of the corresponding data sequence, and calculate the corresponding data block according to the digital characteristic value H of the data sequence. Cy Categorize data blocks; 14) According to the (H comR ,F x ,H x,n ) According to the first data group H comR -Re-data sequence F x - Post data block H x,n The data blocks are organized in order and the connections of some or all data blocks in the data structure are restored.

8. The method according to claim 7, characterized in that The data block also includes H x,(n-1) ,H x,Co , R x,n .

9. The method according to claim 8, characterized in that According to the dimension identification R of the obtained data block t Compare with the dimension identifier of the reconstructed data to determine the consistency of the reconstructed data.

Images