A big data information hierarchical management method and system based on fusion intelligent analysis

Abstract

The application discloses a kind of big data information hierarchical management method and system based on fusion intelligent analysis, it is related to big data analysis technical field, including the following steps: S1, constructs multi-source information uniform expression base;S2, using uniform information representation matrix is coupled to build across source semantics-behavior;S3, using information correlation topological structure carries out multi-level fusion deduction analysis;S4, using fusion risk representation vector carries out grade mapping matrix calculation;S5, using information hierarchical result and information correlation topological structure carries out structure back-embedded mapping.The application sets semantics-behavior coupling mapping calculation mechanism, semantic similarity matrix and behavior correlation intensity matrix are linearly coupled according to fixed coupling weight in relation construction stage, to couple result directly as the information correlation topological structure of edge generation rule generation, realize the synchronous fusion modeling of semantic attribute and behavior attribute under the condition of uniform expression base, improve the consistency and stability of correlation structure.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, specifically to a hierarchical management method and system for big data information based on fusion intelligent analytics. Background Technology

[0002] With the development of big data technology, massive amounts of information are constantly converging in government platforms, financial systems, internet business platforms, and industry data centers. The scale of data is growing exponentially, and the sources of information are becoming increasingly multi-source, spanning departments, systems, and business domains. Against this backdrop, how to uniformly represent, correlate, and securely manage large-scale heterogeneous information has become a crucial issue that urgently needs to be addressed in the fields of information security and data governance.

[0003] In existing technologies, there are shortcomings in the mechanism for constructing a synchronous coupling structure of cross-source attributes under a unified expression basis. This results in the inability to achieve synchronous fusion between semantic association and behavioral association during the relationship construction stage, making it difficult to form a single and unified association topology. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a hierarchical management method and system for big data information based on fusion intelligent analysis, in order to solve the problems mentioned in the background.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] In a first aspect, embodiments of the present invention provide a method and system for hierarchical management of big data information based on fusion intelligent analysis, comprising the following steps:

[0007] S1. Construct a unified representation basis for multi-source information to obtain a unified information representation matrix;

[0008] S2. Use a unified information representation matrix to construct cross-source semantic-behavioral coupling and obtain the information association topology.

[0009] S3. Use the information association topology to perform multi-level fusion inference and analysis to obtain the fusion risk representation vector;

[0010] S4. Use the fused risk representation vector to calculate the level mapping matrix and obtain the information classification results;

[0011] S5. Use the information hierarchical results and information association topology to perform structural back-embedding mapping to obtain the hierarchical management structure model.

[0012] To further optimize this technical solution, step S1 is used to convert multi-source heterogeneous information into a unified structural expression, eliminating field conflicts and inconsistencies in units caused by differences in data sources, and constructing a standardized input matrix that can be directly used for association topology construction and fusion analysis. This matrix is ​​a unified information representation matrix. Its expression is:

[0013] ;

[0014] in,

[0015] Indicates the total number of information objects;

[0016] Indicates the first A unified representation vector for each information object;

[0017] Each It consists of three parts assembled in sequence:

[0018]

[0019] in:

[0020] Represents the structural identifier vector. ;

[0021] Represents a semantic attribute vector. ;

[0022] Represents a behavioral attribute vector. .

[0023] To further optimize this technical solution, step S2 involves constructing a unified expression basis in step S1. Based on this, the semantic and behavioral attributes between information objects are uniformly coupled and calculated to form a structured association network. ;

[0024] Step S2, in constructing cross-source semantic-behavior coupling, includes the following steps:

[0025] Semantic vector set construction and dimension confirmation;

[0026] Semantic similarity matrix construction;

[0027] Construction of behavioral vector sets and time dimension correction;

[0028] Construction of behavioral correlation strength matrix;

[0029] Semantic-behavior coupling mapping computation;

[0030] Candidate neighborhood selection and threshold-based edge construction;

[0031] Topology construction.

[0032] To further optimize this technical solution, in step S2, during the construction of the semantic vector set and dimension confirmation, from... Extract:

[0033]

[0034] in:

[0035] The fixed dimension is the semantic field concatenation result from step S1;

[0036] like If it is a zero vector, then add a minimal positive number to all its dimensions. To avoid the denominator being zero;

[0037] Step S2 involves setting the semantic similarity function during the construction of the semantic similarity matrix:

[0038] ;

[0039] in:

[0040] and Both are information object serial numbers;

[0041] "·" represents the dot product of vectors;

[0042] It is a 2-norm;

[0043] ;

[0044] Thus, the semantic similarity matrix is ​​obtained:

[0045] .

[0046] To further optimize this technical solution, in step S2, during the construction of the behavior vector set and the time dimension correction, from... Extract:

[0047] ;

[0048] in:

[0049] For a fixed number of time windows;

[0050] The time window length has been uniformly set in step S1;

[0051] If the statistical period for a certain object is insufficient, it will be padded with zero values ​​to reach the required period. dimension;

[0052] Step S2 uses the Pearson correlation coefficient when constructing the behavioral correlation strength matrix:

[0053] ;

[0054] in:

[0055] for Standard deviation.

[0056] like or If it is zero, then set: ;

[0057] To unify the dimensions, it is linearly mapped to :

[0058] ;

[0059] Obtain the behavioral correlation matrix:

[0060] .

[0061] To further optimize this technical solution, in step S2, a fixed coupling weight is set during the semantic-behavior coupling mapping calculation:

[0062] ;

[0063] ;

[0064] ;

[0065] , Set as a constant during the system initialization phase;

[0066] Construct a coupling function based on the set coupling weights:

[0067] ;

[0068] The coupling matrix is ​​obtained as follows:

[0069] .

[0070] Candidate neighborhood selection and threshold-based edge construction:

[0071] Partitioning is performed based on business domain identifiers, and calculations are only performed within the same domain or within a specified cross-domain scope. ;

[0072] Set a fixed threshold , indicating the coupling strength Lower limit control value:

[0073] ;

[0074] like and Then, weighted edges are constructed, with edge weights being... .

[0075] To further optimize this technical solution, in step S2, during topology construction, a hash mapping function is set to generate a vertex set representing the graph. This is used to represent all information objects involved in the calculation of coupling relationships:

[0076] ;

[0077] And set the edge set:

[0078] ;

[0079] A graph structure is defined as:

[0080] Undirected weighted graph;

[0081] It does not contain self-loops;

[0082] Repeated edges are not allowed.

[0083] The final output of step S2 is the information association topology. Its expression is:

[0084] ;

[0085] in:

[0086] ;

[0087] ;

[0088] .

[0089] To further optimize this technical solution, step S3 involves constructing a unified coupled topology in step S2. Based on this, a hierarchical propagation calculation is performed on the coupling relationships between nodes, and the direct coupling effects and indirect path propagation effects are unified and integrated to form a structural fusion risk quantification result for each information object. .

[0090] To further optimize this technical solution, step S4 maps continuous fusion risk values ​​to discrete level labels, constructs structured information grading results, and transforms risk characterization results into deterministic grading outputs.

[0091] Step S5, while keeping the original graph structure and topology unchanged, embeds the node hierarchical results back into the graph structure to form a unified structural model with hierarchical attributes, so that the information association structure and hierarchical results are expressed in the same model.

[0092] A big data information hierarchical management system based on fusion intelligent analysis is constructed based on the aforementioned big data information hierarchical management method based on fusion intelligent analysis, and includes the following modules:

[0093] Information access and standardized processing module;

[0094] Information association modeling module;

[0095] Integrate risk calculation module;

[0096] Level mapping analysis module;

[0097] Structural embedding modeling module;

[0098] Hierarchical management and control module.

[0099] In a second aspect, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program instructions, when executed by the processor, implement the steps of a big data information hierarchical management method and system based on fusion intelligent analysis as described in the first aspect of the present invention.

[0100] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, they implement the steps of a big data information hierarchical management method and system based on fusion intelligent analysis as described in the first aspect of the present invention.

[0101] Compared with existing technologies, this invention provides a method and system for hierarchical management of big data information based on fusion intelligent analysis, which has the following beneficial effects:

[0102] This big data information hierarchical management method and system based on fusion intelligent analysis sets up a semantic-behavior coupling mapping mechanism, which enables the semantic similarity matrix and the behavioral association strength matrix to be linearly coupled according to fixed coupling weights during the relationship construction stage. The coupling result is directly used as the edge construction rule to generate the information association topology, thereby realizing the synchronous fusion modeling of semantic attributes and behavioral attributes under a unified expression basis, avoiding cross-model mapping process, and improving the consistency and stability of the association structure. Attached Figure Description

[0103] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0104] Figure 1 This is a flowchart illustrating a hierarchical management method for big data information based on fusion intelligent analysis proposed in this invention.

[0105] Figure 2 This is a schematic diagram of the cross-source semantic-behavior coupling construction process of a big data information hierarchical management method based on fusion intelligent analysis proposed in this invention;

[0106] Figure 3 This is a schematic diagram of the multi-level fusion inference and analysis process of a big data information hierarchical management method based on fusion intelligent analysis proposed in this invention;

[0107] Figure 4 This is a schematic diagram of the structure back-embedding mapping process for a big data information hierarchical management method based on fusion intelligent analysis proposed in this invention. Detailed Implementation

[0108] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0109] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0110] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0111] Example 1:

[0112] Reference Figures 1-4 This is the first embodiment of the present invention, which provides a hierarchical management method for big data information based on fusion intelligent analysis, including the following steps:

[0113] S1. Construct a unified representation basis for multi-source information to obtain a unified information representation matrix;

[0114] Step S1 is used to convert multi-source heterogeneous information into a unified structural expression, eliminate field conflicts and inconsistencies in units caused by differences in data sources, and construct a standardized input matrix that can be directly used for association topology construction and fusion analysis.

[0115] Step S1, when constructing the same expression, includes the following steps:

[0116] Field mapping and structural unification:

[0117] Unify data fields from different systems.

[0118] Create a global field definition table Predefine standard field names, field types, and field lengths;

[0119] Using mature field matching technology, similarity matching is performed on field names from different sources;

[0120] Establish a field mapping table to map source fields to... Standard fields in;

[0121] For data with conflicting fields, the primary field is retained according to the preset priority rules;

[0122] Missing fields are filled with fixed fill values, which are predetermined in the field definition table.

[0123] This leads to a unified set of field structures. .

[0124] Semantic attribute field construction:

[0125] Express information content numerically.

[0126] A global vocabulary was constructed based on all sample data;

[0127] By using the mature TF-IDF model, text fields are converted into fixed-dimensional semantic vectors;

[0128] The vocabulary dimension is fixed during the construction phase and remains consistent across all information objects;

[0129] The category field is converted into a numerical vector using a mature one-hot encoding method;

[0130] Continuous numerical fields are processed using the minimum-maximum normalization method, with the normalization interval uniformly set to 0 to 1;

[0131] Normalized extreme values ​​are calculated based on the full sample data;

[0132] Determine the set of semantic attribute vectors .

[0133] Construction of behavioral attribute fields:

[0134] Structured statistics on information behavior trajectories;

[0135] Extract timestamps, operation types, and access subjects using mature log parsing technology;

[0136] Set a fixed time window length ;

[0137] A sliding window statistical method is used to count the number of visits within each time window;

[0138] Time window length The sliding step size is fixed during system initialization;

[0139] All information objects use the same time window segmentation rules;

[0140] The behavioral statistics indicators were subjected to min-max normalization, with a range of 0 to 1.

[0141] Derive the set of behavioral attribute vectors .

[0142] Structure identifier field construction:

[0143] Generate a unique structural identifier for each information object;

[0144] The primary key field is encoded using a mature hashing technique to generate a unique node identifier.

[0145] A hierarchical coding method is used to generate business domain identifiers;

[0146] The unique node identifier is combined with the business domain identifier to form a structural identifier field;

[0147] Thus, the set of structural identifiers is derived. .

[0148] Construction of a unified representation matrix:

[0149] Combine the three sets of fields mentioned above.

[0150] Concatenate fields in a fixed order: structural identifier field → semantic attribute field → behavioral attribute field;

[0151] All information objects use the same field order;

[0152] Perform a field quantity consistency check before splicing;

[0153] If any fields are missing, they will be filled with preset values.

[0154] All fields are numerical.

[0155] The final output of step S1 is the unified information representation matrix. Its expression is:

[0156] ;

[0157] in,

[0158] Indicates the total number of information objects;

[0159] Indicates the first A unified representation vector for each information object;

[0160] Each It consists of three parts assembled in sequence:

[0161]

[0162] in:

[0163] Represents the structural identifier vector. ;

[0164] Represents a semantic attribute vector. ;

[0165] Represents a behavioral attribute vector. .

[0166] For one OK, A two-dimensional matrix of columns, where:

[0167] Number of lines The number of information objects;

[0168] Number of columns The total dimensions of the fixed field;

[0169] All fields are numeric;

[0170] The units of each field are unified to the range of 0 to 1 or a fixed encoding space;

[0171] The dimensions remain unchanged after they are determined during the system initialization phase.

[0172] S2. Use a unified information representation matrix to construct cross-source semantic-behavioral coupling and obtain the information association topology.

[0173] Step S2: The unified expression basis constructed in step S1 Based on this, the semantic and behavioral attributes between information objects are uniformly coupled and calculated to form a structured association network. .

[0174] Step S2, in constructing cross-source semantic-behavior coupling, includes the following steps:

[0175] Semantic vector set construction and dimension confirmation:

[0176] from Extract:

[0177]

[0178] in:

[0179] The fixed dimension is the semantic field concatenation result from step S1;

[0180] all Consistent dimensions;

[0181] like If it is a zero vector, then add a minimal positive number to all its dimensions. To avoid the denominator being zero.

[0182] Semantic similarity matrix construction:

[0183] Set the semantic similarity function:

[0184] ;

[0185] in:

[0186] "·" represents the dot product of vectors;

[0187] It is a norm 2;

[0188] ;

[0189] Thus, the semantic similarity matrix is ​​obtained:

[0190] ;

[0191] The matrix is ​​a symmetric matrix.

[0192] Construction of behavioral vector sets and time dimension correction:

[0193] from Extract:

[0194] ;

[0195] in:

[0196] For a fixed number of time windows;

[0197] The time window length has been uniformly set in step S1;

[0198] If the statistical period for a certain object is insufficient, it will be padded with zero values ​​to reach the required period. dimension.

[0199] Construction of behavioral association strength matrix:

[0200] Using Pearson correlation coefficient:

[0201] ;

[0202] in:

[0203] for Standard deviation.

[0204] like or If it is zero, then set: ;

[0205] To unify the dimensions, it is linearly mapped to :

[0206] ;

[0207] Obtain the behavioral correlation matrix:

[0208] .

[0209] Semantic-behavior coupling mapping computation:

[0210] Set fixed coupling weights:

[0211] ;

[0212] ;

[0213] ;

[0214] , Set as a constant during the system initialization phase;

[0215] Construct a coupling function based on the set coupling weights:

[0216] ;

[0217] The coupling matrix is ​​obtained as follows:

[0218] .

[0219] Candidate neighborhood selection and threshold-based edge construction:

[0220] Partitioning is performed based on business domain identifiers, and calculations are only performed within the same domain or within a specified cross-domain scope. ;

[0221] Set a fixed threshold , indicating the coupling strength Lower limit control value:

[0222] ;

[0223] Set during the system initialization phase;

[0224] like and Then, weighted edges are constructed, with edge weights being... .

[0225] Topology construction:

[0226] Set a hash mapping function to generate a set of vertices representing the graph. This is used to represent all information objects involved in the calculation of coupling relationships:

[0227] ;

[0228] And set the edge set:

[0229]

[0230] A graph structure is defined as:

[0231] Undirected weighted graph;

[0232] It does not contain self-loops;

[0233] Repeated edges are not allowed.

[0234] The final output of step S2 is the information association topology. Its expression is:

[0235] ;

[0236] in:

[0237] ;

[0238] ;

[0239] .

[0240] Compared to existing technologies that typically construct semantic similarity networks and behavioral association networks separately and then overlay the results in the post-processing stage, this step unifies the representation basis. Semantic and behavioral vectors are directly extracted and linearly coupled within the same node space. The coupling result directly determines the generation rules of weighted edges, thereby constructing a unified information association topology in one go. The node identifiers, attribute representations, and relation weights of this structure all originate from the same basic data system, avoiding cross-model result mapping processes. Therefore, the difference lies not in the single similarity algorithm itself, but in the construction mechanism that completes cross-source attribute synchronous coupling and directly forms the topology structure under the unified expression condition.

[0241] S3. Use the information association topology to perform multi-level fusion inference and analysis to obtain the fusion risk representation vector;

[0242] Step S3: The unified coupled topology constructed in step S2 Based on this, a hierarchical propagation calculation is performed on the coupling relationships between nodes, and the direct coupling effects and indirect path propagation effects are unified and integrated to form a structural fusion risk quantification result for each information object. .

[0243] Step S3, in the multi-level fusion simulation analysis, includes the following steps:

[0244] Node number mapping and adjacency matrix construction:

[0245] To ensure consistency in matrix calculations, according to The original sequential numbering of objects in the middle, establish a number mapping function:

[0246] ;

[0247] Construct a weighted adjacency matrix:

[0248] ;

[0249] Two nodes in the matrix The edge weights of the formed edges are set as follows: ;

[0250] If no edge exists, then ;

[0251] in:

[0252] ;

[0253] Derived from the coupling strength in step S2, It is a symmetric matrix and does not contain self-loops.

[0254] Primary structural impact calculation:

[0255] Set up first-level adjacency set calculation:

[0256] ;

[0257] Set up the calculation of the first-level direct impact quantity:

[0258] ;

[0259] If a node has no adjacent edges, then: ;

[0260] Represents a node The direct coupling affects the total amount.

[0261] Calculation of the impact of secondary path propagation:

[0262] Set up a second-level adjacency set:

[0263] ;

[0264] Set the propagation strength of the secondary path:

[0265] ;

[0266] The calculation is performed only based on existing edges, without recalculation. ;

[0267] If node If there are no first-level adjacent nodes, then: ;

[0268] This calculation reflects the propagation effect with a path length of 2, a fixed layer depth of 2, and no higher-order propagation.

[0269] Hierarchical fusion computing:

[0270] Set a fixed attenuation coefficient: ;

[0271] Set as a constant during the system initialization phase;

[0272] Set the influence value of the merged structure:

[0273] ;

[0274] Represents a node The multi-level structure fusion value.

[0275] Normalization and vector construction:

[0276] set up This represents the maximum value of the "fusion structure impact value" among all current information objects.

[0277] ;

[0278] like ,but:

[0279] ;

[0280] like ,but:

[0281] ;

[0282] Finally, a fusion risk representation vector is constructed:

[0283] ,

[0284] ;

[0285] Vector order and numbering mapping Consistent.

[0286] Compared to existing technologies that typically rank nodes based solely on node weighting, single-propagation steady-state values, or centrality indices, this approach, on a unified topology formed by semantic-behavioral coupling, clearly distinguishes between direct hierarchical influences and secondary path propagation influences. It achieves hierarchical progressive fusion through path weight product calculation and introduces fixed depth and attenuation coefficients for structured control, allowing risk characterization results to be directly derived from coupled topological relationships rather than relying on additional external attribute models. Its core difference lies not in a single graph calculation formula, but in constructing a hierarchical propagation fusion calculation framework based on a coupled graph structure, ultimately forming a unified fusion risk characterization vector.

[0287] S4. Use the fused risk representation vector to calculate the level mapping matrix and obtain the information classification results;

[0288] Step S4 maps continuous fusion risk values ​​to discrete level labels, constructs structured information grading results, transforms risk characterization results into deterministic grading outputs, and provides clear grading basis for subsequent grading management implementation.

[0289] Step S4, when calculating the level mapping matrix, includes the following steps:

[0290] Number of levels and setting of interval boundaries:

[0291] Set a fixed number of levels:

[0292] , ;

[0293] It is set to a fixed positive integer during the system initialization phase and does not change with the data size.

[0294] Using a mature equidistant binning method, closed intervals are... Divide evenly:

[0295] The interval width is:

[0296]

[0297] Build A continuous interval:

[0298] Interval 1:

[0299] Second interval:

[0300] …

[0301] No. Interval: .

[0302] The interval rule is:

[0303] The first interval contains 0;

[0304] The last interval contains 1;

[0305] Each interval is non-overlapping and continuously covers [0,1].

[0306] Form a level mapping matrix Its essence is:

[0307] A table showing the one-to-one correspondence between each risk value range and its unique level number;

[0308] The grade number is set as follows:

[0309] ;

[0310] The numerical value of the number increases as the risk value range increases.

[0311] Risk value range matching calculation:

[0312] right Each element in Perform range matching;

[0313] Use mature interval search algorithms to locate intervals and determine... The interval into which the element falls is assigned a corresponding level number based on the interval it belongs to.

[0314] like If so, it will be classified as Level 1;

[0315] like Then it belongs to the first grade.

[0316] Hierarchical result construction:

[0317] For all After performing interval matching on each information object, a ranking result is generated;

[0318] Construct a set of information classification results: Its expression is:

[0319] ;

[0320] in:

[0321] For the node identifier in step S2 ;

[0322] The level number obtained by interval mapping;

[0323] Sequence and Number Mapping Maintain consistency.

[0324] like If all elements are 0, then all information objects have a level of 1.

[0325] Allowing for the absence of object distribution in certain grade intervals does not affect the validity of the results.

[0326] S5. Use the information hierarchical results and information association topology to perform structural back-embedding mapping to obtain the hierarchical management structure model;

[0327] Step S5, while keeping the original graph structure and topology unchanged, embeds the node hierarchical results back into the graph structure to form a unified structural model with hierarchical attributes. This allows the information association structure and hierarchical results to be expressed in the same model, providing a unified data foundation for subsequent hierarchical management, control, or scheduling.

[0328] Step S5, during the structural back-embedding mapping, includes the following steps:

[0329] Consistency check:

[0330] right and Perform node consistency verification, and the verification rules satisfy:

[0331] The set of nodes in the middle is equal to ;

[0332] The number of nodes is ;

[0333] Each node corresponds to only one level value.

[0334] If there are missing or duplicate entries, then use The mapping is rearranged based on the baseline to ensure that the one-to-one correspondence is maintained.

[0335] Construction of ranking functions:

[0336] based on Construct a node-level function, which is a deterministic mapping function, and its expression is:

[0337] ;

[0338] ;

[0339] in:

[0340] The domain is a set of nodes. ;

[0341] The value range is a discrete set of levels;

[0342] Each node corresponds to a unique level value.

[0343] Node attribute extension:

[0344] Construct a set of nodes with hierarchical attributes to form an expanded node set. Its expression is:

[0345] ;

[0346] During the construction process:

[0347] Node identifier No change;

[0348] Add only attribute fields .

[0349] Construction of the structural back-embedding model:

[0350] While keeping the original edge set and edge weight function unchanged, construct a hierarchical management structure model:

[0351]

[0352] in:

[0353] The original set of nodes;

[0354] This is the original set of associated edges;

[0355] The original edge weight function;

[0356] This is a node-level function.

[0357] The model satisfies:

[0358] The number of nodes remains unchanged. ;

[0359] The number of edges remains unchanged;

[0360] The edge weight function remains unchanged;

[0361] Each node has an attached level attribute.

[0362] This construction process is called "structural back-embedding mapping".

[0363] Compared to existing mature hierarchical techniques, the difference in this step lies in the uniformity of the structural representation. Traditional methods typically store hierarchical results as independent label sets or data tables, separating the hierarchical results from the original association structure, making graph structure modeling and hierarchical partitioning processes independent of each other; while this method uses hierarchical functions... By directly embedding the hierarchical results into the original association graph model as node attributes, the hierarchical results become the intrinsic generating components of the structural model, constructing a weighted graph unified model with node level attributes, thereby realizing the integrated expression of hierarchical information and association structure, rather than external additional storage.

[0364] Example 2:

[0365] This embodiment provides a big data information hierarchical management system based on fusion intelligent analysis, including the following modules:

[0366] The information access and standardization processing module, corresponding to step S1, is used to complete the unified access and standardization processing of multi-source information data. By cleaning, formatting and extracting unique identifiers from structured or semi-structured data, a unified information object identification system is generated. Node identifiers are constructed through a hash mapping mechanism, thereby establishing a consistent data expression foundation and providing structured input for subsequent association modeling.

[0367] The information association modeling module, corresponding to step S2, is used to construct the association structure between information objects. By analyzing attribute similarity or pre-set association rules, it establishes the relationship between nodes, forms a set of nodes, a set of edges and corresponding edge weight functions, and constructs a weighted information association graph structure. This realizes the formal expression of the topological relationship between information and provides a structural foundation for subsequent fusion analysis.

[0368] The risk calculation module, corresponding to step S3, extracts the structural and relational features of nodes based on the constructed association graph structure, calculates the risk measure of each node, and performs comprehensive calculation and normalization through a multi-factor fusion mechanism to generate continuous risk characterization results, thereby realizing the quantitative expression of the risk level of information objects.

[0369] The grade mapping analysis module, corresponding to step S4, is used to convert continuous risk values ​​into discrete grade labels. By presetting a fixed number of grades and dividing the risk value interval equally, the risk value of each node is mapped to the corresponding grade interval, thereby realizing the discrete graded expression of risk results and providing a clear grade basis for structured management modeling.

[0370] The structural embedding modeling module, corresponding to step S5, is used to embed the hierarchical results into the original information association graph structure. While keeping the node set, edge set, and edge weight function unchanged, it constructs the node hierarchical function and expands the node attribute space, so that the hierarchical results become the endogenous attributes of the graph structure, forming a unified hierarchical management structure model, and realizing the integrated expression of hierarchical information and association structure.

[0371] Hierarchical management and control module, corresponding to The execution module implements management strategies based on a hierarchical management structure model. It achieves access control and permission division through node level information, and supports cross-level correlation analysis, level statistical analysis, level distribution display, and dynamic model update functions, thereby realizing hierarchical management and control of big data information objects.

[0372] Example 3:

[0373] This embodiment provides a practical application scenario for a hierarchical management method and system for big data information based on fusion intelligent analysis:

[0374] A city has built a unified data sharing platform to integrate data from multiple departments.

[0375] This platform adopts a big data information hierarchical management method and system based on fusion intelligent analysis to establish a unified information hierarchical management mechanism. The platform implementation process includes:

[0376] Information access and standardization stage:

[0377] Data from various departments is accessed through interfaces to the platform, and the system performs standardized format processing on data from different sources, including:

[0378] Extract unique identifier fields (such as ID number, company code, etc.);

[0379] Deduplication and consistency verification;

[0380] Construct a unified collection of information objects;

[0381] The goal of this stage is to ensure that all information objects can be uniformly identified and managed, and that data that was originally scattered across various departments is organized into a set of data nodes with a unified structure.

[0382] Information association modeling stage:

[0383] The system analyzes the relationships between different data, including:

[0384] Data records of the same citizen across multiple departments are linked;

[0385] The relationship between a company and its legal representative;

[0386] Equity relationships between companies;

[0387] The link between medical records and health insurance records;

[0388] These relationships are used to construct an information-related network structure;

[0389] The platform is able to identify complex relationships between data, rather than just isolated data entries.

[0390] Integration Risk Analysis Phase:

[0391] The system conducts a comprehensive analysis based on the following factors:

[0392] The sensitivity of the information itself (e.g., an ID number higher than a general business number);

[0393] The breadth of data connections (the more departments involved, the higher the risk);

[0394] Historical access frequency and scope of calls;

[0395] Does it involve cross-departmental sharing?

[0396] Through a fusion analysis mechanism, the system provides a risk intensity assessment value for each information object.

[0397] The system can identify:

[0398] Highly sensitive and widely interconnected data objects;

[0399] Data that is not highly sensitive but carries a high risk of cross-contagion;

[0400] Data with low sensitivity and small impact.

[0401] Level mapping phase:

[0402] The platform management department has pre-defined a five-level hierarchical system, including:

[0403] Level 1: Publicly available data;

[0404] Level 2: Internally shared data;

[0405] Level 3: Restricted access data;

[0406] Level 4: Sensitive data;

[0407] Level 5: Highly sensitive data;

[0408] Based on the results of the integrated risk analysis, the system maps each information object to the corresponding level.

[0409] Each piece of data has a clearly defined security level label in the system.

[0410] Structural embedding modeling stage:

[0411] The system embeds the hierarchical results into the original associated network structure, so that each node in the associated network has hierarchical attributes at the same time; at this time, a unified hierarchical management structure model is formed.

[0412] Platform implementation:

[0413] Data at the same level can be shared directly;

[0414] Cross-level calls require approval;

[0415] High-level data triggers an audit mechanism when accessed across departments;

[0416] Analyze cross-level risk chains.

[0417] In actual business operations, this system can achieve the following processes:

[0418] Dynamic access control:

[0419] When a department requests access to a certain piece of data:

[0420] The system automatically determines whether access is permitted or whether approval is required based on the data level.

[0421] Cross-level risk warning:

[0422] If lower-level data is indirectly connected to higher-level data through a chain of associations:

[0423] The system can automatically identify potential risk propagation paths and alert management personnel.

[0424] Data sharing for decision support:

[0425] Management departments can base their decisions on the overall distribution of levels:

[0426] Optimize sharing strategies;

[0427] Adjusting data access boundaries;

[0428] Identify high-risk data concentration areas.

[0429] Audit and traceability:

[0430] When an abnormal data call occurs:

[0431] The system can trace related paths based on a hierarchical structure model and quickly locate high-risk nodes.

[0432] Example 4:

[0433] This embodiment also provides a computer device applicable to a big data information hierarchical management method and system based on fusion intelligent analysis, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the big data information hierarchical management method and system based on fusion intelligent analysis proposed in the above embodiment.

[0434] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements a big data information hierarchical management method and system based on fusion intelligent analysis as proposed in the above embodiments.

[0435] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0436] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0437] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0438] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0439] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0440] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A hierarchical management method for big data information based on fusion intelligent analysis, characterized in that, Includes the following steps: S1. Construct a unified representation basis for multi-source information to obtain a unified information representation matrix; S2. Use a unified information representation matrix to construct cross-source semantic-behavioral coupling and obtain the information association topology. S3. Use the information association topology to perform multi-level fusion inference and analysis to obtain the fusion risk representation vector; S4. Use the fused risk representation vector to calculate the level mapping matrix and obtain the information classification results; S5. Use the information hierarchical results and information association topology to perform structural back-embedding mapping to obtain the hierarchical management structure model.

2. The big data information hierarchical management method based on fusion intelligent analysis according to claim 1, characterized in that, Step S1 is used to convert multi-source heterogeneous information into a unified structural expression, eliminate field conflicts and inconsistencies in units caused by differences in data sources, and construct a standardized input matrix that can be directly used for association topology construction and fusion analysis. This matrix is ​​a unified information representation matrix. Its expression is: ； in, Indicates the total number of information objects; Indicates the first A unified representation vector for each information object; Each It consists of three parts assembled in sequence: ； in: Represents the structural identifier vector. ; Represents a semantic attribute vector. ; Represents a behavioral attribute vector. .

3. The big data information hierarchical management method based on fusion intelligent analysis according to claim 1, characterized in that, The unified expression basis constructed in step S1 in step S2 Based on this, the semantic and behavioral attributes between information objects are uniformly coupled and calculated to form a structured association network. ; Step S2, in constructing cross-source semantic-behavior coupling, includes the following steps: Semantic vector set construction and dimension confirmation; Semantic similarity matrix construction; Construction of behavioral vector sets and time dimension correction; Construction of behavioral correlation strength matrix; Semantic-behavior coupling mapping computation; Candidate neighborhood selection and threshold-based edge construction; Topology construction.

4. The big data information hierarchical management method based on fusion intelligent analysis according to claim 3, characterized in that, Step S2, during the construction of the semantic vector set and dimension confirmation, starts from... Extract: ； in: The fixed dimension is the semantic field concatenation result from step S1; like If it is a zero vector, then add a minimal positive number to all its dimensions. To avoid the denominator being zero; Step S2 involves setting the semantic similarity function during the construction of the semantic similarity matrix: ； in: and Both are information object serial numbers; "·" represents the vector dot product; It is a 2-norm; ； Thus, the semantic similarity matrix is ​​obtained: 。 5. A hierarchical management method for big data information based on fusion intelligent analysis according to claim 3, characterized in that, In step S2, during the construction of the behavior vector set and the time dimension correction, from Extract: ； in: For a fixed number of time windows; The time window length has been uniformly set in step S1; If the statistical period for a certain object is insufficient, it will be padded with zero values ​​to reach the required period. dimension; Step S2 uses the Pearson correlation coefficient when constructing the behavioral correlation strength matrix: ； in: for Standard deviation; like or If it is zero, then set: ; To unify the dimensions, it is linearly mapped to : ； Obtain the behavioral correlation matrix: 。 6. The big data information hierarchical management method based on fusion intelligent analysis according to claim 3, characterized in that, In step S2, when calculating the semantic-behavior coupling mapping, a fixed coupling weight is set: ； ； ； , Set as a constant during the system initialization phase; Construct a coupling function based on the set coupling weights: ； The coupling matrix is ​​obtained as follows: ； Candidate neighborhood selection and threshold-based edge construction: Partitioning is performed based on business domain identifiers, and calculations are only performed within the same domain or within a specified cross-domain scope. ; Set a fixed threshold , indicating the coupling strength Lower limit control value: ； like and Then, weighted edges are constructed, with edge weights being... .

7. A hierarchical management method for big data information based on fusion intelligent analysis according to claim 3, characterized in that, In step S2, during topology construction, a hash mapping function is set to generate a vertex set representing the graph. This is used to represent all information objects involved in the calculation of coupling relationships: ； And set the edge set: ； A graph structure is defined as: Undirected weighted graph; It does not contain self-loops; Duplicate edges are not allowed; The final output of step S2 is the information association topology. Its expression is: ； in: ； ； 。 8. The big data information hierarchical management method based on fusion intelligent analysis according to claim 1, characterized in that, The unified coupled topology structure constructed in step S2 in step S3 Based on this, a hierarchical propagation calculation is performed on the coupling relationships between nodes, and the direct coupling effects and indirect path propagation effects are unified and integrated to form a structural fusion risk quantification result for each information object. .

9. A hierarchical management method for big data information based on fusion intelligent analysis according to claim 1, characterized in that, Step S4 maps continuous fusion risk values ​​to discrete level labels, constructs structured information grading results, and transforms risk characterization results into deterministic grading outputs. Step S5, while keeping the original graph structure and topology unchanged, embeds the node hierarchical results back into the graph structure to form a unified structural model with hierarchical attributes, so that the information association structure and hierarchical results are expressed in the same model.

10. A big data information hierarchical management system based on fusion intelligent analysis, constructed based on the big data information hierarchical management method based on fusion intelligent analysis as described in any one of claims 1-9, characterized in that, Includes the following modules: Information access and standardized processing module; Information association modeling module; Integrate risk calculation module; Level mapping analysis module; Structural embedding modeling module; Hierarchical management and control module.

Images