Multi-source heterogeneous data fusion method and system based on cloud computing

Abstract

The invention relates to the technical field of data fusion, in particular to a multi-source heterogeneous data fusion method and system based on cloud computing, and the method comprises the steps: carrying out the feature extraction of multi-source heterogeneous data, and generating standardized metadata comprising a data type, a transmission protocol and a semantic tag; an adaptive interface is dynamically generated based on the metadata, the adaptive interface comprises a data format conversion module, a protocol packaging module and a semantic analysis module, and cross-domain compatible data streams are output; performing data fragment encryption and priority marking on the cross-domain compatible data stream to generate a secure transmission block carrying a transmission feature vector; incremental fusion is executed at the cloud fusion node, semantic conflicts are detected by constructing a knowledge graph, and a fusion intermediate state is generated; and performing dynamic verification on the fusion intermediate state based on the credibility evaluation model, and outputting final fusion data with a credibility label. According to the method, the dynamic verification and hierarchical labeling of the fusion result are realized, and the semantic consistency and expandability of the fusion intermediate state are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of data fusion, and in particular to a multi-source heterogeneous data fusion method and system based on cloud computing. Background Art

[0002] Against the backdrop of the continuous expansion of information and digital infrastructure, various distributed terminals, intelligent devices, business systems, and industry platforms have generated a large amount of multi-source heterogeneous data. These data usually have characteristics such as diverse formats, inconsistent semantics, and different transmission protocols, making them difficult to directly manage and analyze in a unified manner. With the development of cloud computing and edge computing environments, cross-system and cross-domain data fusion has become a key means to improve comprehensive decision-making capabilities and business intelligence levels. However, existing data fusion solutions often rely on static rules or a single data type, making it difficult to cope with data streams with large structural differences and rapid dynamic changes. In particular, there are obvious bottlenecks in semantic understanding, protocol adaptation, and fusion credibility.

[0003] On the other hand, in the process of multi-source heterogeneous data fusion, how to ensure the structural compatibility, semantic consistency and credibility of the fusion results of the data has become a core issue affecting the stable operation of the system and the interpretability of the results; current technology still lacks a complete fusion architecture with metadata-driven dynamic adaptation capabilities, executable sharding encryption and priority control mechanisms, knowledge graph-based semantic conflict detection capabilities and fusion result credibility labeling mechanisms. Therefore, it is urgent to propose a multi-source heterogeneous data fusion method and system based on a cloud computing platform with dynamic, controllable and verifiable characteristics to achieve deep data fusion and intelligent decision-making support for complex application scenarios. Summary of the Invention

[0004] Based on the above objectives, the present invention provides a multi-source heterogeneous data fusion method based on cloud computing.

[0005] The multi-source heterogeneous data fusion method based on cloud computing includes the following steps:

[0006] S1: Extract features from multi-source heterogeneous data through distributed probes to generate standardized metadata including data types, transmission protocols, and semantic labels;

[0007] S2: Dynamically generate an adaptation interface based on metadata, the adaptation interface includes a data format conversion module, a protocol encapsulation module, and a semantic parsing module, and outputs a cross-domain compatible data stream;

[0008] S3: Performs data sharding encryption and priority marking on cross-domain compatible data streams to generate secure transmission blocks that carry transmission feature vectors.

[0009] S4: Perform incremental fusion at the cloud fusion node, detect semantic conflicts by building a knowledge graph, and generate a fusion intermediate state;

[0010] S5: Dynamically verify the fusion intermediate state based on the credibility evaluation model and output the final fusion data with credibility labels.

[0011] Furthermore, the S1 includes:

[0012] S11, constructing sliding windows of raw data streams: dividing the raw data streams collected by each distributed probe into sliding windows to construct a set of raw data segments;

[0013] S12, structural feature vector extraction: extracting the structural feature vector of each original data segment;

[0014] S13, semantic label generation: inputting the structural feature vector into a semantic encoding model, performing semantic mapping on it through a neural network embedding mechanism, and generating a semantic label vector;

[0015] S14, aggregates the structural feature vector, protocol encoding and semantic label vector to construct a standardized metadata unit.

[0016] Furthermore, the S2 includes:

[0017] S21, data format conversion module: according to the data type label in the metadata, call the matching format conversion function to unify the original data fragments into a standard structure;

[0018] S22, protocol encapsulation module: selects the corresponding encapsulation template according to the protocol code in the metadata, and packages the formatted data into a standard protocol frame;

[0019] S23, semantic parsing module: Based on the semantic tags in the metadata, it performs semantic annotation operations, adds structured semantic information to the data frame, and generates a cross-domain compatible data stream with semantic self-description capabilities.

[0020] Furthermore, the S3 includes:

[0021] S31, cross-domain data sharding and encryption: Split the cross-domain compatible data stream into fixed-length segments to form a set of data segments of equal length. Use a symmetric encryption algorithm to encrypt each data segment to generate encrypted data blocks.

[0022] S32, priority tagging and secure transmission block construction: Based on the semantic label and the system-defined priority policy function, a dynamic priority weight is assigned to each encrypted data shard, and combined with the index and integrity check information, a secure transmission block is packaged and generated.

[0023] Furthermore, the S31 includes:

[0024] S311, data stream sharding: split the cross-domain compatible data stream into equal lengths according to a fixed size to obtain a data shard set;

[0025] S312, data shard encryption: encrypt each data shard using a symmetric encryption algorithm to generate an encrypted data shard.

[0026] Furthermore, the S32 includes:

[0027] S321, priority tagging: assigning a priority weight to each encrypted data shard based on the semantic tag and the system policy function;

[0028] S322, transmission feature vector construction and security block packaging: Combine the encrypted data fragments with the priority weights to construct a secure transmission block.

[0029] Furthermore, the S4 includes:

[0030] S41, incremental data injection and fusion unit construction: unpack the secure transmission block set and extract the original data content, decrypt and restore it to standard structure data fragments, and build a fusion unit based on its metadata;

[0031] S42, semantic graph construction and conflict detection: extract semantic labels and structural feature vectors from the fused unit set, map them to a unified semantic space, construct a dynamic knowledge graph, and use a conflict scoring function based on tensor decomposition to calculate the semantic conflict degree of any entity to the fused entity set;

[0032] S43, fusion intermediate state generation: Based on the knowledge graph and conflict set, perform semantic consistency correction to generate a fusion intermediate state with semantic consistency annotations as a structured result before the final fusion data output.

[0033] Furthermore, the S43 includes:

[0034] S431, conflict weighted analysis: For each pair of conflicting entities, the conflict severity is evaluated using the conflict scoring function, and the conflict impact is calculated in combination with the importance weight;

[0035] S432, graph structure adjustment and semantic consistency correction: perform structural correction of the fusion graph based on the conflict impact;

[0036] S433, fusion intermediate state generation and structure output: Using the revised knowledge graph as the core semantic relationship network, remap the data fragments in the fusion unit set to the graph entity nodes, and generate the fusion intermediate state based on the semantic consistency constraints.

[0037] Furthermore, the S5 includes:

[0038] S51, fusion intermediate state feature encoding: extract the semantic vector, data source credibility factor and time consistency index of each fusion entity in the fusion intermediate state, and construct a unified credibility feature vector;

[0039] S52, credibility assessment model reasoning: input all feature vectors in the fusion intermediate state into the credibility assessment model and output the credibility score of each entity;

[0040] S53, generate credible labels and output fusion results: according to the set credibility label threshold η, each fused entity is divided into "high credibility", "medium credibility" or "low credibility" categories to generate the final fused data set with labels.

[0041] The multi-source heterogeneous data fusion system based on cloud computing is used to implement the multi-source heterogeneous data fusion method based on cloud computing, and includes the following modules:

[0042] Distributed probe collection and feature extraction module: collects raw data from multiple heterogeneous data sources and extracts data type, transmission protocol and semantic label metadata information;

[0043] Metadata-driven adapter interface generation module: Dynamically builds adapter interfaces based on extracted metadata, completes data format conversion, protocol encapsulation, and semantic parsing, and outputs cross-domain compatible data streams with unified structure and consistent semantics.

[0044] Secure transmission block generation module: performs data sharding encryption and priority marking on cross-domain compatible data streams, generating secure transmission blocks that carry transmission feature vectors;

[0045] Cloud-based incremental fusion and graph construction module: This module performs incremental data fusion at the fusion node, detects semantic conflicts by constructing a knowledge graph, and generates a fusion intermediate state.

[0046] Credibility assessment and output module: Dynamically verifies the fusion intermediate state based on the credibility assessment model and outputs the final fusion data with credibility labels.

[0047] Beneficial effects of the present invention:

[0048] The present invention provides a multi-source heterogeneous data fusion method based on cloud computing, which can efficiently extract features from data from different sources through distributed probes, and dynamically generate adaptation interfaces based on standardized metadata, thereby realizing automatic identification and unified processing of data formats, transmission protocols and semantic tags; the method introduces a cross-domain compatibility mechanism, and performs structural standardization, encryption protection and priority marking on the data before fusion, effectively improving the security and scheduling efficiency of data cross-system interaction; at the same time, by utilizing the incremental fusion mechanism and the knowledge graph-driven semantic conflict detection capability, it can dynamically identify and correct the semantic conflict relationship between data, ensuring the semantic consistency and scalability of the fusion intermediate state.

[0049] The present invention realizes dynamic verification and hierarchical labeling of fusion results by constructing a credibility assessment model. The model comprehensively considers multi-dimensional features such as data semantics, source credibility and time consistency, and outputs the credibility score and label level of the fused entity, thereby providing credible protection for subsequent data analysis, risk warning and business decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 is a flow chart of a method according to an embodiment of the present invention;

[0052] Figure 2 2 is a system module diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0054] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0055] like Figure 1 As shown, the multi-source heterogeneous data fusion method based on cloud computing includes the following steps:

[0056] S1: Extract features from multi-source heterogeneous data through distributed probes to generate standardized metadata including data types, transmission protocols, and semantic labels;

[0057] S2: Dynamically generates an adaptation interface based on metadata. The adaptation interface includes a data format conversion module, a protocol encapsulation module, and a semantic parsing module, and outputs a cross-domain compatible data stream.

[0058] S3: Performs data sharding encryption and priority marking on cross-domain compatible data streams to generate secure transmission blocks that carry transmission feature vectors.

[0059] S4: Perform incremental fusion at the cloud fusion node, detect semantic conflicts by building a knowledge graph, and generate a fusion intermediate state;

[0060] S5: Dynamically verify the fusion intermediate state based on the credibility evaluation model and output the final fusion data with credibility labels.

[0061] S1 includes:

[0062] S11, original data stream sliding window construction: the original data stream D collected by each distributed probe i Perform sliding window division to construct a set of original data segments {W i,t}, where D i represents the original data of the i-th data source, W i,t =D i [t:t+Δt] represents the raw data segment collected by sliding with a step size of Δt;

[0063] S12, structural feature vector extraction: extract each original data segment W i,t The structural feature vector F i,t , expressed as:

[0064]

[0065] Among them, F i,t is the structural feature vector, μ(W) is the mean value, reflecting the central trend of the data, σ(W) is the standard deviation, indicating the degree of fluctuation, κ(W) is the kurtosis, used to distinguish the data form, and λ(W) is the transmission protocol code, which is generated by the probe automatically identifying the protocol stack type;

[0066] S13, semantic label generation: inputting the structural feature vector into the semantic encoding model Through the neural network embedding mechanism, it is semantically mapped to generate the semantic label vector T i,t , expressed as:

[0067]

[0068] Among them, T i,t is the semantic label vector of the i-th data source at time t, is a semantic encoding function used to map structural features into semantic label vectors, W s is the semantic embedding weight matrix, b s is a bias term, and softmax is a normalization function used to output the probability distribution of semantic categories;

[0069] S14, the structural feature vector, protocol code and semantic label vector are aggregated to construct a standardized metadata unit, which is expressed as:

[0070] M i,t = <type i ,λ(W i,t ),semTag i >

[0071] Among them, M i,t Is a standardized metadata unit, type i Is the data type label, which is determined by the type identification function (F i,t ) output, λ(W i,t ) indicates the transmission protocol code, semTag i =argmax(T i,t ), indicating semantic labels.

[0072] S2 includes:

[0073] S21, data format conversion module: According to the data type label in the metadata, the matching format conversion function is called to unify the original data fragments into a standard structure, which is expressed as:

[0074]

[0075] Among them, W i,t is the original data fragment, is a data segment after format standardization. It is a data format conversion function;

[0076] S22, protocol encapsulation module: According to the protocol code in the metadata, the corresponding encapsulation template is selected and the formatted data is packaged into a standard protocol frame, which is represented as:

[0077]

[0078] in, is the data segment after format standardization, λ(W i,t ) is the protocol type code, For the protocol encapsulation function, P i,t It is the encapsulated standard data frame;

[0079] S23, semantic parsing module: Based on the semantic tags in the metadata, it performs semantic annotation operations, adds structured semantic information to the data frame, and generates a cross-domain compatible data stream with semantic self-description capabilities, which is expressed as:

[0080]

[0081] in, The final generated cross-domain compatible data stream for multi-system interaction can be directly recognized and processed by subsequent cloud fusion nodes. i is a semantic tag.

[0082] S3 includes:

[0083] S31, cross-domain data sharding and encryption: Split the cross-domain compatible data stream into fixed-length segments to form a set of data segments of equal length. Use a symmetric encryption algorithm to encrypt each data segment to generate encrypted data blocks.

[0084] S32, priority tagging and secure transmission block construction: Based on the semantic label and the system-defined priority policy function, a dynamic priority weight is assigned to each encrypted data shard, and combined with the index and integrity check information, a secure transmission block is packaged and generated.

[0085] S31 includes:

[0086] S311, data stream sharding: the cross-domain compatible data stream is divided into equal lengths according to a fixed size L to obtain a data shard set, which is expressed as:

[0087]

[0088] in, Indicates the kth data shard, K is the total number of data shards,

[0089] S312, data shard encryption: encrypt each data shard using a symmetric encryption algorithm to generate an encrypted data shard, which is expressed as:

[0090]

[0091] Where Enc(·) represents the symmetric encryption function, K sym The symmetric encryption key generated for negotiation, is the kth data shard after using the key.

[0092] S32 includes:

[0093] S321, priority tagging: assign a priority weight to each encrypted data shard based on the semantic tag and the system policy function, expressed as:

[0094]

[0095] in, represents the transmission priority of the kth data slice. A larger value indicates a higher scheduling priority. ψ is the priority policy function, which takes as input the semantic label and the data slice sequence number and outputs the scheduling priority.

[0096] S322, transmission feature vector construction and security block packaging: Combine the encrypted data fragments with the priority weights to construct a secure transmission block, which is expressed as:

[0097]

[0098] in, It is the data fragment index and integrity check code, used for data reconstruction and integrity verification at the receiving end. For safe transmission of blocks.

[0099] S4 includes:

[0100] S41, incremental data injection and fusion unit construction: unpack the secure transmission block set and extract the original data content, decrypt and restore it to standard structure data fragments, and combine its metadata to build a fusion unit, expressed as:

[0101]

[0102] in, is a fusion unit, which represents a fusionable data instance with semantic labels and protocol information. It is the standard structure data fragment after decryption and restoration;

[0103] S42, Semantic Graph Construction and Conflict Detection: Extracting Fusion Unit Sets The semantic labels and structural feature vectors in are mapped to a unified semantic space to construct a dynamic knowledge graph. The conflict scoring function based on tensor decomposition is used to calculate the semantic conflict degree of any entity pair and fusion entity set, which is expressed as:

[0104]

[0105] (e a ,e b )∈ε;

[0106]

[0107] in, is a knowledge graph, including entity sets and semantic relationship sets, ε represents the fusion entity set, Represents a set of relationships, determined by the semantic similarity between entities, represents the entity embedding vector function, cos(·,·) represents the vector cosine similarity, e a and e b are two entity nodes in the constructed knowledge graph, γ∈[0,2], γ(e a ,e b ) is the conflict scoring function. The larger the value, the greater the conflict. If γ(e a ,e b )>θ, it is considered that there is a semantic conflict and recorded in the conflict pair set

[0108] S43, fusion intermediate state generation: Based on the knowledge graph and conflict set, perform semantic consistency correction to generate a fusion intermediate state with semantic consistency annotations as a structured result before the final fusion data output.

[0109] S43 includes:

[0110] S431, conflict weighted analysis: for each pair of conflicting entities The conflict severity is evaluated by the conflict scoring function, and the conflict impact is calculated by combining the importance weights, which is expressed as:

[0111]

[0112] Among them, α(e) is the entity importance weight, which reflects the entity’s connectivity or semantic coreness in the graph, δ(ea ,e b ) is the conflict impact, which indicates the potential interference degree of the conflict on the overall semantic fusion result;

[0113] S432, Graph structure adjustment and semantic consistency correction: Based on the conflict impact δ(e a ,e b ) performs structural correction of the fusion graph. Specific strategies include:

[0114] (1) Implement semantic merging for conflicting entity pairs. If the conflict degree is lower than a certain threshold θ1, their representation vectors are merged, which can be expressed as:

[0115]

[0116] Among them, β∈[0,1] is the merging weight, θ1∈[0.1,0.3];

[0117] (2) Implementing Entity Decoupling for entity pairs whose conflict degree is higher than the threshold θ2 (θ2∈[0.6,0.8]), disconnecting the direct relationship from the graph and retaining their respective branches;

[0118] (3) For moderate conflicts (θ1<δ<θ2), semantic alignment algorithms (such as WordNet mapping, ontology mapping) are used to try to automatically correct labels or reconstruct edge weights.

[0119] The corrected knowledge graph is represented as A fusion graph indicating that semantic conflict mediation has been completed;

[0120] S433, Fusion Intermediate State Generation and Structure Output: Based on the Corrected Knowledge Graph As the core semantic relationship network, the fusion unit set Data fragment in Remap to the graph entity nodes and generate the fusion intermediate state based on the semantic consistency constraint, which is expressed as:

[0121]

[0122] Aggregate(·) is a fusion operator, which represents the fusion method of multi-source data under the same semantic entity (average, weighted, retain all), and Context(e) is the context information of the entity node, including source, timestamp, and weight. The intermediate state of fusion is a temporary fusion result set with consistent semantics and standardized structure, which can be further used for trusted verification.

[0123] S5 includes:

[0124] S51, fusion intermediate state feature encoding: Extract the semantic vector, data source credibility factor and time consistency index of each fusion entity in the fusion intermediate state, and construct a unified credibility feature vector, which is expressed as:

[0125]

[0126] Among them, z j is the credible feature vector of the jth fusion entity, used as the model input, is the semantic embedding vector of the fused entity, f j is the jth fusion entity in the fusion intermediate state, ρ j is the data source credibility factor of the fusion entity, reflecting the historical credibility level of its source equipment / system (such as equipment failure rate, network stability), τ j It is the temporal consistency indicator of the fused entity, measuring its degree of synchronization with other entities on the timeline;

[0127] S52, credibility assessment model reasoning: All feature vectors in the fusion intermediate state are input into the credibility assessment model, and the credibility score of each entity is output, which is expressed as:

[0128]

[0129] Among them, c j ∈[0,1], represents the fusion entity f j The closer to 1, the more reliable it is. It is a credibility assessment model;

[0130] S53, credible label generation and output fusion result: according to the set credibility label threshold η∈[η low ,η high ], classify each fused entity into "high confidence", "medium confidence" or "low confidence" categories, and generate the final fused dataset with labels, which is expressed as:

[0131]

[0132]

[0133] in, is the final fused dataset with credibility labels, label j is the credibility level classification label of the jth fusion entity, which is divided into high, medium and low, η low and η high is the threshold for classifying credible labels, η low ∈[0.2,0.4],η high ∈[0.6,0.8].

[0134] like Figure 2 As shown, the multi-source heterogeneous data fusion system based on cloud computing is used to implement the above multi-source heterogeneous data fusion method based on cloud computing, and includes the following modules:

[0135] Distributed probe collection and feature extraction module: collects raw data from multiple heterogeneous data sources and extracts data type, transmission protocol and semantic label metadata information;

[0136] Metadata-driven adapter interface generation module: Dynamically builds adapter interfaces based on extracted metadata, completes data format conversion, protocol encapsulation, and semantic parsing, and outputs cross-domain compatible data streams with unified structure and consistent semantics.

[0137] Secure transmission block generation module: performs data sharding encryption and priority marking on cross-domain compatible data streams, generating secure transmission blocks that carry transmission feature vectors;

[0138] Cloud-based incremental fusion and graph construction module: This module performs incremental data fusion at the fusion node, detects semantic conflicts by constructing a knowledge graph, and generates a fusion intermediate state.

[0139] Credibility assessment and output module: Dynamically verifies the fusion intermediate state based on the credibility assessment model and outputs the final fusion data with credibility labels.

[0140] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention is limited to these examples. Within the scope of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity.

Claims

1. A multi-source heterogeneous data fusion method based on cloud computing, characterized by: The following steps are involved: S1: Extract features from multi-source heterogeneous data through distributed probes to generate standardized metadata including data types, transmission protocols, and semantic labels; S2: Dynamically generate an adaptation interface based on metadata, the adaptation interface includes a data format conversion module, a protocol encapsulation module, and a semantic parsing module, and outputs a cross-domain compatible data stream; S3: Performs data sharding encryption and priority marking on cross-domain compatible data streams to generate secure transmission blocks that carry transmission feature vectors. S4: Perform incremental fusion at the cloud fusion node, detect semantic conflicts by building a knowledge graph, and generate a fusion intermediate state; S5: Dynamically verify the fusion intermediate state based on the credibility evaluation model and output the final fusion data with credibility labels.

2. The multi-source heterogeneous data fusion method based on cloud computing according to claim 1 is characterized in that: Said S1 comprises: S11, constructing sliding windows of raw data streams: dividing the raw data streams collected by each distributed probe into sliding windows to construct a set of raw data segments; S12, structural feature vector extraction: extracting the structural feature vector of each original data segment; S13, semantic label generation: inputting the structural feature vector into a semantic encoding model, performing semantic mapping on it through a neural network embedding mechanism, and generating a semantic label vector; S14, aggregates the structural feature vector, protocol encoding and semantic label vector to construct a standardized metadata unit.

3. The multi-source heterogeneous data fusion method based on cloud computing according to claim 2 is characterized in that: The S2 includes: S21, data format conversion module: according to the data type label in the metadata, call the matching format conversion function to unify the original data fragments into a standard structure; S22, protocol encapsulation module: selects the corresponding encapsulation template according to the protocol code in the metadata, and packages the formatted data into a standard protocol frame; S23, semantic parsing module: Based on the semantic tags in the metadata, it performs semantic annotation operations, adds structured semantic information to the data frame, and generates a cross-domain compatible data stream with semantic self-description capabilities.

4. The multi-source heterogeneous data fusion method based on cloud computing according to claim 3 is characterized in that: The S3 includes: S31, cross-domain data sharding and encryption: Split the cross-domain compatible data stream into fixed-length segments to form a set of data segments of equal length. Use a symmetric encryption algorithm to encrypt each data segment to generate encrypted data blocks. S32, priority tagging and secure transmission block construction: Based on the semantic label and the system-defined priority policy function, a dynamic priority weight is assigned to each encrypted data shard, and combined with the index and integrity check information, a secure transmission block is packaged and generated.

5. The multi-source heterogeneous data fusion method based on cloud computing according to claim 4 is characterized in that: The S31 includes: S311, data stream sharding: split the cross-domain compatible data stream into equal lengths according to a fixed size L to obtain a data shard set; S312, data shard encryption: encrypt each data shard using a symmetric encryption algorithm to generate an encrypted data shard.

6. The multi-source heterogeneous data fusion method based on cloud computing according to claim 4 is characterized in that: The S32 includes: S321, priority tagging: assigning a priority weight to each encrypted data shard based on the semantic tag and the system policy function; S322, transmission feature vector construction and security block packaging: Combine the encrypted data fragments with the priority weights to construct a secure transmission block.

7. The multi-source heterogeneous data fusion method based on cloud computing according to claim 6 is characterized in that: The S4 includes: S41, incremental data injection and fusion unit construction: unpack the secure transmission block set and extract the original data content, decrypt and restore it to standard structure data fragments, and build a fusion unit based on its metadata; S42, semantic graph construction and conflict detection: extract semantic labels and structural feature vectors from the fused unit set, map them to a unified semantic space, construct a dynamic knowledge graph, and use a conflict scoring function based on tensor decomposition to calculate the semantic conflict degree of any entity to the fused entity set; S43, fusion intermediate state generation: Based on the knowledge graph and conflict set, perform semantic consistency correction to generate a fusion intermediate state with semantic consistency annotations as a structured result before the final fusion data output.

8. The multi-source heterogeneous data fusion method based on cloud computing according to claim 7 is characterized in that: The S43 includes: S431, conflict weighted analysis: For each pair of conflicting entities, the conflict severity is evaluated using the conflict scoring function, and the conflict impact is calculated in combination with the importance weight; S432, graph structure adjustment and semantic consistency correction: perform structural correction of the fusion graph based on the conflict impact; S433, fusion intermediate state generation and structure output: Using the revised knowledge graph as the core semantic relationship network, remap the data fragments in the fusion unit set to the graph entity nodes, and generate the fusion intermediate state based on the semantic consistency constraints.

9. The multi-source heterogeneous data fusion method based on cloud computing according to claim 8 is characterized in that: The S5 includes: S51, fusion intermediate state feature encoding: extract the semantic vector, data source credibility factor and time consistency index of each fusion entity in the fusion intermediate state, and construct a unified credibility feature vector; S52, credibility assessment model reasoning: input all feature vectors in the fusion intermediate state into the credibility assessment model and output the credibility score of each entity; S53, generate credible labels and output fusion results: according to the set credibility label threshold η, each fused entity is divided into "high credibility", "medium credibility" or "low credibility" categories to generate the final fused data set with labels.

10. A multi-source heterogeneous data fusion system based on cloud computing, used to implement the multi-source heterogeneous data fusion method based on cloud computing according to any one of claims 1 to 9, characterized in that: Includes the following modules: Distributed probe collection and feature extraction module: collects raw data from multiple heterogeneous data sources and extracts data type, transmission protocol and semantic label metadata information; Metadata-driven adapter interface generation module: Dynamically builds adapter interfaces based on extracted metadata, completes data format conversion, protocol encapsulation, and semantic parsing, and outputs cross-domain compatible data streams with unified structure and consistent semantics. Secure transmission block generation module: performs data sharding encryption and priority marking on cross-domain compatible data streams, generating secure transmission blocks that carry transmission feature vectors; Cloud-based incremental fusion and graph construction module: This module performs incremental data fusion at the fusion node, detects semantic conflicts by constructing a knowledge graph, and generates a fusion intermediate state. Credibility assessment and output module: Dynamically verifies the fusion intermediate state based on the credibility assessment model and outputs the final fusion data with credibility labels.

Images