A method and apparatus for determining the importance of network nodes based on dual-anchor point iteration.

By employing a dual-anchor-point iterative method to perform information propagation and deterministic iteration on the hypergraph, the problem of high computational and storage overhead in existing technologies is solved, and efficient and stable evaluation of node importance is achieved.

CN123115476APending Publication Date: 2026-11-17CHINA AGRI UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611332913.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-31
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing graph or hypergraph scoring methods based on random walks require pre-calculation and storage of transition probabilities between nodes or between a node and a hyperedge when calculating node relationships. This results in high computational and storage overhead, and the scoring results are affected by random sampling paths, making it impossible to accurately assess the importance of nodes in complex networks.

Method used

A dual-anchor-point iterative approach is adopted. By constructing a node-hyperedge weighted association model, information is propagated using the hypergraph direct propagation operator, and deterministic iterative updates are performed to generate a third-order symmetric tensor. The importance of nodes is directly calculated, avoiding the construction of transition probabilities and random path sampling.

Benefits of technology

It reduces processor computations and memory access operations related to transition probability calculations and random sampling, improves the repeatability and stability of scoring results, and can accurately assess node importance in large-scale networks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115476A_ABST
    Figure CN123115476A_ABST
Patent Text Reader

Abstract

This invention provides a method and apparatus for determining the importance of network nodes based on dual-anchor-point iteration, relating to the fields of graph data processing and high-order relation scoring. The method constructs a node-hyperedge weighted association model, utilizes the hypergraph direct propagation operator to propagate high-order relation information between nodes, and establishes a deterministic iterative update mechanism for ordered anchor pairs, ensuring that node scores are simultaneously influenced by propagation relations, initial anchor constraints, and attracting anchor constraints, thus obtaining the convergent node score vectors corresponding to each ordered anchor pair. Further, by fusing node contributions from different anchor directions, a triplet score tensor is generated. Finally, for each node, all distinct triplet scores containing that node are extracted from the triplet score tensor, and the average of the extracted triplet scores is calculated to obtain the node's importance score. This method obtains triplet scores and node importance scores deterministically, avoiding result fluctuations caused by random path sampling.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of graph data processing and high-order relation scoring technology, and more specifically, to a method and apparatus for determining the importance of network nodes based on dual-anchor point iteration. Background Technology

[0002] In complex network analysis, especially in scenarios with high-order interactions, node importance assessment is one of the core issues in understanding network structure and function. Accurately identifying key nodes in a network is of great significance for practical applications such as bioinformatics analysis, social network analysis, and recommender systems.

[0003] However, existing graph or hypergraph scoring methods based on random walks require pre-calculating and storing the transition probabilities between nodes or between a node and a hyperedge when calculating node relationships, and obtaining the scoring results through random path sampling. For weighted hypergraphs with a large number of nodes and hyperedges, the calculation, storage, and retrieval of transition probabilities increase the computational load and memory access overhead. To obtain more stable results, multiple random samplings are usually required, further increasing redundant calculations and causing fluctuations in scoring results under the same input conditions due to the influence of random sampling paths. Summary of the Invention

[0004] In view of this, the present invention provides a method and apparatus for determining the importance of network nodes based on dual-anchor point iteration.

[0005] One aspect of the present invention provides a method for determining the importance of network nodes based on dual-anchor-point iteration, comprising: acquiring node data and multivariate relation data in the network to be analyzed, and generating a hypergraph based on the node data and multivariate relation data, wherein the hypergraph includes a set of nodes, a set of hyperedges, and a weighted correlation matrix between nodes and hyperedges; calculating the node weighting degree and the hyperedge weighting degree based on the weighted correlation matrix, and determining a propagation operator based on the node weighting degree and the hyperedge weighting degree, wherein the propagation operator is used to perform information propagation on the hypergraph; for any two ordered anchor point pairs consisting of different nodes, selecting one node as the starting anchor point and the other node as the second anchor point. Anchor points are attracted, and the unit vector corresponding to the initial anchor point is used as the score vector of the initial node. A deterministic iterative update is performed using the propagation operator to obtain the score vector corresponding to each ordered anchor pair. The score vector corresponding to each ordered anchor pair represents the numerical value of each node's contribution to the anchor relationship in the current iteration state. For any three distinct nodes, target node components are extracted from the score vectors corresponding to their ordered anchor pairs, and these target node components are aggregated to obtain a third-order symmetric tensor. This third-order symmetric tensor represents the scoring result of the node triplet relationship. Based on the third-order symmetric tensor, the importance of each node in the network to be analyzed is determined.

[0006] According to an embodiment of the present invention, each element value in the weighted association matrix represents the association weight between the corresponding node and the corresponding hyperedge; wherein, when the element value is greater than zero, it indicates that there is an association relationship between the corresponding node and the corresponding hyperedge, and when the element value is equal to zero, it indicates that there is no association relationship between the corresponding node and the corresponding hyperedge; the method further includes: removing nodes with a node weight of zero and hyperedges with a hyperedge weight of zero from the hypergraph.

[0007] According to an embodiment of the present invention, the node weighting degree and the hyperedge weighting degree are calculated based on the weighted association matrix, including: for any node, summing up the association weights of all hyperedges corresponding to the node to obtain the node weighting degree; for any hyperedge, summing up the association weights of all nodes corresponding to the hyperedge to obtain the hyperedge weighting degree.

[0008] According to an embodiment of the present invention, the unit vector corresponding to the starting anchor point is used as the score vector of the initial node, and deterministic iterative updates are performed using a propagation operator, including: deterministic iteration using a first update method and a second update method; wherein, the weight of the first update method is ρ, which is used to propagate node association information according to the structure of the hypergraph; the weight of the second update method is (1-ρ), wherein the second update method returns to the starting anchor point with a conditional probability of 1 / 2 and reaches the attracting anchor point with a conditional probability of 1 / 2; the weights of returning to the starting anchor point and reaching the attracting anchor point in the second update method are both (1-ρ) / 2.

[0009] According to an embodiment of the present invention, the unit vector corresponding to the starting anchor point is used as the score vector of the initial node, and deterministic iterative updates are performed using the propagation operator. The method further includes: when the node score vector satisfies the preset convergence condition or reaches the preset maximum number of iterations, stopping the iteration of the current ordered anchor pair and saving the corresponding converged score vector to obtain the score vector corresponding to each ordered anchor pair.

[0010] According to an embodiment of the present invention, for any three distinct nodes, the target node component is extracted from the score vector corresponding to their ordered anchor pairs, and the target node component is aggregated to obtain a third-order symmetric tensor. This includes: for any three distinct nodes, extracting the target node component from the score vector corresponding to six ordered anchor pairs, and performing average aggregation processing on the target node component to obtain the original triplet score; and performing min-max normalization processing on the original triplet score to obtain a third-order symmetric tensor.

[0011] According to an embodiment of the present invention, the importance of each node in the network to be analyzed is determined based on a third-order symmetric tensor, including: for any node, obtaining the triple scores corresponding to all distinct triples containing the current node from the third-order symmetric tensor, and averaging the triple scores to obtain the importance score of the current node; traversing all nodes in the node set to obtain the importance scores of all nodes.

[0012] Another aspect of the present invention provides a network node importance determination device based on dual-anchor point iteration, comprising: an acquisition module, configured to acquire node data and multivariate relation data in the network to be analyzed, and generate a hypergraph based on the node data and multivariate relation data, wherein the hypergraph includes a set of nodes, a set of hyperedges, and a weighted correlation matrix between nodes and hyperedges; a determination module, configured to calculate the node weighting degree and the hyperedge weighting degree based on the weighted correlation matrix, and determine a propagation operator based on the node weighting degree and the hyperedge weighting degree, wherein the propagation operator is used to perform information propagation on the hypergraph; and a deterministic iteration module, configured to select one node as the starting anchor point and the other node as the starting anchor point for any two different ordered anchor point pairs. Points are used as attraction anchors, and the unit vector corresponding to the initial anchor is used as the score vector of the initial node. A deterministic iterative update is performed using a propagation operator to obtain the score vector corresponding to each ordered anchor pair. The score vector corresponding to each ordered anchor pair represents the numerical value of each node's contribution to the anchor relationship in the current iteration state. An aggregation module is used to extract target node components from the score vectors corresponding to any three distinct nodes' ordered anchor pairs, and aggregate these target node components to obtain a third-order symmetric tensor. This third-order symmetric tensor represents the scoring result of the node triplet relationship. A node importance scoring module is used to determine the importance of each node in the network under analysis based on the third-order symmetric tensor.

[0013] Another aspect of the present invention provides an electronic device comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method described above.

[0014] Another aspect of the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed, are used to implement the method described above.

[0015] Another aspect of the present invention provides a computer program product including computer-executable instructions that, when executed, are used to implement the method described above.

[0016] The method and apparatus for determining the importance of network nodes based on dual-anchor point iteration provided in this invention have at least the following beneficial effects:

[0017] The network node importance determination method and apparatus based on dual-anchor-point iteration provided in this invention constructs a node-hyperedge weighted association model, utilizes the hypergraph direct propagation operator to propagate high-order association information of nodes, and establishes a deterministic iterative update mechanism for ordered anchor pairs, so that node scores are simultaneously affected by propagation relationships, initial anchor constraints, and attracting anchor constraints, obtaining converged node score vectors for each ordered anchor pair. Further, by fusing node contributions under different anchor directions, a triplet score tensor is generated. Finally, for each node, all distinct triplet scores containing that node are extracted from the triplet score tensor, and the average of the extracted triplet scores is calculated to obtain the importance score of that node. This method does not require the construction and storage of a node transition probability table, nor does it require random path sampling, thus saving processor operations and memory access operations related to transition probability calculation, probability data storage, and random sampling. Under the same input data, initial vector, and iteration parameters, consistent convergence results can be obtained, improving the repeatability and stability of the scoring results. Attached Figure Description

[0018] The above and other objects, features and advantages of the present invention will become more apparent from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

[0019] Figure 1 A flowchart illustrating a method for determining the importance of network nodes based on dual-anchor point iteration according to an embodiment of the present invention is shown.

[0020] Figure 2 This schematic diagram illustrates the structural block diagram of a network node importance determination device based on dual-anchor point iteration according to an embodiment of the present invention.

[0021] Figure 3 The diagram illustrates a structural block diagram of an electronic device suitable for implementing a method for determining the importance of network nodes based on dual-anchor-point iteration according to an embodiment of the present invention. Detailed Implementation

[0022] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the invention. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the invention for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0023] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0024] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0025] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0026] In the embodiments of this invention, the collection, updating, analysis, processing, use, transmission, provision, disclosure, and storage of data (e.g., including but not limited to user personal information) comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures have been taken to prevent unauthorized access to user personal information data and to maintain the security of user personal information and network security.

[0027] Explanation of key terms:

[0028] (1) Graph (G): A data structure consisting of a set of nodes V and a set of edges E, denoted as G=(V, E), where each edge connects two nodes.

[0029] (2) Hypergraph (H): A generalized form of graph, denoted as H=(V, E), where each hyperedge can connect any number of nodes, i.e. .

[0030] (3) Node Importance: An indicator that measures the influence or criticality of nodes in a network, also known as node centrality.

[0031] Currently, common methods for evaluating node importance in existing technologies mainly include degree centrality methods, PageRank algorithm, and node2vec algorithm.

[0032] (1) Regarding Degree Centrality (DC):

[0033] Degree centrality is the simplest method for evaluating node importance. Its core idea is that the higher the degree of a node, the more important the node is.

[0034] The formula for calculating degree centrality is: DC(v) = deg(v) / (n-1).

[0035] Where deg(v) represents the degree of node v, and n represents the total number of nodes in the network.

[0036] (2) Regarding the PageRank algorithm:

[0037] The PageRank algorithm was originally used to rank the importance of web pages. Its core idea is that the more important a node is, the more important it is.

[0038] The formula for calculating PageRank is: PR(v) = (1-d) / n + d × Σ [PR(u) / out_deg(u)].

[0039] Where d represents the damping coefficient (usually taken as 0.85), and out_deg(u) represents the out-degree of node u.

[0040] (3) Regarding the node2vec algorithm:

[0041] node2vec is a network representation learning algorithm based on random walks, which introduces a return parameter p and input / output parameters q to control the walk strategy:

[0042] p(Return Parameter): Controls the probability of returning to the previous node.

[0043] q (In-Out Parameter): Controls the probability of exploring the local or global area.

[0044] The above three methods have the following problems:

[0045] Problem 1: Unable to effectively represent higher-order interaction relationships.

[0046] For example, existing technologies are based on ordinary graph modeling, where each edge can only connect two nodes. However, in complex real-world systems, many relationships are multi-dimensional (higher-order). Existing technologies forcibly decompose these multi-dimensional relationships into multiple binary relationships, resulting in significant information loss and an inability to accurately capture the true structure of the system.

[0047] Problem 2: Lack of a two-way bias control mechanism.

[0048] While the node2vec algorithm introduces p and q parameters, it only considers walks based on binary edges and lacks a bias mechanism designed for the specific structure of the hypergraph. In the real world, information propagation often exhibits directionality and bias, and current technologies cannot flexibly implement such bidirectional bias control.

[0049] Question 3: Limited application scenarios and poor universality.

[0050] Many existing methods are designed for specific domains; for example, some drug synergistic prediction methods are specifically designed for drug-disease networks and are difficult to transfer to other domains. This leads to problems such as the need to redevelop algorithms for each domain, difficulty in transferring cross-domain knowledge, high R&D costs, and long development cycles.

[0051] Problem 4: Low computational efficiency.

[0052] Hypergraph structures are more complex than ordinary graphs, and performing random walks directly on hypergraphs often faces computational efficiency issues, especially for large-scale networks: the number and size of hyperedges are large, the calculation of transition probabilities is complex, and the sampling process is time-consuming, which limits the application of existing technologies in real-world big data scenarios.

[0053] In summary, existing graph or hypergraph scoring methods based on random walks require pre-calculating and storing the transition probabilities between nodes or between a node and a hyperedge when calculating node relationships, and then obtaining the scoring results through random path sampling. For weighted hypergraphs with a large number of nodes and hyperedges, the calculation, storage, and retrieval of transition probabilities increase the computational load and memory access overhead. To obtain more stable results, multiple random samplings are usually required, further increasing redundant calculations and causing fluctuations in scoring results under the same input conditions due to the influence of random sampling paths.

[0054] Based on this, the present invention provides a method for obtaining triplet scoring of any three different nodes through deterministic dual-anchor point iteration without constructing a transition probability table between nodes or performing random sampling. This method aims to reduce the computational and storage overhead caused by the construction and storage of transition probabilities and random path sampling in the weighted hypergraph node relationship scoring process, and to reduce the impact of random sampling on the stability of the scoring results.

[0055] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0056] Figure 1 A flowchart illustrating a method for determining the importance of network nodes based on dual-anchor point iteration according to an embodiment of the present invention is shown.

[0057] like Figure 1 As shown, the network node importance determination method based on dual-anchor point iteration in this embodiment may include, for example, operations S1 to S5.

[0058] In operation S1, node data and multivariate relation data in the network to be analyzed are obtained, and a hypergraph is generated based on the node data and multivariate relation data. The hypergraph includes a set of nodes, a set of hyperedges, and a weighted correlation matrix between nodes and hyperedges.

[0059] In operation S2, the node weighted degree and hyperedge weighted degree are calculated based on the weighted correlation matrix, and the propagation operator is determined based on the node weighted degree and hyperedge weighted degree, wherein the propagation operator is used to perform information propagation on the hypergraph.

[0060] In operation S3, for any two different nodes forming an ordered anchor pair, one node is selected as the starting anchor and the other node is selected as the attracting anchor. The unit vector corresponding to the starting anchor is used as the score vector of the initial node. The propagation operator is used to perform deterministic iterative updates to obtain the score vector corresponding to each ordered anchor pair. The score vector corresponding to each ordered anchor pair represents the numerical value of the contribution of each node to the anchor relationship in the current iteration state.

[0061] In operation S4, for any three distinct nodes, the target node component is extracted from the score vector corresponding to their ordered anchor pairs, and the target node component is aggregated to obtain a third-order symmetric tensor, which represents the scoring result of the node triplet relationship.

[0062] In operation S5, the importance of each node in the network to be analyzed is determined based on the third-order symmetric tensor.

[0063] For example, in this embodiment, the application scenario of this network node importance determination method based on dual-anchor iteration can be drug collaborative prediction, identifying key drugs, diseases, and targets. That is, the network to be analyzed can be a drug-disease-target network.

[0064] To make the formulas involved in this invention clearer, the symbols and meanings used throughout the text will be explained below, as shown in Table 1.

[0065] Table 1. Symbols and Meanings in the Full Text

[0066]

[0067] In this embodiment, the network node importance determination method based on dual-anchor point iteration can be divided into three levels. For example:

[0068] The input layer can be a constructed hypergraph (i.e., a weighted hypergraph). , where the set of nodes is The set of hyperedges is The weighted correlation matrix is ,and .

[0069] The intermediate layer can be the convergent score vector of ordered anchor pairs and the third-order symmetric score tensor. .

[0070]

[0071] in, Represents a node The score of the triplet.

[0072] The final output layer can be a node importance score vector I.

[0073] .

[0074] The technical link of this method for determining the importance of network nodes based on dual-anchor point iteration is as follows:

[0075] Weighted hypergraph data → Deterministic iteration with two anchors → Triple raw scores → Third-order symmetric score tensor S → Node importance score I i → Node importance score vector I.

[0076] The network node importance determination method based on dual-anchor point iteration provided in this invention constructs a node-hyperedge weighted association model, utilizes the hypergraph direct propagation operator to propagate high-order association information between nodes, and establishes a deterministic iterative update mechanism for ordered anchor pairs. This ensures that node scores are simultaneously influenced by propagation relationships, initial anchor constraints, and attracting anchor constraints, obtaining converged node score vectors for each ordered anchor pair. Furthermore, by fusing node contributions from different anchor directions, a triplet score tensor is generated. Finally, for each node, all distinct triplet scores containing that node are extracted from the triplet score tensor, and the average of these extracted triplet scores is calculated to obtain the node's importance score. This method eliminates the need to construct and store a transition probability table between nodes, and avoids random path sampling, thus saving processor computations and memory access operations related to transition probability calculation, probability data storage, and random sampling. With identical input data, initial vectors, and iteration parameters, it achieves consistent convergence results, improving the repeatability and stability of the scoring results.

[0077] According to an embodiment of the present invention, the weighted correlation matrix Each element value in the array represents the association weight between the corresponding node and the corresponding hyperedge.

[0078] When the element value is greater than zero, it indicates that the corresponding node and the corresponding hyperedge are associated; when the element value is equal to zero, it indicates that the corresponding node and the corresponding hyperedge are not associated.

[0079] Accordingly, the method for determining the importance of network nodes based on dual-anchor point iteration may, for example, include:

[0080] Nodes with a weighted degree of zero and hyperedges with a weighted degree of zero are removed from the hypergraph to prevent unrelated nodes or invalid hyperedges from participating in subsequent weighted hypergraph propagation operations.

[0081] According to an embodiment of the present invention, in operation S2, the node weighting degree and the hyperedge weighting degree are calculated based on the weighted correlation matrix, which may include, for example:

[0082] For any node, the weights of all hyperedges associated with the node are summed to obtain the node's weighted degree; for any hyperedge, the weights of all nodes associated with the hyperedge are summed to obtain the hyperedge's weighted degree.

[0083] For example, in this embodiment, the weighted degree of the calculated nodes is... and super-edge weighted degree Specifically, the following methods can be used:

[0084]

[0085] The nodes and hyperedges involved in the calculation should satisfy the following conditions: and .

[0086] According to an embodiment of the present invention, the propagation operator is determined in operation S2 based on the node weighted degree and the hyperedge weighted degree, for example, in the following manner:

[0087] For any node score vector x, the propagation operator Φ(·) is defined as:

[0088]

[0089] This method directly performs weighted aggregation of "node to hyperedge and hyperedge to node", without needing to calculate or store the transition probability between any two nodes.

[0090] According to an embodiment of the present invention, in operation S3, the unit vector corresponding to the initial anchor point is used as the score vector of the initial node, and a deterministic iterative update is performed using the propagation operator to obtain the score vector corresponding to each ordered anchor point pair. For example, it may include:

[0091] Deterministic iteration is performed using the first and second update methods.

[0092] For example, in this embodiment, two different nodes can be selected sequentially as ordered anchor point pairs. Where o is the initial anchor point and a is the attraction anchor point. Then, the unit vector of the initial anchor point is used. Perform initialization:

[0093]

[0094] Then, deterministic iterative updates are performed as follows:

[0095]

[0096] Specifically, this deterministic iterative update method includes two update methods, namely the first update method and the second update method.

[0097] Regarding the first update method:

[0098] The weight of this first update method is ρ, which is used to propagate node association information according to the structure of the hypergraph.

[0099] Regarding the second update method:

[0100] The total weight of the second update method is (1-ρ), where the second update method returns to the starting anchor point with a conditional probability of 1 / 2 and reaches the attracting anchor point with a conditional probability of 1 / 2. The weights for returning to the starting anchor point and reaching the attracting anchor point in the second update method are both (1-ρ) / 2.

[0101] In this embodiment, the second update method simultaneously reflects the starting anchor point and the attracting anchor point, and the weights of the two components are clearly defined.

[0102] In this deterministic iterative update method, the above probabilities are only used to illustrate the weights of the three components in the update formula. It is not necessary to generate random numbers, simulate paths, or count the number of visits, nor is it necessary to calculate the transition probabilities between nodes separately.

[0103] According to an embodiment of the present invention, in operation S3, the unit vector corresponding to the initial anchor point is used as the score vector of the initial node, and deterministic iterative updates are performed using the propagation operator. For example, it may also include:

[0104] When the node score vector meets the preset convergence condition or reaches the preset maximum number of iterations, the iteration of the current ordered anchor pair is stopped, and the corresponding converged score vector is saved to obtain the score vector corresponding to each ordered anchor pair.

[0105] For example, in this embodiment, for each ordered anchor pair All methods use deterministic iterative formulas composed of propagation components, initial anchor components, and attraction anchor components. Finally, convergence is checked and the results are saved. For example:

[0106] If the following conditions are met: If so, then stop iterating over the current anchor pair and let .

[0107] If it has not yet converged, then let Then, return to the deterministic iterative update process and continue the calculation.

[0108] In addition, when the preset maximum number of iterations is reached... When that happens, stop the current iteration and save the last result.

[0109] After completing one anchor pair, continue processing the next one that satisfies the condition. The ordered anchor point pairs are processed until all are completed.

[0110] According to an embodiment of the present invention, operation S4, for any three distinct nodes, extracts the target node component from the score vector corresponding to their ordered anchor pairs, and aggregates the target node components to obtain a third-order symmetric tensor, which may include, for example:

[0111] For any three distinct nodes, extract the target node component from the corresponding score vector of the six ordered anchor pairs, and perform average aggregation on the target node component to obtain the original triplet score; perform min-max normalization on the original triplet score to obtain a third-order symmetric tensor.

[0112] For example, in this embodiment, for any three distinct nodes i, j, k, the corresponding node components can be extracted from the results of the six ordered anchor pairs, and the triplet score can be calculated. :

[0113]

[0114] The average of the above six factors ensures that the order of the nodes does not affect the result, that is:

[0115]

[0116] In this embodiment, the triple score, which is independent of the node arrangement order, can be directly obtained through six-item symmetric aggregation.

[0117] Then, the raw scores of all distinct triples are subjected to min-max normalization:

[0118]

[0119] in, and Let and represent the minimum and maximum original scores of all distinct triples, respectively. When they are equal, let represent the minimum and maximum scores of all distinct triples. .when When there are duplicate indices, let .

[0120] Finally, a third-order symmetric tensor is obtained. Each of them satisfies The tensor elements are triples. The final score result.

[0121] In this embodiment, the final result is explicitly represented as a third-order tensor. ,in, It can be used directly for sorting or filtering triples.

[0122] For example, taking a drug synergy prediction scenario, let ρ = 0.8. Then, in each update, the weight of the hypergraph propagation component is 0.8, the weight of returning to the starting anchor point is 0.1, and the weight of reaching the attraction anchor point is 0.1. After iterating over all ordered anchor pairs, for any distinct nodes i, j, k, the six corresponding components are extracted according to operation S4, averaged, and normalized to obtain the final score s of the triple in the third-order symmetric tensor S. ijk .

[0123] Finally, through this final score s ijk The importance of each node in the network to be analyzed can be determined.

[0124] According to an embodiment of the present invention, operation S5 is based on a third-order symmetric tensor to determine the importance of each node in the network to be analyzed, and may include, for example:

[0125] For any node, obtain the triple scores corresponding to all distinct triples containing the current node from the third-order symmetric tensor, and average the triple scores to obtain the importance score of the current node; traverse all nodes in the node set to obtain the importance scores of all nodes.

[0126] For example, in this embodiment, for node i, a set of distinct triplet indexes containing node i is defined. :

[0127]

[0128] The importance score of node i is defined as the average score of the triplet corresponding to the index set. :

[0129]

[0130] Given that the current algorithm scores all distinct triples... Therefore, the average It can also be written as:

[0131]

[0132] After traversing all nodes, a node importance score vector is formed. :

[0133]

[0134] because ,average value Also located Therefore, normalization is usually not required again. This metric represents the nodes. The average degree of participation in high-scoring triples. Under a complete third-order tensor, each node participates in the same number of triples, so using summation or averaging will not change the node ranking; averaging is easier to interpret and preserves the dimensional range.

[0135] In this embodiment, by averaging the scores of all distinct triples in which a node participates, the high-order association scores at the triple level can be converted into importance scores at the node level, while maintaining repeatability of calculation results under the same inputs and parameters.

[0136] Figure 2 The diagram illustrates the structure of a network node importance determination device based on dual-anchor point iteration according to an embodiment of the present invention.

[0137] like Figure 2 As shown, the network node importance determination device 200 based on dual-anchor point iteration in this embodiment of the invention includes: an acquisition module 210, a determination module 220, a deterministic iteration module 230, an aggregation module 240, and a node importance scoring module 250.

[0138] The acquisition module 210 is used to acquire node data and multivariate relationship data in the network to be analyzed, and generate a hypergraph based on the node data and multivariate relationship data. The hypergraph includes a set of nodes, a set of hyperedges, and a weighted correlation matrix between nodes and hyperedges.

[0139] The determination module 220 is used to calculate the node weighting degree and the hyperedge weighting degree according to the weighted association matrix, and to determine the propagation operator based on the node weighting degree and the hyperedge weighting degree, wherein the propagation operator is used to perform information propagation on the hypergraph.

[0140] The deterministic iteration module 230 is used to select one node as the starting anchor point and the other node as the attraction anchor point for any two different nodes forming an ordered anchor point pair. The starting anchor point is used as the initial node, and the propagation operator is used to perform deterministic iterative updates to obtain the score vector corresponding to each ordered anchor point pair. The score vector represents the numerical value of the contribution of each node to the anchor point relationship in the current iteration state.

[0141] The aggregation module 240 is used to extract target node components from the score vectors corresponding to the ordered anchor pairs of any three distinct nodes, and to aggregate the target node components to obtain a third-order symmetric tensor, wherein the third-order symmetric tensor represents the scoring result of the node triplet relationship.

[0142] The node importance scoring module 250 is used to determine the importance of each node in the network to be analyzed based on the third-order symmetric tensor.

[0143] Any one or more of the modules, submodules, units, and subunits according to embodiments of the present invention, or at least part of the functions of any one or more of them, can be implemented in a single module. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present invention can be implemented by being divided into multiple modules. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present invention can be at least partially implemented as hardware circuits, such as field-programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems-on-a-chip, systems-on-a-substrate, systems-on-package, application-specific integrated circuits (ASICs), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuits, or implemented in software, hardware, and firmware, or in any suitable combination of any of these three implementation methods. Alternatively, one or more of the modules, submodules, units, and subunits according to embodiments of the present invention can be at least partially implemented as computer program modules, which, when run, can perform corresponding functions.

[0144] For example, any multiple of the acquisition module 210, determination module 220, deterministic iteration module 230, aggregation module 240, and node importance scoring module 250 can be combined into one module / unit / subunit, or any one of these modules / units / subunits can be split into multiple modules / units / subunits. Alternatively, at least part of the functionality of one or more of these modules / units / subunits can be combined with at least part of the functionality of other modules / units / subunits and implemented in one module / unit / subunit. According to embodiments of the present invention, at least one of the acquisition module 210, determination module 220, deterministic iteration module 230, aggregation module 240, and node importance scoring module 250 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or any other reasonable means of integrating or packaging the circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the acquisition module 210, determination module 220, deterministic iteration module 230, aggregation module 240 and node importance scoring module 250 may be implemented at least partially as a computer program module that can perform corresponding functions when the computer program module is run.

[0145] It should be noted that the network node importance determination device based on dual-anchor point iteration in the embodiments of the present invention corresponds to the network node importance determination method based on dual-anchor point iteration in the embodiments of the present invention. For a detailed description of the network node importance determination device based on dual-anchor point iteration, please refer to the network node importance determination method based on dual-anchor point iteration, which will not be repeated here.

[0146] Figure 3 The diagram illustrates a structural block diagram of an electronic device suitable for implementing a method for determining the importance of network nodes based on dual-anchor-point iteration according to an embodiment of the present invention. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0147] like Figure 3As shown, an electronic device 300 according to an embodiment of the present invention includes a processor 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage portion 308 into a random access memory (RAM) 303. The processor 301 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 301 may also include onboard memory for caching purposes. The processor 301 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0148] The storage section 308 stores various programs and data required for the operation of the electronic device 300. The processor 301, ROM 302, and storage section 308 are interconnected via bus 304. The processor 301 executes various operations of the method flow according to embodiments of the present invention by executing programs in ROM 302 and / or storage section 308. It should be noted that the programs may also be stored in one or more memories other than ROM 302 and storage section 308. The processor 301 may also execute various operations of the method flow according to embodiments of the present invention by executing programs stored in said one or more memories.

[0149] According to an embodiment of the present invention, the electronic device 300 may further include an input / output (I / O) interface 305, which is also connected to a bus 304. The electronic device 300 may also include one or more of the following components connected to the input / output (I / O) interface 305: an input section 306 including a keyboard, mouse, etc.; an output section 307 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN card, modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the input / output (I / O) interface 305 as needed. A removable medium 311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 310 as needed so that computer programs read from it can be installed into the storage section 308 as needed.

[0150] According to embodiments of the present invention, the method flow according to embodiments of the present invention can be implemented as a computer software program. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the method shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by processor 301, it performs the functions defined in the system of the embodiments of the present invention. According to embodiments of the present invention, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0151] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of the present invention.

[0152] According to embodiments of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium. Examples include, but are not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0153] For example, according to embodiments of the present invention, a computer-readable storage medium may include the ROM 302 and / or storage portion 308 described above and / or one or more memories other than ROM 302 and storage portion 308.

[0154] Embodiments of the present invention also include a computer program product comprising a computer program containing program code for performing the methods provided in the embodiments of the present invention. When the computer program product is run on an electronic device, the program code is used to enable the electronic device to implement the methods provided in the embodiments of the present invention.

[0155] When the computer program is executed by the processor 301, it performs the functions defined in the system / apparatus of this embodiment of the invention. According to embodiments of the invention, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0156] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via communication section 309, and / or installed from removable medium 311. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0157] According to embodiments of the present invention, program code for executing the computer programs provided in the embodiments of the present invention can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0158] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions. Those skilled in the art will understand that the features described in the various embodiments of the present invention can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, the features described in the various embodiments of the present invention can be combined and / or combined in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or pairings fall within the scope of this invention.

[0159] The embodiments of the present invention have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of the invention. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of the invention, and all such substitutions and modifications should fall within the scope of the invention.

Claims

1. A method for determining the importance of network nodes based on dual-anchor point iteration, characterized in that, The method includes: Obtain node data and multivariate relation data in the network to be analyzed, and generate a hypergraph based on the node data and multivariate relation data, wherein the hypergraph includes a set of nodes, a set of hyperedges, and a weighted association matrix between nodes and hyperedges; Based on the weighted correlation matrix, the node weighted degree and the hyperedge weighted degree are calculated, and a propagation operator is determined based on the node weighted degree and the hyperedge weighted degree, wherein the propagation operator is used to perform information propagation on the hypergraph; For any two different nodes forming an ordered anchor pair, one node is selected as the starting anchor and the other node is selected as the attracting anchor. The unit vector corresponding to the starting anchor is used as the score vector of the initial node. The propagation operator is used to perform deterministic iterative updates to obtain the score vector corresponding to each ordered anchor pair. The score vector corresponding to each ordered anchor pair represents the value of the contribution of each node to the anchor relationship in the current iteration state. For any three distinct nodes, extract the target node component from the score vector corresponding to their ordered anchor pairs, and aggregate the target node components to obtain a third-order symmetric tensor, wherein the third-order symmetric tensor represents the scoring result of the node triplet relationship. Based on the aforementioned third-order symmetric tensor, the importance of each node in the network to be analyzed is determined.

2. The method according to claim 1, characterized in that, Each element value in the weighted association matrix represents the association weight between the corresponding node and the corresponding hyperedge; Wherein, when the element value is greater than zero, it indicates that the corresponding node and the corresponding hyperedge are associated; when the element value is equal to zero, it indicates that the corresponding node and the corresponding hyperedge are not associated. The method further includes: Remove nodes with a weighted degree of zero and hyperedges with a weighted degree of zero from the hypergraph.

3. The method according to claim 2, characterized in that, The step of calculating the node weighted degree and the hyperedge weighted degree based on the weighted correlation matrix includes: For any given node, the weights of all hyperedges associated with that node are summed to obtain the node's weighted degree. For any hyperedge, the weights of all nodes associated with the hyperedge are summed to obtain the weighted degree of the hyperedge.

4. The method according to claim 1, characterized in that, The step of using the unit vector corresponding to the starting anchor point as the score vector of the initial node and performing deterministic iterative updates using the propagation operator includes: performing deterministic iteration using a first update method and a second update method; The weight of the first update method is ρ, which is used to propagate node association information according to the structure of the hypergraph; The weight of the second update method is (1-ρ), wherein the second update method returns to the starting anchor point with a conditional probability of 1 / 2 and reaches the attraction anchor point with a conditional probability of 1 / 2; the weights of returning to the starting anchor point and reaching the attraction anchor point in the second update method are both (1-ρ) / 2.

5. The method according to claim 4, characterized in that, The step of using the unit vector corresponding to the starting anchor point as the score vector of the initial node and performing deterministic iterative updates using the propagation operator further includes: When the node score vector meets the preset convergence condition or reaches the preset maximum number of iterations, the iteration of the current ordered anchor pair is stopped, and the corresponding converged score vector is saved to obtain the score vector corresponding to each ordered anchor pair.

6. The method according to claim 1, characterized in that, The process of extracting target node components from the score vectors corresponding to any three distinct nodes' ordered anchor pairs, and aggregating these target node components to obtain a third-order symmetric tensor, includes: For any three distinct nodes, extract the target node component from the score vector corresponding to the six ordered anchor pairs, and perform average aggregation on the target node component to obtain the original triplet score. The original scores of the triples are subjected to min-max normalization to obtain the third-order symmetric tensor.

7. The method according to claim 1, characterized in that, The determination of the importance of each node in the network to be analyzed based on the third-order symmetric tensor includes: For any node, obtain the triplet scores corresponding to all distinct triplets containing the current node from the third-order symmetric tensor, and average the triplet scores to obtain the importance score of the current node. Traverse all nodes in the node set to obtain the importance score of all nodes.

8. A network node importance determination device based on dual-anchor point iteration, characterized in that, The device includes: The acquisition module is used to acquire node data and multivariate relationship data in the network to be analyzed, and generate a hypergraph based on the node data and the multivariate relationship data, wherein the hypergraph includes a set of nodes, a set of hyperedges, and a weighted association matrix between nodes and hyperedges; The determination module is used to calculate the node weighting degree and the hyperedge weighting degree according to the weighted association matrix, and determine the propagation operator based on the node weighting degree and the hyperedge weighting degree, wherein the propagation operator is used to perform information propagation on the hypergraph; The deterministic iteration module is used to select one node as the starting anchor point and the other node as the attraction anchor point for any two different nodes forming an ordered anchor point pair. The unit vector corresponding to the starting anchor point is used as the score vector of the initial node. The propagation operator is used to perform deterministic iterative updates to obtain the score vector corresponding to each ordered anchor point pair. The score vector corresponding to each ordered anchor point pair represents the numerical value of the contribution of each node to the anchor point relationship in the current iteration state. The aggregation module is used to extract target node components from the score vectors corresponding to the ordered anchor pairs of any three distinct nodes, and to aggregate the target node components to obtain a third-order symmetric tensor, wherein the third-order symmetric tensor represents the scoring result of the node triplet relationship. The node importance scoring module is used to determine the importance of each node in the network to be analyzed based on the third-order symmetric tensor.

9. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having executable instructions stored thereon, characterized in that, When executed by the processor, this instruction causes the processor to perform the method according to any one of claims 1 to 7.