CN111669288B - Directional network link prediction method and device based on directional heterogeneous neighbor - Google Patents
Directional network link prediction method and device based on directional heterogeneous neighbor Download PDFInfo
- Publication number
- CN111669288B CN111669288B CN202010449774.8A CN202010449774A CN111669288B CN 111669288 B CN111669288 B CN 111669288B CN 202010449774 A CN202010449774 A CN 202010449774A CN 111669288 B CN111669288 B CN 111669288B
- Authority
- CN
- China
- Prior art keywords
- network
- neighbors
- directed
- nodes
- calculating
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/147—Network analysis or design for predicting network behaviour
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention belongs to the technical field of complex network link prediction, and particularly relates to a directed network link prediction method and a directed network link prediction device based on directed heterogeneous neighbors, wherein the method comprises the steps of inputting a data set to construct a network model; counting the edge connection probability of the directed triplet isomers in the network; selecting a pair of unconnected candidate nodes, and calculating the number of common neighbors existing between the two nodes; calculating the index similarity value of the two-node directed heterogeneous neighbors; calculating corresponding index similarity scores for all unconnected node pairs; and sorting the indexes in a descending order according to the index similarity values, and taking the node pairs corresponding to the first scores as prediction continuous edges. The invention utilizes CN index thought, introduces the edge connecting direction of the local structure, and integrates and calculates the score of each heterogeneous neighbor, so that the method has better prediction effect in the directed network.
Description
Technical Field
The invention belongs to the technical field of complex network link prediction, and particularly relates to a directed network link prediction method and device based on directed heterogeneous neighbors, which are suitable for the problem of continuous edge prediction in various types of real directed networks, and are particularly used for the fields of friend recommendation, traffic planning and the like.
Background
The complex network is an abstract tool for mining the intrinsic mechanism of a real network in real life, and the important direction link prediction of the complex network has important application in the fields of recommendation systems, traffic planning, network reconstruction analysis and the like by predicting unknown connection, future connection or finding wrong connection among nodes by using the known information of the network.
In the real world, factors such as technology, manpower and privacy attributes are limited, most of observed networks lack a lot of information, namely network data are usually lacked, in the last decades, numerous excellent prediction indexes are provided for undirected networks (as shown in fig. 2), indexes based on similarity are excellent in effect due to concise ideas, and various prediction schemes can be developed and can be classified into prediction methods of local similarity and global similarity. The similarity thought considers that the higher the similarity degree of two nodes in the network is, the higher the probability of generating connecting edges between the nodes is. Representative similarity indicators include CN (Common Neighbors), AA (adaptive-Adar), RA (Resource Allocation), LP (Local Path), katz, and the like. The schemes utilize information such as common neighbors among nodes, node degrees, node paths and the like to predict, and have excellent prediction effect in undirected networks.
However, real network nodes are not simply connected without direction, and the direction has more important meanings, such as the one-way attention relationship of social networks, the food chain relationship of natural creatures, the reference relationship between papers, and the like, and the connection direction has an irreversible characteristic. In the existing scheme, a directed network is regarded as an undirected network, and undirected indexes are further applied to prediction, so that the prediction effect is greatly reduced.
Disclosure of Invention
The invention provides a directed network link prediction method and device based on directed heterogeneous neighbors, aiming at the problem that the existing directed network has poor effect due to the fact that network edge connecting direction information is ignored based on local similarity indexes.
In order to solve the technical problems, the invention adopts the following technical scheme:
the invention provides a directed network link prediction method based on directed heterogeneous neighbors, which is characterized by comprising the following steps:
step 1, inputting a data set to construct a network model;
step 2, counting the edge connection probability of the directed triplet isomers in the network;
step 3, selecting a pair of unconnected candidate nodes, and calculating the number of common neighbors existing between the two nodes;
step 4, according to the step 2 and the step 3, calculating the index similarity value of the two-node directed heterogeneous neighbors;
and 6, sorting in a descending order according to the index similarity scores, and taking node pairs corresponding to the first scores as prediction continuous edges.
Further, step 1 constructs a network model G (V, E), and represents the network connection state by an adjacency matrix A, a ij And =1 indicates that the network has a one-way side of i → j.
Further, edge connection probability p of nine directed triplex isomers in the network is counted through the adjacency matrix A k The process is as follows:
step 201, the first-order neighbors of the nodes in the network are calculated in sequence as follows, and the reciprocal neighbors are gamma bila (x)＝A*A T The out-degree neighbor is gamma out (x)＝A-(A*A T ) The approach neighbors are gamma in (x)＝A T -(A*A T )；
Step 202, the matrix operation of nine different heterogeneous second-order neighbors is S k ＝Γ i (x)*Γ j (x)，i,j∈{out,in,bila}；
Step 203, calculating the edge connection probability of the nine directional triple isomers
k＝1,2,…,9。
Further, a pair of candidate nodes x and y which are not connected is selected, and the number of common neighbors existing between the two nodes is calculated
i,j∈{out,in,bila}。
Further, step 4, calculating similarity scores of the x and y directed heterogeneous neighbor indexes of the two nodes
Further, in step 6, node pairs corresponding to the first L scores are taken as predicted connected edges, L is a preset positive integer, L = N (N-1) -M, N is the total number of network nodes, M is the number of known connected edges of the network, and N (N-1) -M is the number of unconnected edges of the network.
The invention also provides a device for predicting the directed network link based on the directed heterogeneous neighbor, which comprises the following steps:
the network model building module is used for inputting a data set to build a network model;
the edge connecting probability calculation module is used for counting the edge connecting probability of the directed triple isomers in the network;
the common neighbor number calculation module is used for selecting a pair of unconnected candidate nodes and calculating the number of common neighbors existing between the two nodes;
the first index similarity value calculation module is used for calculating the index similarity values of the two-node directed heterogeneous neighbors;
the second index similarity score calculation module is used for calculating corresponding index similarity scores for all unconnected node pairs;
and the prediction continuous edge selection module is used for sorting in a descending order according to the index similarity values, and taking node pairs corresponding to the first plurality of values as prediction continuous edges.
Compared with the prior art, the invention has the following advantages:
the method is suitable for various types of directed networks, local edge-connected heterogeneous differences of the directed networks are further mined on the basis of the undirected common neighbors by calculating probability scores of possibly established connections among unconnected nodes and sequencing the unconnected nodes in a descending order, and the effect of promoting edge connection generation by different structures is expressed quantitatively through edge-connected probability, so that an excellent prediction effect is shown in the directed networks.
Drawings
In order to more clearly illustrate the embodiments or technical solutions of the present invention, the drawings used in the embodiments or technical solutions in the prior art are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is a flowchart of a method for predicting a directed network link based on a directed heterogeneous neighbor according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of the mutual neighbors of nodes x and y in a undirected network;
FIG. 3 is a schematic diagram of a heterogeneous neighbor of a directed network according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a directional network link prediction apparatus based on a directional heterogeneous neighbor according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer and more complete, the technical solutions in the embodiments of the present invention will be described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention, and based on the embodiments of the present invention, all other embodiments obtained by a person of ordinary skill in the art without creative efforts belong to the scope of the present invention.
As shown in fig. 1, the method for predicting a directed network link based on a directed heterogeneous neighbor according to this embodiment includes the following steps:
step S11, inputting the data set to construct a network model G (V, E), and representing the network connection edge state by an adjacency matrix A, a ij =1 indicates that the network has a one-way side of i → j;
step S12, counting the edge connecting probability p of the nine directional triple isomers in the network through the adjacency matrix A k The specific process is as follows:
step S121, sequentially calculating first-order neighbors of nodes in the network as follows, wherein the reciprocal neighbors are gamma bila (x)＝A*A T The emergence neighbor is gamma out (x)＝A-(A*A T ) The approach neighbors are gamma in (x)＝A T -(A*A T )；
Step S122, as shown in FIG. 3, the matrix operation of the nine different directed heterogeneous second-order neighbors is S k ＝Γ i (x)*Γ j (x) I, j ∈ { out, in, bila }; for example, the matrix operation for structure S1 in fig. 3 is described as S1= Γ bila (x)*Γ bila (x)＝(A*A T )*(A*A T ) The matrix operation for structure S2 is described as S2= Γ in (x)*Γ bila (x)＝(A T -(A*A T ))*(A*A T )……。
Step S123, calculating the edge connecting probability of S1-S9
k＝1,2,…,9。
S13, selecting a pair of unconnected candidate nodes x and y, and calculating the number of common neighbors existing between the two nodes
i,j∈{out,in,bila}。
Step S14, according to the step S12 and the step S13, calculating the similarity score of the index of the directed heterogeneous neighbor of the node x and the node y
And S15, repeating the step S13 and the step S14 for all unconnected node pairs, and calculating corresponding index similarity scores.
And S16, sorting the indexes in a descending order according to the index similarity values, and taking node pairs corresponding to the first L values as predicted connected edges according to actual requirements, wherein L is a preset positive integer, L = N (N-1) -M usually, N is the total number of network nodes, M is the known connected edge number of the network, and N (N-1) -M is the number of unconnected edges of the network.
As shown in fig. 4, this embodiment further provides a device for predicting a directed network link based on a directed heterogeneous neighbor, including:
a network model building module 41, configured to input a data set to build a network model;
the edge connecting probability calculating module 42 is used for counting the edge connecting probability of the directional triple isomers in the network;
a common neighbor number calculation module 43, configured to select a pair of unconnected candidate nodes and calculate the number of common neighbors existing between the two nodes;
the first index similarity value calculation module 44 is configured to calculate an index similarity value of the two-node directed heterogeneous neighbor;
a second index similarity score calculation module 45, configured to calculate corresponding index similarity scores for all unconnected node pairs;
and the prediction continuous edge selecting module 46 is used for sorting according to the index similarity scores in a descending order, and taking node pairs corresponding to the first scores as prediction continuous edges.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Those of ordinary skill in the art will understand that: all or part of the steps of implementing the method embodiments may be implemented by hardware related to program instructions, and the program may be stored in a computer-readable storage medium, and when executed, executes the steps including the method embodiments; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.
Finally, it should be noted that: the above description is only a preferred embodiment of the present invention, and is only used to illustrate the technical solutions of the present invention, and not to limit the protection scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.
Claims (2)
1. A directed network link prediction method based on directed heterogeneous neighbors is characterized by comprising the following steps:
step 1, inputting a data set to construct a network model G (V, E), and representing a network connection edge state by an adjacency matrix A, a ij =1 indicates that the network has a unidirectional connection edge of i → j;
step 2, counting the edge connection probability p of the nine directed triple isomers in the network through the adjacency matrix A k The process is as follows:
step 201, the first-order neighbors of the nodes in the network are calculated in sequence as follows, and the reciprocal neighbors are gamma bila (x)＝A*A T The emergence neighbor is gamma out (x)＝A-(A*A T ) The approach neighbors are gamma in (x)＝A T -(A*A T )；
Step 202, the matrix operation of the nine different directed heterogeneous second-order neighbors is S k ＝Γ i (x)*Γ j (x)，i,j∈{out,in,bila}；
Step 203, calculating the edge connecting probability of the nine directional triple isomers
k＝1,2,…,9；
Step 3, selecting a pair of candidate nodes x and y which are not connected, and calculating the number of common neighbors existing between the two nodes
i,j∈{out,in,bila}；
Step 4, according to the step 2 and the step 3, the existence of the two nodes x and y is calculatedSimilarity score to heterogeneous neighbor index
Step 5, repeating the step 3 and the step 4 for all unconnected node pairs, and calculating corresponding index similarity scores;
and 6, sorting the indexes in a descending order according to the index similarity scores, taking node pairs corresponding to the first L scores as prediction connected edges, wherein L is a preset positive integer, L = N (N-1) -M, N is the total number of network nodes, M is the known connected edge number of the network, and N (N-1) -M is the number of unconnected edges of the network.
2. A directed network link prediction device based on directed heterogeneous neighbors, comprising:
a network model construction module for inputting data set to construct network model G (V, E) and using adjacency matrix A to represent network connection state, a ij =1 indicates that the network has a one-way side of i → j;
a connecting edge probability calculation module for counting the connecting edge probability p of the nine directional triple isomers in the network through the adjacency matrix A k The process is as follows:
step 201, the first-order neighbors of the nodes in the network are calculated in sequence as follows, and the reciprocal neighbors are gamma bila (x)＝A*A T The emergence neighbor is gamma out (x)＝A-(A*A T ) The approach neighbors are gamma in (x)＝A T -(A*A T )；
Step 202, the matrix operation of the nine different directed heterogeneous second-order neighbors is S k ＝Γ i (x)*Γ j (x)，i,j∈{out,in,bila}；
Step 203, calculating the edge connecting probability of the nine directional triple isomers
k＝1,2,…,9；
A common neighbor number calculation module for selecting a pair of unconnected candidate nodes x and y and calculating the number of common neighbors existing between the two nodes
i,j∈{out,in,bila}；
A first index similarity value calculation module for calculating the similarity values of the x and y directed heterogeneous neighbor indexes of the two nodes
The second index similarity value calculation module is used for calculating corresponding index similarity values of all unconnected node pairs;
and the prediction connected edge selection module is used for sorting the nodes in a descending order according to the index similarity scores, taking the node pairs corresponding to the first L scores as prediction connected edges, wherein L is a preset positive integer, L = N (N-1) -M, N is the total number of network nodes, M is the number of known connected edges of the network, and N (N-1) -M is the number of unconnected edges of the network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010449774.8A CN111669288B (en) | 2020-05-25 | 2020-05-25 | Directional network link prediction method and device based on directional heterogeneous neighbor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010449774.8A CN111669288B (en) | 2020-05-25 | 2020-05-25 | Directional network link prediction method and device based on directional heterogeneous neighbor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111669288A CN111669288A (en) | 2020-09-15 |
| CN111669288B true CN111669288B (en) | 2023-02-14 |
Family
ID=72384479
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010449774.8A Active CN111669288B (en) | 2020-05-25 | 2020-05-25 | Directional network link prediction method and device based on directional heterogeneous neighbor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111669288B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112700056B (en) * | 2021-01-06 | 2023-09-15 | 中国互联网络信息中心 | Complex network link prediction method, device, electronic equipment and medium |
| CN112966155B (en) * | 2021-03-23 | 2023-03-21 | 西安电子科技大学 | Link prediction method based on path correlation |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107018027A (en) * | 2017-05-23 | 2017-08-04 | 浙江工业大学 | Link prediction method based on Bayesian estimation and common neighbor node degree |
| CN110493045A (en) * | 2019-08-19 | 2019-11-22 | 大连民族大学 | A kind of directed networks link prediction method merging multimode body information |
| CN110569885A (en) * | 2019-08-19 | 2019-12-13 | 大连民族大学 | multi-order motif directed network link prediction method based on naive Bayes |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8650139B2 (en) * | 2011-07-01 | 2014-02-11 | Google Inc. | Predicting user navigation events |
-
2020
- 2020-05-25 CN CN202010449774.8A patent/CN111669288B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107018027A (en) * | 2017-05-23 | 2017-08-04 | 浙江工业大学 | Link prediction method based on Bayesian estimation and common neighbor node degree |
| CN110493045A (en) * | 2019-08-19 | 2019-11-22 | 大连民族大学 | A kind of directed networks link prediction method merging multimode body information |
| CN110569885A (en) * | 2019-08-19 | 2019-12-13 | 大连民族大学 | multi-order motif directed network link prediction method based on naive Bayes |
Non-Patent Citations (2)
| Title |
|---|
| 一种有向复杂网络生成模型的建立方法;刘大伟等;《小型微型计算机系统》;20181031;第39卷(第10期);第2197-2201页 * |
| 有向网络链路预测研究进展;赵学磊等;《信息工程大学学报》;20191215(第06期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111669288A (en) | 2020-09-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102413029B (en) | Method for partitioning communities in complex dynamic network by virtue of multi-objective local search based on decomposition | |
| Lee et al. | Can Twitter data be used to validate travel demand models | |
| Nie et al. | Network traffic prediction based on deep belief network and spatiotemporal compressive sensing in wireless mesh backbone networks | |
| CN111669288B (en) | Directional network link prediction method and device based on directional heterogeneous neighbor | |
| Guo et al. | Network intelligent control and traffic optimization based on SDN and artificial intelligence | |
| CN111325340B (en) | Information network relation prediction method and system | |
| Liu et al. | An oriented spanning tree based genetic algorithm for multi-criteria shortest path problems | |
| CN102262702A (en) | Decision-making method for maintaining middle and small span concrete bridges | |
| CN114638021B (en) | Security evaluation method for lightweight block chain system of Internet of things | |
| CN111988786B (en) | Sensor network covering method and system based on high-dimensional multi-target decomposition algorithm | |
| Li et al. | Network topology optimization via deep reinforcement learning | |
| CN115456093A (en) | High-performance graph clustering method based on attention-graph neural network | |
| CN117315331A (en) | Dynamic graph anomaly detection method and system based on GNN and LSTM | |
| Almarzoqi et al. | Re-learning exp3 multi-armed bandit algorithm for enhancing the massive iot-lorawan network performance | |
| CN115168654A (en) | Multi-attribute decision-based super-network important node identification method | |
| Al-Saadi et al. | SDN-based routing framework for elephant and mice flows using unsupervised machine learning | |
| CN106341258A (en) | Method for predicting unknown network connection edges based on second-order local community and seed node structure information | |
| Meena et al. | Overlapping community detection in social network using disjoint community detection | |
| CN115965795A (en) | Deep darknet group discovery method based on network representation learning | |
| CN116070385A (en) | Automatic risk identification method and system for overseas mineral resource supply chain | |
| CN113283243B (en) | Entity and relationship combined extraction method | |
| CN114596030A (en) | Express delivery network operation prediction model based on data mining | |
| CN114912927A (en) | Block chain anti-fraud analysis method and system | |
| Zhang et al. | Near-optimal distributed detection in balanced binary relay trees | |
| Kalam et al. | 5g traffic forecasting using federated learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |