KR20210121773A - Apparatus and method for detecting community in large scale network - Google Patents

Abstract

Disclosed are device and method for detecting a community. The method for detecting the community according to one disclosed embodiment comprises: a step of selecting one or more critical nodes in a network comprising a plurality of nodes; a step of calculating node similarity between the critical node and the remaining node; a step of generating a classification result of classifying a plurality of nodes into one or more communities based on the critical node and based on the node similarity; a step of calculating modularity for the classification result; a step of repeated-performing the above steps until a preset condition is satisfied; and a step of determining, when the preset condition is satisfied, the classification result with the greatest modularity among the plurality of classification results generated through repeated performance as a final classification result. Therefore, the present invention is capable of improving a performance of detecting the community.

Description

COMMUNITY DETECTION DEVICES AND METHOD {APPARATUS AND METHOD FOR DETECTING COMMUNITY IN LARGE SCALE NETWORK}

The disclosed embodiments relate to techniques for detecting communities.

As society becomes more complex, not only human-to-human connections but also devices-to-device connections are becoming very complex, and the concept of a large-scale network, which means a connection structure between these complex entities, is emerging.

In order to understand the internal structure of such a large-scale network and to obtain effective information, it is very important to understand how the community is structured according to the connection relationships of entities in the network.

The conventional community detection method uses a graph in which an entity is expressed as a node and a connection between nodes is expressed as an edge, but there is a limit in that the detection result is greatly affected by the location of the initial node.

In order to overcome this problem, recent attempts have been made to find the most influential node, but since the most influential node is found after the community is detected, prior knowledge for community detection is required, and macroscopic connection patterns are considered at the small sub-network level. There is a downside to not being able to.

Republic of Korea Patent Publication No. 10-1977231 (2019.05.03.)

The disclosed embodiments are intended to provide a means for effectively detecting a community in a large network.

A community detection method according to an embodiment includes selecting one or more important nodes in a network including a plurality of nodes, calculating a node similarity between each of the one or more important nodes and the remaining nodes, the node Generating a classification result obtained by classifying the plurality of nodes into one or more communities based on each of the one or more important nodes based on the similarity, calculating modularity for the classification result, and satisfying a preset condition Repeatedly performing the selecting step, calculating the node similarity, generating the classification result, and calculating the modularity until the preset condition is satisfied. and determining a classification result having the greatest modularity among the plurality of classification results as a final classification result.

The selecting may include calculating an eigenvector centrality value of each of the plurality of nodes and determining the one or more important nodes among the plurality of nodes based on the eigenvector centrality value. In addition, the one or more important nodes may be further selected by a preset number each time the selecting step is repeatedly performed.

The eigenvector centrality value may be calculated based on an adjacency matrix indicating whether each of the plurality of nodes is connected and an eigenvalue of the adjacency matrix.

The calculating of the node similarity may include calculating the node similarity based on whether there is a connection between the one or more important nodes and the remaining nodes and a Jaccard similarity between the one or more important nodes and the remaining nodes.

The generating may include classifying the plurality of nodes into the one or more communities based on the number of important nodes directly connected through an edge.

The generating may include, in the case of a first type of node that is not directly connected to the important node among the plurality of nodes, the node similarity with the first type of node among the nodes directly connected to the first type of node is The largest node and the first type node may be classified into the same community, and in the case of a second type node directly connected to only one important node among the plurality of nodes, the directly connected one important node and the second type node nodes of a type may be classified into the same community, and in the case of a node of a third type directly connected to two or more important nodes among the plurality of nodes, the node with the third type of nodes among the two or more important nodes An important node having the greatest similarity and the third type node may be classified into the same community.

In the calculating of the modularity, the modularity of the classification result may be calculated using Equations 1 to 3 below.

[Equation 1]

[Equation 2]

[Equation 3]

In this case, Q is modularity, m is the total number of edges in the network, and Ni and Nj are arbitrary nodes in the network, respectively.

The preset condition may be that all of the plurality of nodes are selected as the important node.

A community detection apparatus according to an embodiment includes one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs The programs include instructions for selecting one or more important nodes in a network including a plurality of nodes, instructions for calculating a node similarity between each of the one or more important nodes and the remaining nodes, and based on the node similarity, the A command for generating a classification result of classifying the plurality of nodes into one or more communities based on each of one or more important nodes, a command for calculating modularity for the classification result, and a predetermined condition When a command for repeatedly performing the steps of selecting, calculating the node similarity, generating the classification result, and calculating the modularity, and the preset condition are satisfied, the plurality of generated through repetition and a command for determining a classification result having the greatest modularity among the classification results of , as a final classification result.

The instruction for selecting includes an instruction for calculating an eigenvector centrality value of each of the plurality of nodes and an instruction for determining the at least one important node among the plurality of nodes based on the eigenvector centrality value may include, and the one or more important nodes may be further selected by a preset number each time the selecting step is repeatedly performed.

The eigenvector centrality value may be calculated based on an adjacency matrix indicating whether each of the plurality of nodes is connected and an eigenvalue of the adjacency matrix.

The command for calculating the node similarity may calculate the node similarity based on whether there is a connection between the one or more important nodes and the remaining nodes and a Jaccard similarity between the one or more important nodes and the remaining nodes.

The command for generating may classify the plurality of nodes into the one or more communities based on the number of important nodes directly connected through an edge.

The generating command may include, in the case of a first type of node not directly connected to the important node among the plurality of nodes, the degree of node similarity with the first type of node among nodes directly connected to the first type of node. In the case of a node of the second type directly connected to only one important node among the plurality of nodes, the node having the maximum , and the first type node may be classified into the same community, and the directly connected one important node and the second type node Two types of nodes may be classified into the same community, and in the case of a third type node directly connected to two or more important nodes among the plurality of nodes, An important node having the maximum node similarity and the third type node may be classified into the same community.

The command for calculating the modularity may calculate the modularity of the classification result using Equations 1 to 3 below.

[Equation 1]

[Equation 2]

[Equation 3]

In this case, Q is modularity, m is the total number of edges in the network, and Ni and Nj are arbitrary nodes in the network, respectively.

The preset condition may be that all of the plurality of nodes are selected as the important node.

According to the disclosed embodiments, since a community is detected after selecting an important node in the network, community detection can be performed without other information only with centrality and similarity between nodes, thereby improving community detection performance.

Also, according to the disclosed embodiments, by clearly identifying the community structure of a part (local) around an important node, it can be effectively applied to a large-scale network in which it is difficult to grasp the entire community structure.

1 is a block diagram illustrating and describing a computing environment including a computing device suitable for use in example embodiments;
2 is a flowchart illustrating a community detection method according to an embodiment;
3 is a flowchart for explaining step 202 in more detail according to an embodiment;
4 is a diagram for explaining a process of generating a classification result of a community according to an embodiment;
5 is an exemplary view of the final classification result of the community

Hereinafter, specific embodiments will be described with reference to the drawings. The following detailed description is provided to provide a comprehensive understanding of the methods, apparatus, and/or systems described herein. However, this is merely an example and the disclosed embodiments are not limited thereto.

In describing the embodiments, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the disclosed embodiments, the detailed description thereof will be omitted. And, the terms to be described later are terms defined in consideration of functions in the disclosed embodiments, which may vary according to intentions or customs of users and operators. Therefore, the definition should be made based on the content throughout this specification. The terminology used in the detailed description is for the purpose of describing the embodiments only, and should in no way be limiting. Unless explicitly used otherwise, expressions in the singular include the meaning of the plural. In this description, expressions such as “comprising” or “comprising” are intended to indicate certain features, numbers, steps, acts, elements, some or a combination thereof, one or more other than those described. It should not be construed to exclude the presence or possibility of other features, numbers, steps, acts, elements, or any part or combination thereof.

1 is a block diagram illustrating and describing a computing environment 10 including a computing device suitable for use in example embodiments. In the illustrated embodiment, each component may have different functions and capabilities other than those described below, and may include additional components in addition to those described below.

The illustrated computing environment 10 includes a computing device 12 . In one embodiment, computing device 12 may be a community detection device in accordance with disclosed embodiments.

Computing device 12 includes at least one processor 14 , computer readable storage medium 16 , and communication bus 18 . The processor 14 may cause the computing device 12 to operate in accordance with the exemplary embodiments discussed above. For example, the processor 14 may execute one or more programs stored in the computer-readable storage medium 16 . The one or more programs may include one or more computer-executable instructions that, when executed by the processor 14, configure the computing device 12 to perform operations in accordance with the exemplary embodiment. can be

Computer-readable storage medium 16 is configured to store computer-executable instructions or program code, program data, and/or other suitable form of information. The program 20 stored in the computer readable storage medium 16 includes a set of instructions executable by the processor 14 . In one embodiment, computer-readable storage medium 16 includes memory (volatile memory, such as random access memory, non-volatile memory, or a suitable combination thereof), one or more magnetic disk storage devices, optical disk storage devices, flash It may be memory devices, other forms of storage medium accessed by computing device 12 and capable of storing desired information, or a suitable combination thereof.

Communication bus 18 interconnects various other components of computing device 12 , including processor 14 and computer readable storage medium 16 .

Computing device 12 may also include one or more input/output interfaces 22 and one or more network communication interfaces 26 that provide interfaces for one or more input/output devices 24 . The input/output interface 22 and the network communication interface 26 are coupled to the communication bus 18 . Input/output device 24 may be coupled to other components of computing device 12 via input/output interface 22 . Exemplary input/output device 24 may include a pointing device (such as a mouse or trackpad), a keyboard, a touch input device (such as a touchpad or touchscreen), a voice or sound input device, various types of sensor devices, and/or imaging devices. input devices and/or output devices such as display devices, printers, speakers and/or network cards. The exemplary input/output device 24 may be included in the computing device 12 as a component constituting the computing device 12 , and may be connected to the computing device 12 as a separate device distinct from the computing device 12 . may be

2 is a flowchart 200 for explaining a method for detecting a community according to an embodiment. The method illustrated in FIG. 2 may be performed, for example, by the computing device 12 described above.

Hereinafter, the term 'community' refers to a group in which objects similar to each other are classified, and in this case, the term 'object' may refer to a person, a creature, a thing, or an abstract object such as a word or an idea. In addition, in the disclosed embodiment, the 'network' may mean a set in which all of these 'objects' are connected to each other by an arbitrary standard, and in the network, an object is a 'node', and the connection between objects is It can be expressed as an 'edge' connecting nodes.

Accordingly, the following 'community detection method' provides a method for efficiently deriving a result of a plurality of nodes in a network classified into one or more communities, respectively. At this time, it is assumed that each node in the network cannot belong to two or more communities at the same time.

Referring to FIG. 2 , in step 202 , the computing device 12 selects one or more important nodes from a network including a plurality of nodes.

In the disclosed embodiments, an 'important node' is a node determined to be important compared to other nodes in consideration of the number of connected nodes and the importance of each edge in the network, and each node when classifying the network into one or more communities It is a criterion for determining which community they will be classified into.

Then, in step 204, the computing device 12 calculates a node similarity between each of the one or more important nodes and the remaining nodes.

In an embodiment, the computing device 12 may calculate the node similarity based on whether or not there is a connection between one or more important nodes and the remaining nodes and the Jaccard similarity.

That is, the computing device 12 may calculate the node similarity using Equation 1 below.

[Equation 1]

는 노드 i와 직접 연결된 노드의 집합,

는 노드 j와 직접 연결된 노드의 집합,

는 노드 i 또는 노드 j와 직접 연결된 노드의 수,

는 노드 i 및 노드 j와 직접 연결된 노드의 수를 나타낸다.In this case, Similarity(i, j) is the node similarity between node i and node j, A _ij is the i-row j-column component of the adjacency matrix,

is the set of nodes directly connected to node i,

is the set of nodes directly connected to node j,

is the number of nodes directly connected to node i or node j,

denotes the number of nodes directly connected to node i and node j.

Thereafter, in step 206 , the computing device 12 classifies the plurality of nodes into one or more communities based on each of one or more important nodes based on the calculated node similarity, thereby generating a classification result.

In an embodiment, the computing device 12 may classify the plurality of nodes into one or more communities based on the number of important nodes directly connected to each of the plurality of nodes through an edge.

Thereafter, in step 208, the computing device 12 calculates modularity for the generated classification result.

In the disclosed embodiments, 'modularity' refers to a quantified value of how many more edges are included in the community under the current classification result, compared to the edge in the network in which each node is randomly connected. Accordingly, it can be estimated that nodes within each community are well connected under the current classification result as compared to a randomly connected network as the modularity value is large.

In an embodiment, the computing device 12 may calculate the modularity of the classification result using Equations 2 to 4 below.

[Equation 2]

[Equation 3]

[Equation 4]

In this case, Q is modularity, m is the total number of edges in the network, and Ni and Nj are arbitrary nodes in the network, respectively.

Thereafter, in step 210, the computing device 12 determines whether a preset condition is satisfied based on the current community classification result. If it is determined that the preset condition is not satisfied, the computing device 12 repeats steps 202 to 208. In this case, whenever step 202 is repeatedly performed, the computing device 12 may further select an important node by a preset number. For example, the computing device 12 may select one more important node each time step 202 is repeatedly performed.

In an embodiment, the preset condition may be that all of a plurality of nodes included in the network are selected as important nodes.

Thereafter, when it is determined that the preset condition is satisfied, in step 212 , the computing device 12 determines a classification result having the greatest modularity among a plurality of classification results generated through the iterative execution as a final classification result.

Specifically, since the computing device 12 calculates modularity for each of the plurality of community classification results by repeating step 208, the final classification of the community classification result corresponding to the maximum modularity value among the plurality of calculated modularity values determined by the result.

In the illustrated flowchart, the method is described by dividing the method into a plurality of steps, but at least some of the steps are performed in a reversed order, are performed together in combination with other steps, are omitted, are performed separately, or are not shown. One or more steps may be added and performed.

3 is a flowchart 300 for explaining a method for setting an important node according to an embodiment.

The method illustrated in FIG. 3 may be performed, for example, in step 202 illustrated in FIG. 2 .

Referring to FIG. 3 , in step 302 , the computing device 12 may calculate an eigenvector centrality value for each of a plurality of nodes included in the network.

In an embodiment, the eigenvector centrality value may be calculated based on an adjacency matrix indicating whether each of a plurality of nodes is connected and an eigenvalue of the adjacency matrix.

Specifically, the adjacency matrix refers to a matrix having 1 as an element value when there is an edge connecting between nodes, and 0 as an element value when there is no edge. For example, if the number of nodes included in the network is n, the adjacency matrix is an nxn square matrix, the main diagonal components are all 1, and the element value of row i and column j indicates whether an edge exists between the i-th node and the j-th node. Depending on whether or not it has a value of 0 or 1.

Accordingly, the computing device 12 may calculate the eigenvector centrality value of each of the plurality of nodes using Equation 5 below.

[Equation 5]

는 인접 행렬의 고유값 중 최대값을 나타낸다.In this case, A _ij is the i-row j-column component of the adjacency matrix, x _i is the eigenvector centrality value of the node i, x _j is the eigenvector centrality value of the node j,

represents the maximum value among the eigenvalues of the adjacency matrix.

Thereafter, in operation 304 , the computing device 12 may determine one or more important nodes among the plurality of nodes based on the calculated eigenvector centrality values for each of the plurality of nodes.

Specifically, the computing device 12 determines the important nodes in the order of increasing the eigenvector centrality value, but in step 210 to be described below, when the computing device 12 repeatedly performs step 202, next to the previously determined important node A predetermined number of nodes having a large eigenvector centrality value may be further selected and included in the important node.

In the illustrated flowchart, the method is described by dividing the method into a plurality of steps, but at least some of the steps are performed in a reversed order, are performed together in combination with other steps, are omitted, are performed separately, or are not shown. One or more steps may be added and performed.

4 is an exemplary diagram 400 for explaining an example of generating a community classification result according to an embodiment, and FIG. 5 is an exemplary diagram 500 illustrating an example of a final community classification result.

Referring to FIG. 4 , the computing device 12 distinguishes cases where the number of important nodes directly connected to each of a plurality of nodes through an edge is 0, 1, and 2 or more, and a plurality of The method of classifying nodes in one or more communities can be different. The classification method for each case is as follows.

(1) In the case of a first type of node (A) that is not directly connected to an important node among a plurality of nodes, the computing device 12 is a first type of node ( The node having the maximum node similarity with A) and the first type node (A) are classified into the same community.

(2) In the case of the second type of node (B) directly connected to only one critical node (eg, critical node 1 in FIG. 4 ) among the plurality of nodes, the computing device 12 is directly connected to the critical node and the second type of node (B). Classify the two types of nodes (B) into the same community.

(3) In the case of a third type of node (C) directly connected to two or more important nodes among a plurality of nodes (eg, important node 1 and important node 2 in FIG. 4 ), the two or more important nodes directly connected Among them, the important node having the highest node similarity with the third type node (C) and the third type node (C) are classified into the same community.

Referring to FIG. 5 , it can be seen that a plurality of nodes included in the network are exemplarily classified into six communities through the process described with reference to FIG. 4 . In this way, even when it is difficult to classify each node because there are so many objects (nodes) in the network, each node can be quickly classified among similar nodes by initially selecting important nodes and calculating node similarity and modularity.

Meanwhile, an embodiment of the present invention may include a program for performing the methods described in this specification on a computer, and a computer-readable recording medium including the program. The computer-readable recording medium may include program instructions, local data files, local data structures, etc. alone or in combination. The medium may be specially designed and configured for the present invention, or may be commonly used in the field of computer software. Examples of computer-readable recording media include hard disks, magnetic media such as floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and program instructions specially configured to store and execute program instructions such as ROMs, RAMs, flash memories, and the like. Hardware devices are included. Examples of the program may include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like.

Although representative embodiments of the present invention have been described in detail above, those of ordinary skill in the art to which the present invention pertains will understand that various modifications are possible without departing from the scope of the present invention with respect to the above-described embodiments. . Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined by the following claims as well as the claims and equivalents.

10: Computing Environment
12: computing device
14: Processor
16: computer readable storage medium
18: communication bus
20: Program
22: input/output interface
24: input/output device
26: network communication interface

Claims (16)

selecting one or more important nodes in a network including a plurality of nodes;
calculating a node similarity between each of the one or more important nodes and the remaining nodes;
generating a classification result of classifying the plurality of nodes into one or more communities based on each of the one or more important nodes based on the node similarity;
calculating modularity for the classification result;
repeatedly performing the selecting step, calculating the node similarity, generating the classification result, and calculating the modularity until a preset condition is satisfied; and
and determining, as a final classification result, a classification result having the greatest modularity among a plurality of classification results generated through the iterative execution when the preset condition is satisfied.
The method according to claim 1,
The selecting step is
calculating an eigenvector centrality value of each of the plurality of nodes; and
determining the at least one important node among the plurality of nodes based on the eigenvector centrality value,
The one or more important nodes are further selected by a preset number each time the selecting step is repeatedly performed.
3. The method according to claim 2,
The eigenvector centrality value is
It is calculated based on an adjacency matrix indicating whether each of the plurality of nodes is connected and an eigenvalue of the adjacency matrix.
The method according to claim 1,
Calculating the node similarity includes:
and calculating the node similarity based on whether there is a connection between the one or more important nodes and the remaining nodes and a Jaccard similarity between the one or more important nodes and the remaining nodes.
The method according to claim 1,
The generating step is
Classifying the plurality of nodes into the one or more communities based on the number of important nodes directly connected through an edge.
6. The method of claim 5,
The generating step is
In the case of a first-type node not directly connected to the one or more important nodes among the plurality of nodes, a node having the highest node similarity with the first-type node among nodes directly connected to the first-type node; classifying the first type of node into the same community;
In the case of a second type of node directly connected to one important node among the plurality of nodes, classifying the directly connected one important node and the second type of node into the same community;
In the case of a third type of node directly connected to a plurality of important nodes among the plurality of nodes, a critical node having the highest node similarity with the third type of nodes among the plurality of important nodes and the third type of node are selected. A community detection method that classifies into the same community.
The method according to claim 1,
Calculating the modularity comprises:
A community detection method for calculating the modularity of the classification result using Equations 1 to 3 below.
[Equation 1]

[Equation 2]

[Equation 3]

(Where Q is modularity, m is the total number of edges in the network, and Ni and Nj are arbitrary nodes in the network, respectively)
The method according to claim 1,
The preset condition is
wherein all of the plurality of nodes are selected as the important nodes.
one or more processors;
Memory; and
one or more programs;
wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors;
the one or more programs,
instructions for selecting one or more important nodes in a network including a plurality of nodes;
instructions for calculating a node similarity between each of the one or more important nodes and the remaining nodes;
instructions for generating a classification result of classifying the plurality of nodes into one or more communities based on each of the one or more important nodes based on the node similarity;
instructions for calculating modularity for the classification result; and
instructions for repeatedly performing the selecting step, calculating the node similarity, generating the classification result, and calculating the modularity until a preset condition is satisfied; and
and a command for determining, as a final classification result, a classification result having the greatest modularity among a plurality of classification results generated through the iterative execution when the preset condition is satisfied.
10. The method of claim 9,
The command to select is:
instructions for calculating an eigenvector centrality value of each of the plurality of nodes; and
instructions for determining the at least one significant node among the plurality of nodes based on the eigenvector centrality value;
The one or more important nodes are further selected by a preset number each time the selecting step is repeatedly performed.
11. The method of claim 10,
The eigenvector centrality value is
The community detection apparatus is calculated based on an adjacency matrix indicating whether each of the plurality of nodes is connected and an eigenvalue of the adjacency matrix.
10. The method of claim 9,
The command for calculating the node similarity is:
and calculating the node similarity based on whether there is a connection between the one or more important nodes and the remaining nodes and a Jaccard similarity between the one or more important nodes and the remaining nodes.
10. The method of claim 9,
The command to generate the above is
Classifying the plurality of nodes into the one or more communities based on the number of important nodes directly connected through an edge.
14. The method of claim 13,
The command to generate the above is
In the case of a first-type node not directly connected to the important node among the plurality of nodes, a node having the highest node similarity with the first-type node among nodes directly connected to the first-type node and the second node Classify nodes of type 1 into the same community,
In the case of a second type of node directly connected to only one important node among the plurality of nodes, classifying the directly connected one important node and the second type of node into the same community;
In the case of a third type node directly connected to two or more important nodes among the plurality of nodes, the third type of node and the critical node having the highest node similarity with the third type node among the two or more important nodes A community detection device that classifies nodes into the same community.
10. The method of claim 9,
The instruction for calculating the modularity is:
A community detection device for calculating the modularity of the classification result using Equations 1 to 3 below.
[Equation 1]

[Equation 2]

[Equation 3]

(Where Q is modularity, m is the total number of edges in the network, and Ni and Nj are arbitrary nodes in the network, respectively)
10. The method of claim 9,
The preset condition is
wherein all of the plurality of nodes are selected as the important nodes.

Images