A method for discovering user communities in social networks based on Markov similarity enhancement

Abstract

A method for discovering user communities in social networks based on Markov similarity enhancement. First, the initial matrix and state transition matrix are extracted through the topological structure of the network. Then, the steady-state Markov similarity enhancement matrix is ​​obtained through the Markov iterative state transition of the initial matrix. The most similar node pair set of the network is obtained according to the Markov similarity index, and the connected components therein are proposed to complete community discovery. The present invention takes into account the topological relationship between user nodes in the social network, and uses the Markov similarity matrix to enhance the accuracy of community discovery. The obtained social network user community structure is stable and has high accuracy.

Description

Technical Field

[0001] The invention relates to the field of network science, and in particular to a social network user community discovery method based on Markov similarity enhancement. Background Art

[0002] With the vigorous development of Internet information technology, human beings have entered the information age and have a deeper understanding of the world. The Internet covers almost every aspect of our work, study, and life. For example, entertainment, friendship, information consumption, and other aspects of life are constantly being networked and intelligent. Among them, social networks are developing rapidly, and various social networks are emerging in an endless stream. For social networks, how to discover the hidden information of social networks and use the hidden information of social networks to divide communities is particularly important. The connection between subjects in a social network system can be abstracted into a network graph and analyzed and studied. And due to the existence of interpersonal relationships, social networks often have obvious community structure characteristics, with more frequent communication between people in the same community and fewer connections between different communities. Community discovery, as an important technology for analyzing social networks, can discover communities in social networks, which not only helps users find other users with the same interests as themselves, but also helps recommendation systems make recommendations based on different user communities.

[0003] In order to reveal the community structure of social networks, there are currently a variety of community discovery algorithms, such as the GN method based on modularity partitioning, the LPA method based on label propagation, the method based on edge clustering, etc. However, these methods often have problems such as unstable partitioning results, the edge information between nodes in the network is often ignored, and the user community partition is not accurate enough. Therefore, it is particularly important to propose an accurate and high-quality social network user community discovery method. Summary of the invention

[0004] In order to overcome the problems of unstable results and low accuracy of existing community discovery methods, the present invention proposes a social network user community discovery method based on Markov similarity enhancement.

[0005] The technical solution adopted by the present invention to solve its technical problem is:

[0006] A method for discovering user communities in social networks based on Markov similarity enhancement includes the following steps:

[0007] Step 1: Construct an undirected and unweighted social network G(V,E) based on social network data. One social user corresponds to one node. If any two social users have exchanged information, there is an edge between the two corresponding nodes. V represents the set of nodes in the network, E represents the set of edges in the network, and N = |V| represents the number of nodes in the network.

[0008] Step 2: According to the topological structure of the social network G, the adjacency matrix A is obtained as the initial state matrix of the social network G;

[0009] Step 3: Randomly select two neighbor nodes v with edges in the social network G i and v j , calculate the similarity index between two neighbor nodes

[0010]

[0011] Where Γ(v i ), Γ(v j ) represent the nodes v i and v j The neighbor node set of G is traversed through the social network G and the similarity index between all neighbor nodes is calculated;

[0012] Step 4: Calculate the state transition matrix of the social network G in its initial state

[0013] in

[0014] Step 5: Normalize the state transfer matrix S, where

[0015] Step 6: Calculate the Markov similarity enhancement matrix of the social network G in the steady state

[0016]

[0017] in It represents the ratio of the number of network edges to the number of network nodes, Se i,j Represents node v i and v j The Markov similarity index of ;

[0018] Step 6: In the social network G, randomly select a node v i , among all its neighboring nodes, select the corresponding Markov similarity index Se i,j The largest node v j As v i The most similar nodes constitute the most similar node pair (v i ,v j ), traverse the social network G and find the most similar node pairs of all nodes;

[0019] Step 7: Without considering the original edges of the social network G, connect all the most similar node pairs. The users in the social network G will naturally form several disconnected community structures. Within the same community, the network nodes are connected to each other; there is no connection between nodes in different communities; thus forming the user community structure of the social network G.

[0020] The technical concept of the present invention is: for social networks, a Markov similarity enhancement matrix and a Markov similarity index are proposed, thereby completing community discovery, obtaining a more accurate community division structure, and improving the quality of community division.

[0021] The present invention first extracts the initial matrix and state transfer matrix through the topological structure of the network, then obtains the steady-state Markov similarity enhancement matrix through the Markov iterative state transfer of the initial matrix, obtains the most similar node pair set of the network according to the Markov similarity index, and proposes the connected components therein to complete community discovery. The present invention takes into account the topological relationship between the user nodes of the social network, and uses the Markov similarity matrix to enhance the accuracy of community discovery. The obtained social network user community structure is stable and has high accuracy.

[0022] The advantages of the present invention are: combining social network topology features and Markov chains, and high accuracy of community discovery. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a simple social network diagram. The nodes in the diagram represent each user in the social network. 0, 1, 2, 3, ..., 12 are the corresponding user numbers. If two users have the same interests and hobbies, there is an edge between the two nodes. DETAILED DESCRIPTION

[0024] The present invention will be further described below in conjunction with the accompanying drawings.

[0025] Reference Figure 1 The method for recommending network users with the same interests using the method of the present invention comprises the following steps:

[0026] Step 1: Construct an internally connected undirected and unweighted network G(V,E) based on social network data. One user is represented by one node, that is, Figure 1 The dots in the graph, 0, 1, 2, 3, ..., 12 are the corresponding user numbers; if two users have the same interest, there is an edge between the two users. V represents the set of nodes in the network, E represents the set of edges in the network, and N = |V| represents the number of nodes in the network;

[0027] Step 2: According to the topological structure of the social network G, the adjacency matrix A is obtained as the initial state matrix of the social network G;

[0028] Step 3: Randomly select two neighbor nodes v with edges in the social network G i and v j , calculate the similarity index between two neighbor nodes

[0029]

[0030] Where Γ(v i ), Γ(v j ) represent the nodes v i and v j The neighbor node set of G is traversed through the social network G and the similarity index between all neighbor nodes is calculated;

[0031] Step 4: Calculate the state transition matrix of the social network G in its initial state

[0032] in

[0033] Step 5: Normalize the state transfer matrix S, where

[0034] Step 6: Calculate the Markov similarity enhancement matrix of the social network G in the steady state

[0035]

[0036] in It represents the ratio of the number of network edges to the number of network nodes, Se i,j Represents node v i and v j The Markov similarity index of ;

[0037] Step 6: In the social network G, randomly select a node v i , among all its neighboring nodes, select the corresponding Markov similarity index Se i,j The largest node v j As v i The most similar nodes constitute the most similar node pair (v i ,v j ), traverse the social network G and find the most similar node pairs of all nodes;

[0038] Step 7: Without considering the original edges of the social network G, connect all the most similar node pairs. The users in the social network G will naturally form several disconnected community structures. Within the same community, the network nodes are connected to each other; there is no connection between nodes in different communities; thus forming the user community structure of the social network G.

[0039] Step 8: Based on the obtained community structure, recommend other node users in the same community to the corresponding users.

[0040] The contents described in the embodiments of this specification are merely an enumeration of the implementation forms of the inventive concept. The protection scope of the present invention should not be regarded as limited to the specific forms described in the embodiments. The protection scope of the present invention also extends to equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

1. A method for discovering user communities in social networks based on Markov similarity enhancement, characterized in that: The steps include: Step 1: Construct an undirected and unweighted social network G(V,E) based on social network data. One social user corresponds to one node. If any two social users have exchanged information, there is an edge between the two corresponding nodes. V represents the set of nodes in the network, E represents the set of edges in the network, and N = |V| represents the number of nodes in the network. Step 2: According to the topological structure of the social network G, the adjacency matrix A is obtained as the initial state matrix of the social network G; Step 3: Randomly select two neighbor nodes v with edges in the social network G i and v j , calculate the similarity index between two neighbor nodes Where Γ(v i ), Γ(v j ) represent the nodes v i and v j The neighbor node set of G is traversed through the social network G and the similarity index between all neighbor nodes is calculated; Step 4: Calculate the state transition matrix of the social network G in its initial state Step 5: Normalize the state transfer matrix S, where Step 6: Calculate the Markov similarity enhancement matrix of the social network G in the steady state in It represents the ratio of the number of network edges to the number of network nodes, Se i,j Represents node v i and v j The Markov similarity index of ; Step 6: In the social network G, randomly select a node v i , among all its neighboring nodes, select the corresponding Markov similarity index Se i,j The largest node v j As v i The most similar nodes constitute the most similar node pair (v i ,v j ), traverse the social network G and find the most similar node pairs of all nodes; Step 7: Without considering the original edges of the social network G, connect all the most similar node pairs. The users in the social network G will naturally form several disconnected community structures. Within the same community, the network nodes are connected to each other; there is no connection between nodes in different communities; thus forming the user community structure of the social network G.

Images