Dynamic social network community evolution analysis method and system

A technology of social networking and analysis methods, applied in instruments, data processing applications, computing, etc., can solve the problems of distinguishing core node types, failing to make full use of network topology information, etc., to achieve high accuracy and efficiency, and reasonable division results , the effect of high execution efficiency

Active Publication Date: 2019-03-19
CHINA UNIV OF MINING & TECH
1 Cites 5 Cited by

AI-Extracted Technical Summary

Problems solved by technology

[0005] In view of the above analysis, the embodiment of the present invention aims to provide a dynamic social network community evolution analysis method and its ...
View more

Method used

Compared with embodiment 1, the present embodiment further limits step S1~S4, makes full use of the strong characteristics of superspreader node dissemination to judge generation, merger and expansion event, makes the judgment of evolution event 1 typ...
View more

Abstract

The invention relates to a dynamic social network community evolution analysis method and system thereof, belongs to the technical field of network identification, and solves the problems that networktopology structure information cannot be fully utilized and types of core nodes are not distinguished in the prior art. The method comprises the following steps of: for a given dynamic social network, dividing a community structure corresponding to a time slice network for each time slice from a first time slice; calculating a superloader set of each time slice network and a superloader set of each community corresponding to the time slice network according to a community structure division result; determining an evolution event 1 type of each community for the superloader set, i.e., generating, merging and expanding events; and for the superblock set, determining the evolution event type 2 of each community, namely disappearing, splitting and reducing events. According to the method, thecharacteristics of high dissemination of the superloader and the disruption connectivity of the superloader are utilized, the dynamic social network community evolution event is analyzed according tothe change conditions of the two types of node sets, and the evolution event recognition accuracy and efficiency are high.

Application Domain

Data processing applications

Technology Topic

Community structureNetwork identification +8

Image

  • Dynamic social network community evolution analysis method and system
  • Dynamic social network community evolution analysis method and system
  • Dynamic social network community evolution analysis method and system

Examples

  • Experimental program(5)

Example Embodiment

[0085] Example 1
[0086] A specific embodiment of the present invention discloses a method for analyzing the evolution of a dynamic social network community, such as figure 1 shown, including the following steps:
[0087] S1. For a given dynamic social network, starting from the first time slice, divide the community structure corresponding to the time slice network for each time slice;
[0088] S2. Calculate the superspreader set of each time slice network and the superblocker set of each community corresponding to the time slice network according to the community structure division results;
[0089] S3. For the above-mentioned superspreader set, determine the evolution event 1 type of each community; the possible types of the evolution event 1 include generation, merger and expansion events;
[0090] S4. For the above superblocker set, determine the evolution event 2 type of each community; the possible types of the evolution event 2 include disappearance, splitting and shrinking events.
[0091] Compared with the prior art, the types of evolution events provided by this embodiment include generation, merger, expansion, disappearance, division and reduction. The above technical solution utilizes the characteristics of superspreader's strong dissemination and superblocker's ability to destroy connectivity, and analyzes dynamic social network community evolution events according to the changes of these two types of node sets. It has been proved by a large number of experiments that the accuracy and efficiency of the discovery are high. The above technical solution solves the problem that the existing dynamic social network community evolution analysis ignores the internal topology information of the community network and does not distinguish the types of core nodes. Generate, merge, and expand events, and use the characteristics of superblockers to destroy network connectivity to discover community disappearance, split, and shrink events.

Example Embodiment

[0092] Example 2
[0093] Optimizing on the basis of Example 1, such as figure 2 As shown, step S1 can be further refined into the following steps:
[0094] S11. For a given dynamic social network, starting from the first time slice, obtain the neighbor relationship between each node in each time slice network;
[0095] S12. According to the neighbor relationship above, divide the community structure corresponding to each time slice network through the QCA algorithm.
[0096] The QCA algorithm is a fast adaptive dynamic community discovery algorithm based on modularity. It formulates different community structure update strategies for each network change, including node addition, node deletion, edge addition, and edge deletion. By maximizing modularity To determine the community affiliation of incremental nodes.
[0097] Preferably, as image 3 As shown, step S2 can be further refined into the following steps:
[0098] S21. Obtain the superspreader set of each time slice network through the Degree Discount algorithm;
[0099] S22. Obtain the superblocker set of each community corresponding to each time slice network through the CoreHD algorithm.
[0100] Specifically, the calculation method of the superspreader is
[0101] dd v = d v -2t v -(d v -t v )t v p
[0102] where dd v is the influence score of superspreader node v, d v is the degree value of superspreader node v, t v is the number of neighbor nodes of superspreader node v selected as seed nodes, and p is the propagation probability of superspreader node information in the social network.
[0103] The Degree Discount algorithm recursively uses the above formula to calculate the influence score of the node, and each time finds out the node with the largest influence score and joins the seed node set. The set of seed nodes is the above-mentioned superspreader set.
[0104] The calculation method of the superblocker is: first delete as few nodes as possible through the CoreHD algorithm, so that there is no cycle in the network after the node is deleted. Then, the node with the highest degree value is adaptively removed from the network based on 2-core decomposition to find the smallest possible set of nodes, namely the superblocker set, so that after removing these nodes, the network is decomposed into several independent connected components.
[0105] Preferably, as Figure 4 As shown, step S3 can be further refined into the following steps:
[0106] S31. Judging the type of evolution event 1 according to the calculation model of the generated event. If the superspreader node of the current time slice t does not exist in the previous time slice t-1 or is not a node in the superspreader set, then it is determined that the evolution event 1 is an event. The community represented by the superspreader node is the newly generated community in the current time slice t.
[0107] S32. Determine the type of evolution event 1 according to the calculation model of the merged event. If the two superspreader nodes in the same community in the current time slice t belong to different communities in the previous time slice t-1, then determine the evolution event 1 as In a merge event, the communities represented by the two superspreader nodes are merged in the current time slice t.
[0108] S33. Determine the type of evolution event 1 according to the calculation model of the expansion event. If the size of the superspreader node in a community in the current time slice t is larger than the size of the superspreader node in the community corresponding to the previous time slice t-1, it is determined that the evolution event 1 is expansion event, the community expands in the current time slice t.
[0109] Preferably, as Figure 5 As shown, step S4 can be further refined to include the following steps:
[0110] S41. Determine the type of evolution event 2 according to the calculation model of the disappearance event. If the superblocker node in the previous time slice t-1 does not exist in the current time slice t or is no longer a node in the superblocker set, then it is determined that the evolution event 2 is disappearance Event, the community once represented by the superblocker node dies in the current time slice t.
[0111] S42. Judging the type of evolution event 2 according to the calculation model of the split event, if the two superblocker nodes in the same community in the previous time slice t-1 are in two different communities in the current time slice t, then determine the evolution event 2 For a split event, the community in the previous time slice t-1 splits in the current time slice t.
[0112] S43. Determine the type of evolution event 2 according to the calculation model of the reduction event. If the size of the superblocker node in a certain community in the previous time slice t-1 is larger than the size of the superblocker node in the community corresponding to the current time slice t, then the evolution event 2 is determined to be reduction event, the community shrinks in the current time slice t.
[0113] Preferably, the calculation model for generating events is
[0114]
[0115]
[0116] In the formula, ss represents a superspreader node, represents the kth community at time t, SS t Represents the superspreader collection of dynamic social networks at time t, SS t-1 Represents the superspreader set of the dynamic social network at time t-1, and Birth()=1 means that the event is true.
[0117] At time t, for any The superspreader node ss is not a superspreader at time t-1, then the community to which the node ss belongs It is a newly generated community.
[0118] The calculation model of the merge event is
[0119]
[0120]
[0121] In the formula, ss 1 、ss 2 Indicates two superspreader nodes, are two communities at time t-1, is the kth community at time t, SS t-1 Represents the superspreader set of the dynamic social network at time t-1, and Merging()=1 indicates that the merge event is true.
[0122] At time t-1, there exists ss 1 、ss 2 two communities within the superspreader, but at time t, ss 1 、ss 2 in the same community within, the community A merge has occurred.
[0123] The calculation model of the expansion event is
[0124]
[0125]
[0126] In the formula, ss represents a superspreader node, is the kth community at time t-1, is the sth community at time t, SS tis the superspreader collection of dynamic social networks at time t, SS t-1 Represents the superspreader set of the dynamic social network at time t-1, and Expansion()=1 means that the expansion event is true.
[0127] At time t-1, any ss is a community The superspreader inside is the community at time t within the superspreader, and The number of superspreaders within is less than within the superspreader number, the community Expansion occurred.
[0128] Preferably, the calculation model of the disappearance event is
[0129]
[0130]
[0131] In the formula, sb represents a superblocker node, Indicates the kth community at time t-1, SB t-1 Indicates the superblocker set of the dynamic social network at time t-1, and Death()=1 indicates that the disappearance event is true.
[0132] At time t-1, any community The superblocker node sb within is no longer a superblocker at time t, then the community where sb is located Dying happened.
[0133] The calculation model of the splitting event is
[0134]
[0135]
[0136] In the formula, sb 1 、sb 2 Indicates two superblocker nodes, is the kth community at time t-1, and are two communities at time t, SB t is the superblocker set of the dynamic social network at time t, SB t-1 is the superblocker set of the dynamic social network at time t, and Splitting()=1 means that the splitting event is true.
[0137] At time t-1, the same community Two superblockersb inside 1 、sb 2 , at time t, in different communities and , then sb 1 、sb 2 the community There was a split.
[0138] The calculation model of the reduction event is
[0139]
[0140]
[0141] In the formula, sb represents a superblocker node, is the kth community at time t-1, is the sth community at time t, SB t is the superblocker set at time t, SB t-1 is the superblocker set of the dynamic social network at time t-1, and Reduction()=1 means that the reduction event is true.
[0142] At time t-1, the community Any supersblcokersb within the community at time t within the superblocker, but the community There are more superblockers in the community than in the community The number of superblockers within the community A shrinkage has occurred.
[0143] Compared with embodiment 1, this embodiment further restricts steps S1 to S4, making full use of the characteristics of superspreader node's strong dissemination to judge generation, merger and expansion events, so that the judgment of evolution event 1 type is more accurate and fully The disappearance, splitting and shrinking events are judged by using the superblocker node's strong destructive characteristics to the social network connectivity, which makes the judgment of the evolution event 2 type more accurate.

Example Embodiment

[0144] Example 3
[0145] Another specific embodiment of the present invention discloses a dynamic social network community evolution analysis system, which uses the method described in Embodiment 1 to perform evolution analysis. Specifically, such as Image 6 As shown, the dynamic social network community evolution analysis system includes a community division module, a superspreader set calculation module, a superblocker set calculation module, a superspreader-based evolution event analysis module, and a superblocker-based evolution event analysis module. Among them, the output of the community division module is connected to the input of the superspreader set calculation module, the superblocker set calculation module, the superspreader-based evolution event analysis module, and the superblocker-based evolution event analysis module. The output end of the superspreader set calculation module is connected to the input end of the superspreader-based evolution event analysis module, and the output end of the superblocker set calculation module is connected to the output end of the superblocker-based evolution event analysis module.
[0146] The community division module is used to divide the community structure corresponding to the time slice network for each time slice from the first time slice according to a given dynamic social network;
[0147] The superspreader set calculation module is used to calculate the superspreader set of each time slice network according to the community structure division result output by the community division module;
[0148] The evolution event analysis module based on superspreader is used to determine the evolution event 1 type of each community for the above-mentioned superspreader set; the possible types of the evolution event 1 include generation, merger and expansion;
[0149] The superblocker-based evolution event analysis module is used to determine the evolution event 2 type of each community for the superblocker set; the possible types of the evolution event 2 include disappearance, division and shrinkage.
[0150] Preferably, the superspreader-based evolution event analysis module includes a generation event judging unit, a merge event judging unit, and an expansion event judging unit.
[0151] Generate an event judging unit, which is used to judge the evolution event 1 type according to the generation event model, if the superspreader node of the current time slice t does not exist in the previous time slice t-1 or is not a node in the superspreader set, then determine the evolution event 1 means an event is generated, and the community represented by the superspreader node is a newly generated community in the current time slice t.
[0152] The merge event judging unit is used to judge the evolution event 1 type according to the calculation model of the merge event. If the two superspreader nodes in the same community in the current time slice t belong to different communities in the previous time slice t-1, then judge Evolution event 1 is a merge event, and the communities represented by the two superspreader nodes merge in the current time slice t.
[0153] The expansion event judging unit is used to judge the type of evolution event 1 according to the calculation model of the expansion event. If the superspreader node scale in a community in the current time slice t is larger than the superspreader node scale in the community corresponding to the previous time slice t-1, then judge Evolution event 1 is an expansion event, and the community expands in the current time slice t.
[0154] Preferably, the superblocker-based evolution event analysis module includes a disappearance event judgment unit, a split event judgment unit, and a reduction event judgment unit.
[0155] The disappearance event judging unit is used to judge the evolution event 2 type according to the calculation model of the disappearance event. If the superblocker node in the previous time slice t-1 does not exist in the current time slice t or is no longer a node in the superblocker set, then judge Evolution event 2 is a disappearance event, and the community once represented by the superblocker node died in the current time slice t;
[0156] The split event judging unit is used to judge the type of evolution event 2 according to the calculation model of the split event. If the two superblocker nodes in the same community in the previous time slice t-1 are in two different communities in the current time slice t, Then it is determined that evolution event 2 is a split event, and the community in the previous time slice t-1 splits in the current time slice t;
[0157] The reduction event judging unit is used to judge the type of evolution event 2 according to the calculation model of the reduction event. If the size of the superblocker node in a certain community in the previous time slice t-1 is larger than the size of the superblocker node in the community corresponding to the current time slice t, then determine Evolution event 2 is a shrinkage event, and the community shrinks in the current time slice t.
[0158] The system shown in this embodiment is based on the same principle as the method in Embodiment 2, and their similarities can be used for reference.
[0159] Preferably, the QCA algorithm is a fast adaptive dynamic community discovery algorithm based on modularity, and formulates different community structure update strategies for each network change, including node addition, node deletion, edge addition, and edge deletion. degree maximization to determine the community affiliation of incremental nodes.
[0160] When implemented, the process of the system is as follows Figure 7 shown.
[0161] Compared with the prior art, the system provided by this embodiment utilizes the characteristics of superspreader's strong dissemination and superblocker's disruptive connectivity, and analyzes dynamic social network community evolution events according to the changes of these two types of node sets. After a large number of experiments, it is found that Higher accuracy and efficiency. This embodiment solves the problem that the existing dynamic social network community evolution analysis ignores the internal topology information of the community network and does not distinguish the core node types. The core nodes are divided into superspreader and superblocker, and the characteristics of superspreader are used to discover the characteristics of the community. Generate, merge, and expand events, and use the characteristics of superblockers to destroy network connectivity to discover community disappearance, split, and shrink events.

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.

Similar technology patents

Method, apparatus, computer device and storage medium for commercial site selection

ActiveCN109345130Areduce computing costImprove efficiency and accuracy
Owner:深圳市和讯华谷信息技术有限公司

Classification and recommendation of technical efficacy words

  • Improve execution efficiency
  • Improve efficiency and accuracy

Bluetooth connecting method and mobile terminal

InactiveCN103067055AReduce scan timeImprove efficiency and accuracy
Owner:SHENZHEN TINNO WIRELESS TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products