CN108492201A - A kind of social network influence power maximization approach based on community structure - Google Patents
A kind of social network influence power maximization approach based on community structure Download PDFInfo
- Publication number
- CN108492201A CN108492201A CN201810269184.XA CN201810269184A CN108492201A CN 108492201 A CN108492201 A CN 108492201A CN 201810269184 A CN201810269184 A CN 201810269184A CN 108492201 A CN108492201 A CN 108492201A
- Authority
- CN
- China
- Prior art keywords
- node
- community
- influence
- network
- follows
- 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.)
- Granted
Links
- 238000013459 approach Methods 0.000 title claims abstract description 16
- 238000000034 method Methods 0.000 claims abstract description 18
- 230000008901 benefit Effects 0.000 claims abstract description 8
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000010586 diagram Methods 0.000 claims description 8
- 239000012141 concentrate Substances 0.000 claims description 6
- 230000010354 integration Effects 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 230000003213 activating effect Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 9
- 230000000644 propagated effect Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The social network influence power maximization approach based on community structure that the invention discloses a kind of, the detailed process of this method are as follows：(1) community is divided, both candidate nodes collection is formed, by being divided to network, identifying the core node and boundary node in network and forming both candidate nodes collection；(2) node is heuristically selected, for each node that both candidate nodes are concentrated, pass through the degree of node, community's scale, community's number of connection and influence power weight verify potential influence, to which heuristically subset is added in the maximum node of selection potential influence；(3) greedy algorithm is executed, subset is added using the maximum node of greedy algorithm selection marginal benefit.The present invention further improves the accuracy rate and operational efficiency for excavating initial seed node, efficiently solves social network influence power maximization problems by analyzing effect of the community structure in influence power propagation.
Description
Technical field
The present invention relates to field of social network, especially a kind of social network influence power maximization side based on community structure
Method.
Background technology
In recent years, with the rise of social networks, more and more social platforms such as Facebook, Twitter and
Google+ etc. causes the extensive concern of people.These platforms make various information in social network as the carrier of social networks
It is propagated on network.How to enable these information maximumlly to be propagated outward by these social platforms, allows more use
Family is gone to receive these information, is referred to as " maximizing influence problem ".The maximizing influence problem of social networks is social network
One hot issue of network research, and in the marketing, the fields such as transmission and rumour control have great application value.
Social network influence power maximization problems is how to choose Top-K seed node to be propagated, final to make
Spread scope it is maximum.The problem is proved to be a np hard problem, therefore, mainly there is two major classes solution at present, a kind of
It is coverage preferably greedy algorithm, another kind of is the higher heuritic approach of efficiency.Due to greedy algorithm need spend compared with
Long run time, and the result of heuritic approach is unstable, therefore the mixing for combining heuristic and greedy algorithm to generate is calculated
Method is to solve the problems, such as a kind of relatively good method of maximizing influence instantly, such algorithm mainly open in the first stage by application
Hairdo algorithm, second stage application greedy algorithm.All these maximizing influence algorithms are typically all based on two kinds of influences
Power propagation model, i.e. linear threshold model (LT) and independent cascade model (IC), wherein independent cascade model be again it is a kind of less
Stable model needs largely to simulate, the advantage that linear threshold model has independent cascade model incomparable, " accumulative
Characteristic " is so that node can activate a large amount of node, specific rules to be in subsequent activation：Each node starts false
It is set as active or inactive node, and a threshold value is assigned to indicate that the node can be by each inactive node
The complexity of influence, when what only the node was all enlivens the sum of neighbours' influence power more than or equal to the Node B threshold, the node
It will be activated.Here each live-vertex can repeatedly participate in activation, therefore, inactive when starting one
When node is not activated, the influence power with its neighbor node is constantly being accumulated, its possibility being activated is increased.
In general, verification maximizing influence problem is mainly there are two index, and one is coverage, and one is to execute
Efficiency.However, current most of work are there is no the actual structure problem of network is considered, each network is with community structure
Characteristic, i.e. community's internal connection is close, is contacted between community sparse.By the analysis to structure, shadow can be further increased
Range and execution efficiency are rung, i.e., the core node inside community can be such that information is propagated inside community as early as possible, and community
Between boundary node can expand again information propagation range, by divide community, to identify that this two classes node can improve
Execution efficiency and accuracy rate.Therefore, for the larger problem of network size in terms of maximizing influence, Louvain algorithms are used
Divide community, the algorithm be it is a kind of quickly and accuracy rate higher algorithm, may be used on large scale network.
Invention content
The technical problem to be solved by the present invention is to：For the deficiency of technology in existing maximizing influence algorithm, one is proposed
Kind is based on community structure social network influence power maximization approach, passes through the work for analyzing community structure in actual information propagation
With further increasing the coverage of seed node under the premise of ensureing efficiency of algorithm.
The technical scheme is that：
A kind of social network influence power maximization approach based on community structure, the method includes the following steps：
(1) social network diagram is built：G=(V, E), wherein G represent social networks, and V represents node set, and E represents network
Line set；
(2) community is divided, both candidate nodes collection is generated：Use the Louvain algorithms that Blondel et al. is proposed to input first
Network G carry out community's division, generate M community, i.e. C=(C1,C2,...CM), next finds out the boundary node of each community
Collect SboundaryAnd core node collection ScoreUnion is taken, both candidate nodes collection CS, wherein S are formedcoreIt is according to degree centrality selection
The 10% of each community number spends larger node as core node collection；
(3) node is heuristically selected：The both candidate nodes heuristically formed from step (2) concentrate selectionIt is a potential
Seed node collection S is added in the maximum node of influence power, while the activation of seed node collection S is executed using linear threshold model
Journey generates initial live-vertex collection A, wherein k and represents the number of destination node set S, and c represents heuristic factor；
(4) greedy algorithm is executed：Continue the both candidate nodes formed from step (2) using greedy algorithm and concentrates selectionSubset S is added in a maximum node of marginal benefit, while being generated new into line activating using linear threshold model
Set A is added in live-vertex.
Further, the concrete operation step of Louvain algorithms is as follows in step (2)：
(a) merge community：By each node in network as a community, it is then based on modularity gain maximization
Standard determines which neighbours community merges, and repeats this process, until modularity gain no longer increases, modularity gain
It is defined as follows：
Wherein, ∑inIndicate all side weights sums in corporations C, ∑totIndicate the power on all sides for being connected to corporations C
The sum of value, ki,inIndicate node i to the weights sum on all sides of corporations C, total number of edges of m expression networks；
(b) new network is built：The new community that step (a) is obtained builds new network weight as a new node
Step (a) is executed again；
(c) the two stages repeat, until modularity gain no longer changes；
Further, seed node collection is added in the maximum node of selection potential influence in the step (3), then each section
The calculating process of the potential influence of point is as follows：
(a) for any one node v in figure G, the community attributes of each node, i.e. core node or side are first determined whether
Boundary's node, and calculate separately based on community structure the community influence of each node；
(b) its community influence is assessed for core node, the degree of integration node and community's number where node, counts
It is as follows to calculate formula：
CI (v)=CD(v)+CS(v)/2
Wherein, CD(v) degree of community, C are representedS(v) community's scale where node is represented；
(c) for boundary node, the degree of integration node, the neighbours community for the community's number and the node that node is connected directly
Community scale mean value assess its community influence, calculation formula is as follows：
CI (v)=CD(v)+CN(v)+AvgNS(v)/3
Wherein, CD(v) degree of community, C are representedN(v) community's number that node is connected directly, AvgN are representedS(v) node is represented
Neighbours community community's scale mean value, calculation formula is：
Wherein, | Ci(w) | the scale of community where representing the neighbours w of node v；
(d) in order to keep each index consistent to the contribution of community influence, optimized using normalization standard, combining step
(b) and (c), the community influence of each node v is defined as follows in network：
Wherein, each index is the later result of normalization；
(e) in addition to the community influence that step (d) obtains, to neighbor node w, there are one directly affect each node v
Power weight bvw, both comprehensive, the potential influence of each node v calculates as follows in network：
Wherein, w ∈ neighbor (v),Indicate that node w is the inactive neighbours of node v.
Further, the concrete operation step of greedy algorithm is as follows in the step (4)：
(a) it initializes：Seed node collection S is initialized；
(b) marginal benefit of each node v is calculated：It is expressed as by the way that institute's energy band in a node v to subset S is added
The final influence power increment come, then calculation formula is as follows：
σ(S+v)-σ(S)
Wherein, σ (g) indicates to influence force function；
(c) seed node is selected：Select the maximum node of influence power gain that subset S is added, and to the shadow of each node
Power is rung to be updated；
(d) repeat step (c), until selection meets k node of target.
The beneficial effects of the invention are as follows：A kind of social network influence power maximization approach based on community structure,
By the community structure characteristic of network, to identify the core node and boundary node of community, and both candidate nodes collection is formed, secondly profit
Community influence and the influence power weight of itself of each node are integrated with the accumulation characteristics of linear threshold heuristically to select
With the maximum seed node of potential influence, finally seed node is selected using greedy algorithm.By such method, into one
Step improves the accuracy rate and operational efficiency for excavating initial seed node, efficiently solves the maximization of social network influence power and asks
Topic.
Description of the drawings
Present invention will be further explained below with reference to the attached drawings and examples.
Fig. 1 is a kind of flow chart of social network influence power maximization approach based on community structure of the present invention；
Fig. 2 is heuristic factor c of the present invention difference heuristic factor coverage design sketch when seed node scale is 50；
Fig. 3 is heuristic factor c of the present invention difference heuristic factor run time design sketch when seed node scale is 50；
Fig. 4 is the coverage effect contrast figure of the present invention and existing algorithm in HepTh social networks；
Fig. 5 is the coverage effect contrast figure of the present invention and existing algorithm in Brightkite social networks；
Fig. 6 is the coverage effect contrast figure of the present invention and existing algorithm in Epinions social networks；
Fig. 7 is the coverage effect contrast figure of the present invention and existing algorithm in Amazon social networks；
Fig. 8 is the present invention and existing algorithm run time comparison diagram on four social networks；
Specific implementation mode
In conjunction with the accompanying drawings, the present invention is further explained in detail.These attached drawings are simplified schematic diagram, only with
Illustration illustrates the basic structure of the present invention, therefore it only shows the composition relevant to the invention.
It is a kind of social network influence power maximization approach based on community structure of the present invention as shown in Figure 1, it is specific to walk
It is rapid as follows：
Step 1, social network diagram is built：G=(V, E),
Wherein G represents social networks, and V represents node set, and E represents the line set of network.
Step 2, community is divided, both candidate nodes collection is generated.
It uses the Louvain algorithms that industry proposes to carry out community's division to the network G of input first, generates M community, i.e.,
C=(C1,C2,...CM), next finds out the boundary node set S of each communityboundaryAnd core node collection ScoreTake union, shape
At both candidate nodes collection CS, wherein ScoreBe according to degree centrality select the 10% larger node of degree of each community's number as
Core node collection.Louvain algorithms described in the step are as follows：
(2a) merges community：By each node in network as a community, it is maximum to be then based on modularity gain
Change standard determines which neighbours community merges, and repeats this process, and until modularity gain no longer increases, modularity increases
Benefit is defined as follows：
Wherein, ∑inIndicate all side weights sums in corporations C, ∑totIndicate the power on all sides for being connected to corporations C
The sum of value, ki,inIndicate node i to the weights sum on all sides of corporations C, total number of edges of m expression networks.
(2b) builds new network：The new community that step (2a) is obtained builds new network as a new node
Repeat step (2a).
(2c) the two stages repeat, until modularity gain no longer changes.
Step 3, node is heuristically selected.
The both candidate nodes heuristically formed from step 2 concentrate selectionKind is added in a maximum node of potential influence
Child node collection S, while using the activation of linear threshold model execution seed node collection S, generating initial live-vertex collection
A, wherein k represent the number of destination node set S, and c represents heuristic factor.The calculating of potential influence described in the step
Journey is as follows：
(3a) first determines whether the community attributes of each node for any one node v in figure G, i.e., core node or
Boundary node, and calculate separately based on community structure the community influence of each node.
(3b) assesses its community influence for core node, the degree of integration node and community's number where node,
Calculation formula is as follows：
CI (v)=CD(v)+CS(v)/2
Wherein, CD(v) degree of community, C are representedS(v) community's scale where node is represented.
(3c) is for boundary node, the degree of integration node, the neighbours society for the community's number and the node that node is connected directly
Community's scale mean value in area assesses its community influence, and calculation formula is as follows：
CI (v)=CD(v)+CN(v)+AvgNS(v)/3
Wherein, CD(v) degree of community, C are representedN(v) community's number that node is connected directly, AvgN are representedS(v) node is represented
Neighbours community community's scale mean value, calculation formula is：
Wherein, | Ci(w) | the scale of community where representing the neighbours w of node v.
(3d) is optimized, comprehensive step to keep each index consistent to the contribution of community influence using normalization standard
Suddenly (3b) and (3c), the community influence of each node v is defined as follows in network：
Wherein, each index is the later result of normalization.
The community influence that (3e) is obtained in addition to step (3d), there are one direct shadows to neighbor node w by each node v
Ring power weight bvw, both comprehensive, the potential influence of each node v calculates as follows in network：
Wherein, w ∈ neighbor (v),Indicate that node w is the inactive neighbours of node v.
Step 4, greedy algorithm is executed.
Continue the both candidate nodes formed from step 2 using greedy algorithm and concentrates selectionA marginal benefit is maximum
Subset S is added in node, while generating new live-vertex into line activating using linear threshold model and set A is added.The step
Described in greedy algorithm be as follows：
(4a) is initialized：Seed node collection S is initialized.
(4b) calculates the marginal benefit of each node v：It is expressed as by the way that institute's energy in a node v to subset S is added
The final influence power increment brought, then calculation formula is as follows：
σ(S+v)-σ(S)
Wherein, σ (g) indicates to influence force function.
(c) seed node is selected：Select the maximum node of influence power gain that subset S is added, and to the shadow of each node
Power is rung to be updated.
(4d) repeats step (4c), until selection meets k node of target.
Embodiment：
One, data set and experimental setup
In this embodiment, using the disclosed data set HepTh data sets of four different scales from SNAP,
Brightkite data sets, Epinions data sets and Amazon data sets.HepTh data sets come from the conjunction of high-energy physics theory
Author's network is a non-directed graph.Brightkite data sets are a location-based social networks, are a non-directed graphs.
Epinions data sets come from trust network, are to be formed by trusting in the member selection part of the websites Epinions to comment on
Linking relationship, therefore be a digraph.Amazon data sets come from Amazon purchase website, if two in website
Product is often bought jointly, will be there are one linking relationship, therefore is also that there are one digraphs.The static state of this four data sets
Structure feature statistics is as shown in table 1.
Table 1：Experimental data static structure characteristic statistics
Its threshold value of linear threshold model used in the present invention is often assigned a random number between [0,1], this
In in order to keep result more clear, the present invention uses classical threshold θ=0.5 of the propositions such as Kempe, and the influence of linear threshold model
Power weight is often provided with as bvw=1/CD(v), this indicates that node v is the same to the contribution of each neighbours, but this is not
The case where meeting real world, therefore we are arranged influence power weight and areWherein CD(v) it represents
The number of degrees of node v, N (v) represent the neighborhood of node v.
All emulation experiments are arranged using a combined method HPG, greedy method Greedy, webpage in following embodiment
Sequence method PageRank, maximal degree method Degree and method of randomization Random and PHG (Partition- of the present invention
Heuristic-Greedy it) compares.
Two, community's performance is divided
For accurate identification key node, the accuracy rate for dividing community is particularly important, and for large-scale society
Network is handed over, run time is also a necessary Consideration, and comprehensive various considerations, the present invention has selected Louvain algorithms
Community is divided, which is one and may be used on extensive social networks and the higher algorithm of accuracy rate, divides knot
Fruit is as shown in table 2.
Table 2：Community discovery result
It is that characterization community divides quality, modularity Q and parameter u there are two parameter by analyzing table 2.Modularity Q is
Compare the Connection Density difference of existing network and baseline network in the case where identical community divides to weigh the quality of network division, modularity
Value it is higher, represent network division it is better.The range of the module angle value of four networks is between 0.76-0.91 in table 2, explanation
Louvain algorithms still have higher accuracy rate for dividing community.And parameter u=(Smin/Smax) represent it is every in network
One node is not with its neighbour in the average probability of same community, which directly determines the power of community structure, the parameter
Smaller to represent that community structure is stronger, the parameter of four data sets is below 0.01 in table 2, illustrates the society that this four networks divide
Area is all strong community structure, is more advantageous to the key node identified in community.Consolidated statement 2 and correlation analysis, it is known that
Louvain algorithms are applied to one proper algorithm of extensive social networks.
Three, the selection of heuristic factor c
It is suitable for each collection selection one by considering that influence power propagates the resultant effect with run time
Heuristic factor c, so that the present invention can reach optimum efficiency.Fig. 2 and Fig. 3 indicates that different heuristic factors are advised in seed node respectively
Coverage and run time variation diagram when mould is 50, as shown in Figure 2, with the increase of heuristic factor, coverage is most
It is gradually reduced on number data set, and the coverage variation on Epinions data sets is little.Likewise, from the figure 3, it may be seen that
With the increase of heuristic factor, the efficiency of run time is stepping up.This is mainly due to the inefficient of heuritic approach
Caused by the high effect and poor efficiency of fruit and high efficiency and greedy algorithm.Therefore, in order to obtain proper run time
And relatively high coverage, the heuristic factor that Amazon data sets and Brightkite data sets is respectively set is 0.4 He
0.4.And Epinions data sets due in coverage gap it is little, according to its run time, its heuristic factor is arranged
It is 1.For HepTh data sets since scale is smaller, run time gap is little, therefore sets its heuristic factor according to coverage
It is set to 0.2.
Four, coverage
Fig. 4 to Fig. 7 respectively shows seed node scale PHG and other five kinds of algorithm HPG of the present invention in 1-50,
Greedy, PageRank, Degree and Random are on tetra- data sets of HepTh, Brightkite, Epinions and Amazon
Coverage comparison diagram.By this four figures as can be seen that Random algorithms are that performance is worst, this is mainly due to the algorithms
Do not account for any factor, and other algorithms it is all different degrees of considered some factors.Heuritic approach PageRank and
Although Degree performances is better than Random algorithm but more far short of what is expected than remaining algorithm Greedy, HPG and PHG.Work as seed
When scale is smaller, it can be found that Greedy and PHG algorithms have as algorithm and the PHG of the present invention of high-impact it is similar
Coverage, but with the increase of node size, the performance of PHG of the present invention is become better and better.For example, being in seed node scale
When 50, PHG algorithms are higher by 10.7% and 35.5% than Greedy algorithm and HPG algorithms on Amazon data sets respectively.These
As a result all illustrate that community structure information plays an important role in information propagation so that the present invention can have been identified effectively
The node of influence power.
Five, run time
Fig. 8 illustrates the present invention and exists with other other five kinds of algorithms HPG, Greedy, PageRank, Degree and Random
Run time comparison diagram when seed node scale is 50.As can be seen from the figure Random, Degree and PageRank algorithm
Run time it is shorter, this is mainly due to these algorithms and unstable, cannot preferably carry out the propagation of influence power.And
In remaining algorithm, the efficiency of PHG of the present invention is more efficient than other two algorithms Greedy and HPG, this is primarily due to
The present invention is to form both candidate nodes collection by dividing community search key node and core node, to reduce feed search
Space, improve operational efficiency.
In conclusion by analyzing effect of the community structure in influence power propagation, the present invention utilizes community structure information
To identify the key node in social networks so that the present invention not only has promotion in coverage, but also in operational efficiency
On have further optimization, excellent effect.
It is enlightenment with above-mentioned desirable embodiment according to the present invention, through the above description, relevant staff is complete
Various changes and amendments can be carried out without departing from the scope of the technological thought of the present invention' entirely.The technology of this invention
Property range is not limited to the contents of the specification, it is necessary to determine its technical scope according to right.
Claims (4)
1. a kind of social network influence power maximization approach based on community structure, which is characterized in that the method includes following
Step：
(1) social network diagram is built：G=(V, E), wherein G represent social networks, and V represents node set, and E represents the side of network
Set；
(2) community is divided, both candidate nodes collection is generated：Louvain algorithms are used to carry out community's division to the network G of input first,
Generate M community, i.e. C=(C1,C2,...CM), next finds out the boundary node set S of each communityboundaryAnd core node
Collect ScoreUnion is taken, both candidate nodes collection CS, wherein S are formedcoreIt is 10% degree that each community's number is selected according to degree centrality
Larger node is as core node collection；
(3) node is heuristically selected：The didactic both candidate nodes formed from step (2) concentrate selectionA potential impact
Seed node collection S is added in the maximum node of power, while the activation of seed node collection S, production are executed using linear threshold model
Raw initial live-vertex collection A, wherein k represent the number of destination node set S, and c represents heuristic factor；
(4) greedy algorithm is executed：Continue the both candidate nodes formed from step (3) using greedy algorithm and concentrates selectionA side
Subset S is added in the node of border Income Maximum, while generating new live-vertex into line activating using linear threshold model and being added
Set A.
2. a kind of social network influence power maximization approach based on community structure according to claim 1, feature exist
In the concrete operation step of Louvain algorithms is as follows in the step (2)：
(a) merge community：By each node in network as a community, it is then based on modularity gain and maximizes standard
It determines which neighbours community merges, repeats this process, until modularity gain no longer increases, modularity gain is determined
Justice is as follows：
Wherein, ∑inIndicate all side weights sums in corporations C, ∑totIndicate all sides for being connected to corporations C weights it
With ki,inIndicate node i to the weights sum on all sides of corporations C, total number of edges of m expression networks；
(b) new network is built：The new community that step (a) is obtained builds new network and repeats to hold as a new node
Row step (a)；
(c) the two stages repeat, until modularity gain no longer changes.
3. a kind of social network influence power maximization approach based on community structure according to claim 1, feature exist
In seed node collection is added in the maximum node of selection potential influence in the step (3), then the potential influence of each node
Calculating process it is as follows：
(a) for any one node v in figure G, the community attributes of each node, i.e. core node or boundary section are first determined whether
Point, and calculate separately based on community structure the community influence of each node；
(b) its community influence is assessed for core node, the degree of integration node and community's number where node, calculates public
Formula is as follows：
CI (v)=CD(v)+CS(v)/2
Wherein, CD(v) degree of community, C are representedS(v) community's scale where node is represented；
(c) for boundary node, the degree of integration node, the society of the neighbours community for the community's number and the node that node is connected directly
Area scale mean value assesses its community influence, and calculation formula is as follows：
CI (v)=CD(v)+CN(v)+AvgNS(v)/3
Wherein, CD(v) degree of community, C are representedN(v) community's number that node is connected directly, AvgN are representedS(v) neighbour of node is represented
Community's scale mean value of community is occupied, calculation formula is：
Wherein, | Ci(w) | the scale of community where representing the neighbours w of node v；
(d) in order to keep each index consistent to the contribution of community influence, optimized using normalization standard, combining step (b)
(c), the community influence of each node v is defined as follows in network：
Wherein, each index is the later result of normalization；
(e) in addition to the community influence that step (d) obtains, there are one direct influences to weigh to neighbor node w by each node v
Weight bvw, both comprehensive, the potential influence of each node v calculates as follows in network：
Wherein, w ∈ neighbor (v),Indicate that node w is the inactive neighbours of node v.
4. a kind of social network influence power maximization approach based on community structure according to claim 1, feature exist
In the concrete operation step of greedy algorithm is as follows in the step (4)：
(a) it initializes：Seed node collection S is initialized；
(b) marginal benefit of each node v is calculated：It is expressed as to bring by being added in node v to subset S
Final influence power increment, then calculation formula is as follows：
σ(S+v)-σ(S)
Wherein, σ (g) indicates to influence force function；
(c) seed node is selected：Select the maximum node of influence power gain that subset S is added, and to the influence power of each node
It is updated；
(d) repeat step (c), until selection meets k node of target.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810269184.XA CN108492201B (en) | 2018-03-29 | 2018-03-29 | Social network influence maximization method based on community structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810269184.XA CN108492201B (en) | 2018-03-29 | 2018-03-29 | Social network influence maximization method based on community structure |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108492201A true CN108492201A (en) | 2018-09-04 |
| CN108492201B CN108492201B (en) | 2022-02-08 |
Family
ID=63317301
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810269184.XA Active CN108492201B (en) | 2018-03-29 | 2018-03-29 | Social network influence maximization method based on community structure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108492201B (en) |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109064348A (en) * | 2018-09-06 | 2018-12-21 | 上海交通大学 | A method of it blocking rumour community in social networks and inhibits gossip propagation |
| CN109345158A (en) * | 2018-12-19 | 2019-02-15 | 重庆百行智能数据科技研究院有限公司 | Business risk recognition methods, device and computer readable storage medium |
| CN109617871A (en) * | 2018-12-06 | 2019-04-12 | 西安电子科技大学 | Network node immunization method based on community structure information and threshold value |
| CN109685355A (en) * | 2018-12-19 | 2019-04-26 | 重庆百行智能数据科技研究院有限公司 | Business risk recognition methods, device and computer readable storage medium |
| CN110162716A (en) * | 2019-05-21 | 2019-08-23 | 湖南大学 | A kind of influence power community search method and system based on community's retrieval |
| CN110222273A (en) * | 2019-05-14 | 2019-09-10 | 上海交通大学 | Shopping center's promotion method and system in social networks based on geographic community |
| CN110263264A (en) * | 2019-06-28 | 2019-09-20 | 南昌航空大学 | A method of obtaining community network key node |
| CN110750721A (en) * | 2019-10-21 | 2020-02-04 | 秒针信息技术有限公司 | Information pushing method and device, electronic equipment and readable storage medium |
| CN110796561A (en) * | 2019-10-19 | 2020-02-14 | 上海大学 | Influence maximization method and device based on three-hop velocity attenuation propagation model |
| CN110855641A (en) * | 2019-10-30 | 2020-02-28 | 支付宝(杭州)信息技术有限公司 | Community attribute information determination method, device and storage medium |
| CN110992195A (en) * | 2019-11-25 | 2020-04-10 | 中山大学 | Social network high-influence user identification method combined with time factors |
| CN111597665A (en) * | 2020-05-15 | 2020-08-28 | 天津科技大学 | Hierarchical network embedding method based on network partition |
| CN111782969A (en) * | 2020-07-06 | 2020-10-16 | 桂林电子科技大学 | Social network maximum influence node selection method based on geographic area |
| CN111988129A (en) * | 2019-05-21 | 2020-11-24 | 中移(苏州)软件技术有限公司 | Influence maximization data set processing method, device and system |
| CN112148989A (en) * | 2020-10-16 | 2020-12-29 | 重庆理工大学 | Social network node influence recommendation system based on local nodes and degree discount |
| CN112214689A (en) * | 2020-10-22 | 2021-01-12 | 上海交通大学 | Method and system for maximizing influence of group in social network |
| WO2021159629A1 (en) * | 2020-02-14 | 2021-08-19 | 山东科技大学 | Social network influence maximization method based on local and global influences |
| CN113282744A (en) * | 2021-06-07 | 2021-08-20 | 南京邮电大学 | Literary work character relation visualization analysis method based on node influence measurement |
| CN113284030A (en) * | 2021-06-28 | 2021-08-20 | 南京信息工程大学 | Urban traffic network community division method |
| CN113436674A (en) * | 2021-06-23 | 2021-09-24 | 兰州大学 | Incremental community detection method-TSEIA based on TOPSIS seed expansion |
| CN115049002A (en) * | 2022-06-15 | 2022-09-13 | 重庆理工大学 | Complex network influence node identification method based on reverse generation network |
| CN115329209A (en) * | 2022-07-18 | 2022-11-11 | 齐齐哈尔大学 | Method for maximizing influence of time sequence social network of improved K-shell |
| CN115659007A (en) * | 2022-09-21 | 2023-01-31 | 浙江大学 | Dynamic influence propagation seed minimization method based on diversity |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050154639A1 (en) * | 2004-01-09 | 2005-07-14 | Zetmeir Karl D. | Business method and model for integrating social networking into electronic auctions and ecommerce venues. |
| US20070198510A1 (en) * | 2006-02-03 | 2007-08-23 | Customerforce.Com | Method and system for assigning customer influence ranking scores to internet users |
| CN106875281A (en) * | 2017-03-13 | 2017-06-20 | 哈尔滨工程大学 | Community network node method for digging based on greedy subgraph |
-
2018
- 2018-03-29 CN CN201810269184.XA patent/CN108492201B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050154639A1 (en) * | 2004-01-09 | 2005-07-14 | Zetmeir Karl D. | Business method and model for integrating social networking into electronic auctions and ecommerce venues. |
| US20070198510A1 (en) * | 2006-02-03 | 2007-08-23 | Customerforce.Com | Method and system for assigning customer influence ranking scores to internet users |
| CN106875281A (en) * | 2017-03-13 | 2017-06-20 | 哈尔滨工程大学 | Community network node method for digging based on greedy subgraph |
Non-Patent Citations (2)
| Title |
|---|
| BIN HU等: "Improving Louvain Algorithm for Community Detection", 《INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND ENGINEERING APPLICATIONS (AIEA 2016)》 * |
| 田家堂等: "一种新型的社会网络影响最大化算法", 《计算机学报》 * |
Cited By (36)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109064348A (en) * | 2018-09-06 | 2018-12-21 | 上海交通大学 | A method of it blocking rumour community in social networks and inhibits gossip propagation |
| CN109064348B (en) * | 2018-09-06 | 2021-10-08 | 上海交通大学 | Method for locking rumor community and inhibiting rumor propagation in social network |
| CN109617871A (en) * | 2018-12-06 | 2019-04-12 | 西安电子科技大学 | Network node immunization method based on community structure information and threshold value |
| CN109345158A (en) * | 2018-12-19 | 2019-02-15 | 重庆百行智能数据科技研究院有限公司 | Business risk recognition methods, device and computer readable storage medium |
| CN109685355A (en) * | 2018-12-19 | 2019-04-26 | 重庆百行智能数据科技研究院有限公司 | Business risk recognition methods, device and computer readable storage medium |
| CN110222273A (en) * | 2019-05-14 | 2019-09-10 | 上海交通大学 | Shopping center's promotion method and system in social networks based on geographic community |
| CN110222273B (en) * | 2019-05-14 | 2021-08-17 | 上海交通大学 | Business point promotion method and system in social network based on geographic community |
| CN111988129A (en) * | 2019-05-21 | 2020-11-24 | 中移(苏州)软件技术有限公司 | Influence maximization data set processing method, device and system |
| CN110162716A (en) * | 2019-05-21 | 2019-08-23 | 湖南大学 | A kind of influence power community search method and system based on community's retrieval |
| CN111988129B (en) * | 2019-05-21 | 2022-07-01 | 中移(苏州)软件技术有限公司 | Influence maximization data set processing method, device and system |
| CN110263264A (en) * | 2019-06-28 | 2019-09-20 | 南昌航空大学 | A method of obtaining community network key node |
| CN110263264B (en) * | 2019-06-28 | 2021-04-27 | 南昌航空大学 | Method for acquiring social network key node |
| CN110796561A (en) * | 2019-10-19 | 2020-02-14 | 上海大学 | Influence maximization method and device based on three-hop velocity attenuation propagation model |
| CN110796561B (en) * | 2019-10-19 | 2023-04-11 | 上海大学 | Influence maximization method and device based on three-hop velocity attenuation propagation model |
| CN110750721A (en) * | 2019-10-21 | 2020-02-04 | 秒针信息技术有限公司 | Information pushing method and device, electronic equipment and readable storage medium |
| CN110855641B (en) * | 2019-10-30 | 2022-07-01 | 支付宝(杭州)信息技术有限公司 | Community attribute information determination method, device and storage medium |
| CN110855641A (en) * | 2019-10-30 | 2020-02-28 | 支付宝(杭州)信息技术有限公司 | Community attribute information determination method, device and storage medium |
| CN110992195B (en) * | 2019-11-25 | 2023-04-21 | 中山大学 | Social network high-influence user identification method combined with time factors |
| CN110992195A (en) * | 2019-11-25 | 2020-04-10 | 中山大学 | Social network high-influence user identification method combined with time factors |
| WO2021159629A1 (en) * | 2020-02-14 | 2021-08-19 | 山东科技大学 | Social network influence maximization method based on local and global influences |
| CN111597665A (en) * | 2020-05-15 | 2020-08-28 | 天津科技大学 | Hierarchical network embedding method based on network partition |
| CN111782969B (en) * | 2020-07-06 | 2023-05-23 | 桂林电子科技大学 | Social network maximum influence node selection method based on geographic area |
| CN111782969A (en) * | 2020-07-06 | 2020-10-16 | 桂林电子科技大学 | Social network maximum influence node selection method based on geographic area |
| CN112148989A (en) * | 2020-10-16 | 2020-12-29 | 重庆理工大学 | Social network node influence recommendation system based on local nodes and degree discount |
| CN112148989B (en) * | 2020-10-16 | 2021-08-24 | 重庆理工大学 | Social network node influence recommendation system based on local nodes and degree discount |
| CN112214689A (en) * | 2020-10-22 | 2021-01-12 | 上海交通大学 | Method and system for maximizing influence of group in social network |
| CN113282744B (en) * | 2021-06-07 | 2022-11-08 | 南京邮电大学 | Literary work character relation visualization analysis method based on node influence measurement |
| CN113282744A (en) * | 2021-06-07 | 2021-08-20 | 南京邮电大学 | Literary work character relation visualization analysis method based on node influence measurement |
| CN113436674B (en) * | 2021-06-23 | 2023-02-17 | 兰州大学 | Incremental community detection method-TSEIA based on TOPSIS seed expansion |
| CN113436674A (en) * | 2021-06-23 | 2021-09-24 | 兰州大学 | Incremental community detection method-TSEIA based on TOPSIS seed expansion |
| CN113284030A (en) * | 2021-06-28 | 2021-08-20 | 南京信息工程大学 | Urban traffic network community division method |
| CN113284030B (en) * | 2021-06-28 | 2023-05-23 | 南京信息工程大学 | Urban traffic network community division method |
| CN115049002A (en) * | 2022-06-15 | 2022-09-13 | 重庆理工大学 | Complex network influence node identification method based on reverse generation network |
| CN115329209A (en) * | 2022-07-18 | 2022-11-11 | 齐齐哈尔大学 | Method for maximizing influence of time sequence social network of improved K-shell |
| CN115659007A (en) * | 2022-09-21 | 2023-01-31 | 浙江大学 | Dynamic influence propagation seed minimization method based on diversity |
| CN115659007B (en) * | 2022-09-21 | 2023-11-14 | 浙江大学 | Dynamic influence propagation seed minimization method based on diversity |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108492201B (en) | 2022-02-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108492201A (en) | A kind of social network influence power maximization approach based on community structure | |
| Gong et al. | Influence maximization in social networks based on discrete particle swarm optimization | |
| Qiu et al. | PHG: A three-phase algorithm for influence maximization based on community structure | |
| Keikha et al. | Influence maximization across heterogeneous interconnected networks based on deep learning | |
| Aghaee et al. | A survey on meta-heuristic algorithms for the influence maximization problem in the social networks | |
| Berahmand et al. | A link-based similarity for improving community detection based on label propagation algorithm | |
| Nguyen et al. | Probability-based multi-hop diffusion method for influence maximization in social networks | |
| Hong et al. | Efficient minimum cost seed selection with theoretical guarantees for competitive influence maximization | |
| CN103116611A (en) | Social network opinion leader identification method | |
| Li et al. | Cinema: conformity-aware greedy algorithm for influence maximization in online social networks | |
| Dey et al. | Influence maximization in online social network using different centrality measures as seed node of information propagation | |
| CN104050245A (en) | Social network influence maximization method based on activeness | |
| Wang et al. | Influential node identification by aggregating local structure information | |
| Qiao et al. | A new blockmodeling based hierarchical clustering algorithm for web social networks | |
| Liu et al. | Maximum likelihood-based influence maximization in social networks | |
| Wang et al. | Discrete particle swarm optimization based influence maximization in complex networks | |
| Liu et al. | Influence nodes identifying method via community-based backward generating network framework | |
| Zarei et al. | Chaotic memetic algorithm and its application for detecting community structure in complex networks | |
| Fan et al. | The probabilistic maximum coverage problem in social networks | |
| Zhang et al. | Overlapping community‐based particle swarm optimization algorithm for influence maximization in social networks | |
| Ge et al. | User topic preferences based influence maximization in overlapped networks | |
| Bidoni et al. | Discovering Community Structure in Dynamic Social Networks using the Correlation Density Rank | |
| Shakya et al. | Opposition-based genetic algorithm for community detection in social networks | |
| Bouzinis et al. | Incentive-based delay minimization for 6G-enabled wireless federated learning | |
| Shang et al. | Factor Analysis for Influence Maximization Problem in Social Networks |
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 |