CN108510115A - A kind of maximizing influence analysis method towards dynamic social networks - Google Patents
A kind of maximizing influence analysis method towards dynamic social networks Download PDFInfo
- Publication number
- CN108510115A CN108510115A CN201810269151.5A CN201810269151A CN108510115A CN 108510115 A CN108510115 A CN 108510115A CN 201810269151 A CN201810269151 A CN 201810269151A CN 108510115 A CN108510115 A CN 108510115A
- Authority
- CN
- China
- Prior art keywords
- node
- influence
- delay
- power
- social networks
- 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.)
- Pending
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 16
- 238000009826 distribution Methods 0.000 claims abstract description 51
- 238000005315 distribution function Methods 0.000 claims abstract description 20
- 230000004913 activation Effects 0.000 claims abstract description 19
- 102100034799 CCAAT/enhancer-binding protein delta Human genes 0.000 claims abstract description 13
- 101000945965 Homo sapiens CCAAT/enhancer-binding protein delta Proteins 0.000 claims abstract description 12
- 238000000034 method Methods 0.000 claims abstract description 11
- 238000010586 diagram Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 6
- 230000006854 communication Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 13
- 238000011160 research Methods 0.000 abstract description 3
- 101001019013 Homo sapiens Mitotic interactor and substrate of PLK1 Proteins 0.000 description 20
- 102100033607 Mitotic interactor and substrate of PLK1 Human genes 0.000 description 20
- 230000003068 static effect Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000000644 propagated effect Effects 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000003612 virological effect Effects 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of maximizing influence analysis method towards dynamic social networks, this method detailed processes：(1) activation probability is obtained, time factor is added in activation probability by using force delay distribution function power law distribution is influenced；(2) influence power propagation model LAIC is established；(3) greedy algorithm is executed, the initial marginal benefit of each node is calculated using greedy algorithm；(4) original greedy algorithm is optimized using CELF algorithms, the efficiency of nodes for research node is improved by the submodule characteristic and influence power Priority Queues that influence force function.Effect of the present invention by analysis time factor in influence power propagation, it has used and a kind of being distributed consistent distribution function power law distribution with true social network node degree, outstanding achievement and rational run time finally are achieved on choosing most influential TOP K nodes, efficiently solves dynamic social network influence power maximization problems.
Description
Technical field
The present invention relates to field of social network, propose a kind of maximizing influence analysis method towards dynamic social networks
GLAIC(Greedy based on LAIC model)。
Background technology
In recent years, with the rise of social networks, more and more users like sharing thinking for oneself in social platform
Method and viewpoint so that social networks plays an increasingly important role in information propagation.Therefore, information is understood in social network
One of the key problem of the person that has become current research research how is propagated and spread in network.Assuming that there are one companies to want
Their new product is publicized, at this moment the said firm will select some influential users, it is desirable to which they can be imitated by public praise
This Products Show should be given to their friend and friends of friends.And the circulation way of this social network influence power is wide
General is applied to many fields, such as viral marketing and commending system.
A critical issue in the propagation of social network influence power is how to choose limited Initial travel seed, to
Keep the final spread scope of the information maximum, is referred to as " maximizing influence problem ".In order to solve this problem, many outstanding
Propagation model and algorithm are suggested, and obtain relatively good achievement.However, currently most of study both for static network,
And have ignored importance of the temporal information in information propagation.In fact, almost all of social networks is all as the time exists
Constantly change, for example, with the continuous variation of time, the linking relationship between social network user can create at any time or
It disappears, therefore by time factor in view of being a rational selection in maximizing influence problem.
Effect based on the time in information propagation, industry propose a new influence propagation model LAIC (Latency
Aware Independent Cascade), which is added to time factor by using influence force delay distribution function sharp
In probability living, the influence force delay distribution function which considers is Poisson distribution and geometry distribution.However true social
It is much distributed in network and all meets power law distribution, including time factor, for example, the admission ticket of concert is before opening concert
It can quickly be sold, but after concert, which just loses its validity.Therefore, it is necessary to be examined in LAIC models
Consider power law distribution, i.e., message can be by fast propagation within a bit of time.
Invention content
The technical problem to be solved by the present invention is to：In the case where ensureing rational operational efficiency, for most of existing influences
Power maximizes algorithm and is limited in problem in static network, proposes a kind of maximizing influence analysis towards dynamic social networks
Method by effect of the analysis time factor in influence power propagation, and power law distribution is applied in influence power propagation model,
So that final result is more accurate and reliable.
A kind of maximizing influence analysis method GLAIC towards dynamic social networks, 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) activation probability is obtained：When towards dynamic social networks, by using influence force delay distribution functionIt will
Time factor is added to activation probability, then the activation probability isWherein puvInitial activation probability is represented, is generally set
For 1/in degree (v), δtRepresenting influences force delay, can be obtained from influence force delay distribution function, used in the present invention
Influence force delay distribution function be power law distribution；
(3) influence power propagation model is established：Shadow is carried out using LAIC models and the activation probability obtained in step (2)
Ring the structure of power propagation model；
(4) greedy algorithm is executed：Influence power is carried out using original greedy algorithm and the LAIC models obtained in step (3)
Propagation；
(5) CELF algorithm optimizations are used：Operational efficiency is carried out to greedy algorithm used in step (4) using CELF algorithms
Optimization.
Further, the power law distribution described in step (2) is a kind of consistent with true social network node degree distribution
Function, and the maximizing influence of dynamic social networks of the present invention also complies with this rule, i.e., within a bit of time
A certain message can fast propagation.
More specifically, the specific communication process of the LAIC models described in step (3) is as follows：
(a) assume that the node in network is divided into three kinds of states：It is active, postpone active and an inactive state.Wherein enliven shape
State refers to user and receives some information, and it is not to receive at once for the message of propagation that delay active state, which refers to user, can be passed through
One time delay, can just receive, and an inactive state refers to user and has rejected some message；
(b) it carves at the beginning, each seed node u can be with probabilityGo to attempt activation its delay it is active or
The inactive neighbours v of person；
If (c) seed node can successfully influence its inactive neighbour, inactive neighbours can be changed into delay and live
Jump, and pass through δtAfterwards, delay live-vertex can be transformed into live-vertex；
(d) when an active node of delay is influenced by multiple neighbor nodes, activationary time can update earliest
That time being affected, influence of remaining neighbour to it ignore that；
(e) this process recycles always progress, until not new active or delay live-vertex occurs.
More specifically, the concrete operation step of the greedy algorithm described in step (4) is as follows：
(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.
More specifically, the concrete operation step of the CELF algorithms described in step (5) is as follows：
(a) greedy algorithm is executed：Starting round, the marginal benefit of each node is calculated using original greedy algorithm；
(b) Priority Queues is created：According to the marginal benefit of each node, a preferential team is added in each node in descending order
Arrange Q；
(c) seed node is selected：According to the submodule characteristic for influencing force function, when round below selects seed node, no
Need to calculate the influence power of all nodes, it is only necessary to judge whether the round of head of the queue element is equal with current round, if equal,
Seed node is added and updates the marginal benefit of the element, and descending sort is re-started to Priority Queues Q if unequal；
(d) in each round below, repeat step (c), until selection meets k node of target.
The beneficial effects of the invention are as follows：A kind of maximizing influence analysis method towards dynamic social networks,
Consider effect of the time in information propagation, power law distribution is applied in LAIC models first, is secondly held on LAIC models
The original greedy algorithm of row, finally optimizes original greedy algorithm using CELF algorithms.Pass through such method so that the present invention
Finally outstanding achievement and rational run time are achieved on choosing most influential TOP-K nodes；Meanwhile the present invention
Used power law distribution is a kind of to be distributed in during influence power is propagated more effectively side than other distributions such as Poisson distribution and geometry
Method.
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 maximizing influence analysis method towards dynamic social networks of the present invention；
Fig. 2 is the comparison figure that power law distribution is distributed in Poisson distribution, geometry on Oregon data sets；
Fig. 3 is the comparison figure that power law distribution is distributed in Poisson distribution, geometry on CA_HepPh data sets；
Fig. 4 is the comparison figure that power law distribution is distributed in Poisson distribution, geometry on Email data sets；
Fig. 5 is the comparison figure that power law distribution is distributed in Poisson distribution, geometry on Slashdot data sets；
Fig. 6 is the comparison figure that power law distribution is distributed in Poisson distribution, geometry on Web_Stanford data sets；
Fig. 7 is the present invention and coverage effect contrast figure of the existing algorithm on Oregon data sets；
Fig. 8 is the present invention and coverage effect contrast figure of the existing algorithm on CA_HepPh data sets；
Fig. 9 is the present invention and coverage effect contrast figure of the existing algorithm on Email data sets；
Figure 10 is the present invention and coverage effect contrast figure of the existing algorithm on Slashdot data sets；
Figure 11 is the present invention and coverage effect contrast figure of the existing algorithm on Web_Stanford data sets；
Figure 12 is the present invention and existing algorithm run time comparison diagram on five data sets；
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 maximizing influence analysis method towards dynamic social networks of the present invention as shown in Figure 1, it is specific to walk
It is rapid as follows：
Step 1：Build social network diagram：G=(V, E), wherein G represent social networks, and V represents node set, and E represents net
The line set of network.
Step 2：Obtain activation probability.
When in face of dynamic social networks, by using influence force delay distribution functionTime factor is added to sharp
Probability living, then the activation probability beWherein puvInitial activation probability is represented, 1/in deg ree are generally set to
(v), δtRepresenting influences force delay, can be obtained from influence force delay distribution function, the influence force delay used in the present invention
Distribution function is power law distribution.The principle of power-law distribution described in the step is：Power law distribution is a kind of and true social network
Network node degree is distributed consistent function, and the maximizing influence of dynamic social networks of the present invention also complies with this rule,
A certain message can fast propagation i.e. within a bit of time.
Step 3：Establish influence power propagation model.
The structure of influence power propagation model is carried out using LAIC models and the activation probability obtained in step 2.The step
The specific communication process of LAIC models described in rapid is as follows：
(3a) assumes that the node in network is divided into three kinds of states：It is active, postpone active and an inactive state.Wherein enliven shape
State refers to user and receives some information, and it is not to receive at once for the message of propagation that delay active state, which refers to user, can be passed through
One time delay, can just receive, and an inactive state refers to user and has rejected some message.
(3b) is carved at the beginning, each seed node u can be with probabilityIt goes to attempt to activate its delay active
Or inactive neighbours v.
(3c) if seed node can successfully influence its inactive neighbour, inactive neighbours can be changed into delay and live
Jump, and pass through δtAfterwards, delay live-vertex can be transformed into live-vertex.
(3d) when an active node of delay is influenced by multiple neighbor nodes, activationary time can update earliest
That time being affected, influence of remaining neighbour to it ignore that.
(3e) this process recycles always progress, until not new active or delay live-vertex occurs.
Step 4：Execute greedy algorithm.
The propagation of influence power is carried out using original greedy algorithm and the LAIC models obtained in step 3.Institute in the step
The greedy algorithm stated is 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.
(4c) selects seed node：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.
Step 5：Use CELF algorithm optimizations.
Operational efficiency optimization is carried out to greedy algorithm used in step 4 using CELF algorithms.Described in the step
CELF algorithms are as follows：
(5a) executes greedy algorithm：Starting round, the marginal benefit of each node is calculated using original greedy algorithm.
(5b) creates Priority Queues：According to the marginal benefit of each node, each node is added one preferentially in descending order
Queue Q.
(5c) selects seed node：According to the submodule characteristic for influencing force function, when round below selects seed node, no
Need to calculate the influence power of all nodes, it is only necessary to judge whether the round of head of the queue element is equal with current round, if equal,
Seed node is added and updates the marginal benefit of the element, and descending sort is re-started to Priority Queues Q if unequal.
(5d) repeats step (5c) in each round below, until selection meets k node of target.
Embodiment：
One, data set and experimental setup
In this embodiment, using the disclosed data set Oregon data sets of five different scales from SNAP, CA_
HepPh data sets, Email data sets, Slashdot data sets and Web_Stanford data sets.Oregon data sets are one
The autonomous system for including 9 figures is a non-directed graph.CA_HepPh data sets come from high-energy physics theory partner networks,
It is a non-directed graph.Email data sets come from safe email communication networks, about 500,000 envelope electronics of the network coverage
The communication of all electronics in mail is a non-directed graph.Slashdot data sets come from a relevant News Network of technology
It stands, the linking relationship between user is by following relationship to be formed between friend, therefore is a digraph.Web_
Stanford data sets come from Stanford University website, the node on behalf website in network, and the hyperlink between website is formed
Oriented link, therefore be also a digraph.The static structure characteristic statistics of this five data sets are as shown in table 1.
Table 1：Experimental data static structure characteristic statistics
The parameter setting used in the present invention for influencing the influence delay distribution that delay is distributed and alignment algorithm uses is such as
Under：
Power law distribution：The parameter of power law distribution is and k-αDirectly proportional, α is the random selection from { 1,2,3,4 }.
Poisson distribution：The parameter of Poisson distribution is the random selection from set { 1,2,3..., 10 }.
Geometry is distributed：The parameter of geometry distribution is to pass throughIt generates.
All emulation experiments are all using following algorithm and GLAIC of the present invention (Greedy based in following embodiment
On LAIC) it makes comparisons：
ISP with Possion：The algorithm assumes that a node is passed merely by unique propagation path that influences
It broadcasts, the influence force delay distribution function used is Poisson distribution.
ISP with Geometric：The algorithm principle is similar with " ISP with Possion " algorithm, but use
It is geometry distribution to influence force delay distribution function.
MISP with Possion：The algorithm is the optimization algorithm of ISP algorithms, in order to make operational efficiency higher, the algorithm
The collective effect of the node of propagation and the activation probability of its out-degree neighbours is considered, the influence force delay which uses is distributed letter
Number is Poisson distribution.
MISP with Geometric：The algorithm is similar to " MISP with Possion " algorithm, but what it was used
Force delay distribution function is influenced to be distributed for geometry.
MIAM and MIAC：The two algorithms are proposed according to biggest impact sub-tree structure, and the influence power used is prolonged
Slow distribution function only has geometry distribution.
Degree Discount：The algorithm is a degree heuritic approach, and activation probability is usually using 0.01.
Random：The algorithm is that seed node collection is added in random selection node, is a base of maximizing influence algorithm
It is accurate.
Two, power law distribution effect
It is that Poisson distribution and geometry are distributed due to influencing force delay distribution function used in ISP and MISP algorithms,
In order to verify the effect of power law distribution used in the present invention in influence power propagation, we utilize ISP and MISP algorithm knots
It closes power law distribution and produces two new algorithms respectively, " ISP with Power-Law " and " MISP with Power-Law ",
And with " ISP with Possion ", " ISP with Geometric ", " MISP with Possion " and " MISP with
These four algorithms of Geometric " compare.Fig. 2 to Fig. 6 respectively shows seed node scale, and in 1-50, these three are different
The comparison diagram being distributed on five different data collection.
Can be seen that " ISP with Power-Law " by this five figures, algorithm behaves oneself best, and " MISP with
The performance of Power-Law " algorithms is similar to " ISP with Power-Law " algorithm, for example, on CA_HepPh data sets,
The coverage of " MISP with Power-Law " algorithm is only lower by 11.2% than " ISP with Power-Law " algorithm.To the greatest extent
Pipe is in this way, the two algorithms are but more much higher than the coverage of remaining algorithm.Specifically, such as one 2,000,000
Data set Web_Stanford, the coverage ratio " ISP with Possion " of " ISP with Power-Law " and " ISP
With Geometric " difference high 178.9% and 381.5%.And the coverage ratio of " MISP with Power-Law "
" MISP with Possion " and " MISP with Geometric " difference high 163.7% and 363.4%.This is primarily due to
" ISP with Power-Law " and " MISP with Power-Law " all considers power law distribution, and the distribution is compared to Poisson
Distribution and geometry distribution more meet the distribution of true social networks, to effectively expand the coverage of influence power propagation.
Three, coverage
Fig. 7 to Figure 11 respectively shows seed node scale GLAIC and other eight kinds of algorithm " ISP of the present invention in 1-50
With Possion ", " ISP with Geometric ", " MISP with Possion ", " MISP with Geometric ",
" MIAM ", " MIAC ", " Degree Discount " and " Random " coverage comparison diagram on five data sets.From this five
A figure can be seen that behaving oneself best for GLAIC algorithms of the present invention, for example, on Email data sets, the influence model of GLAIC algorithms
It encloses respectively than remaining algorithm " ISP with Possion ", " ISP with Geometric ", MISP with Possion ",
" MISP with Geometric ", " MIAM ", " MIAC ", " Degree Discount " and " Random " are high by 112.8%,
636.1%, 124.5%, 651.5%, 618.7%, 638.4%, 1575.9% and 3776.8%.This is primarily due to the present invention
Power law distribution and greedy algorithm are combined, and CELF algorithms has been used to optimize.On the one hand, greedy algorithm is opened compared to other
Hairdo algorithm such as " Degree Discount " has higher coverage, and on the other hand, the present invention is the influence power used
It is power law distribution to postpone distribution function, by second part, the present invention have been verified that power law distribution compared to Poisson distribution and
Geometry, which is distributed in during influence power is propagated, has better effect.
Four, run time
Figure 12 illustrates GLAIC of the present invention and other other eight kinds of algorithms " ISP with Possion ", " ISP with
Geometric ", " MISP with Possion ", " MISP with Geometric ", " MIAM ", " MIAC ", " Degree
The run time comparison diagram of Discount " and " Random " when seed node scale is 50.It can be seen from the figure that GLAIC
It is an algorithm based on greed, is required for recalculating a marginal benefit before selecting seed node due to every time, i.e.,
Make that present invention uses CELF algorithms to optimize, but GLAIC is still than other some algorithms, especially heuritic approaches
Operational efficiency it is relatively lower, operational efficiency of the GLAIC in small data set is outstanding, such as in Oregon data sets,
CA_HepPh data sets, the run time on Email data sets is 32s, 48s and 358s respectively, this demonstrates GLAIC and is applied to
Excellent properties when small data set.
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 (5)
1. a kind of maximizing influence analysis method towards dynamic social networks, 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) activation probability is obtained：When towards dynamic social networks, by using influence force delay distribution functionBy the time because
Element is added to activation probability, then the activation probability isWherein puvInitial activation probability is represented, is set as 1/
Indegree (v), δ t, which are represented, influences force delay, and from influencing to obtain in force delay distribution function, influencing force delay distribution function is
Power law distribution；
(3) influence power propagation model is established：Influence power is carried out using LAIC models and the activation probability obtained in step (2)
The structure of propagation model；
(4) greedy algorithm is executed：The biography of influence power is carried out using original greedy algorithm and the LAIC models obtained in step (3)
It broadcasts；
(5) CELF algorithm optimizations are used：It is excellent that operational efficiency is carried out to greedy algorithm used in step (4) using CELF algorithms
Change.
2. a kind of maximizing influence analysis method towards dynamic social networks according to claim 1, feature exist
It is a kind of to be distributed consistent function with true social network node degree in, the power law distribution described in step (2).
3. a kind of maximizing influence analysis method towards dynamic social networks according to claim 1, feature exist
In the specific communication process of the LAIC models described in step (3) is as follows：
(a) assume that the node in network is divided into three kinds of states：Active, delay is active and an inactive state, wherein active state refer to
User receives some information, and it is not to receive at once for the message of propagation that delay active state, which refers to user, can pass through one
Time delay can just receive, and an inactive state refers to user and has rejected some message；
(b) it carves at the beginning, each seed node u can be with probabilityIt goes to attempt to activate its delay active or non-
Active neighbours v；
If (c) seed node can successfully influence its inactive neighbour, inactive neighbours can be changed into delay actively, and
And pass through δtAfterwards, delay live-vertex can be transformed into live-vertex；
(d) when an active node of delay is influenced by multiple neighbor nodes, activationary time can update earliest by shadow
That loud time, influence of remaining neighbour to it ignore that；
(e) this process recycles always progress, until not new active or delay live-vertex occurs.
4. a kind of maximizing influence analysis method towards dynamic social networks according to claim 1, feature exist
In the concrete operation step of the greedy algorithm described in step (4) is as follows：
(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.
5. a kind of maximizing influence analysis method towards dynamic social networks according to claim 1, feature exist
In the concrete operation step of the CELF algorithms described in step (5) is as follows：
(a) greedy algorithm is executed：Starting round, the marginal benefit of each node is calculated using original greedy algorithm；
(b) Priority Queues is created：According to the marginal benefit of each node, a Priority Queues Q is added in each node in descending order；
(c) seed node is selected：It is not needed according to the submodule characteristic for influencing force function when round below selects seed node
Calculate the influence power of all nodes, it is only necessary to judge whether the round of head of the queue element is equal with current round, if equal, is added
Seed node updates the marginal benefit of the element, and re-start descending sort to Priority Queues Q if unequal；
(d) in each round below, repeat step (c), until selection meets k node of target.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810269151.5A CN108510115A (en) | 2018-03-29 | 2018-03-29 | A kind of maximizing influence analysis method towards dynamic social networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810269151.5A CN108510115A (en) | 2018-03-29 | 2018-03-29 | A kind of maximizing influence analysis method towards dynamic social networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108510115A true CN108510115A (en) | 2018-09-07 |
Family
ID=63379146
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810269151.5A Pending CN108510115A (en) | 2018-03-29 | 2018-03-29 | A kind of maximizing influence analysis method towards dynamic social networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108510115A (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109560966A (en) * | 2018-12-03 | 2019-04-02 | 中国科学院信息工程研究所 | A kind of cascade propagation early detection method based on social sensor optimization |
| 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 |
| CN109741198A (en) * | 2018-11-28 | 2019-05-10 | 中国科学院计算技术研究所 | Spreading network information influence power measure, system and maximizing influence method |
| 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 |
| CN110972227A (en) * | 2019-11-19 | 2020-04-07 | 三峡大学 | Seed node selection method for offloading cellular traffic through opportunistic mobile networks |
| CN111339444A (en) * | 2020-03-11 | 2020-06-26 | 嘉兴学院 | Method for solving social network influence problem based on coulomb force model |
| CN111445291A (en) * | 2020-04-01 | 2020-07-24 | 电子科技大学 | Method for providing dynamic decision for social network influence maximization problem |
| CN112101996A (en) * | 2020-09-14 | 2020-12-18 | 哈尔滨工程大学 | Social network dynamic influence maximization method based on region |
| CN112100241A (en) * | 2020-09-14 | 2020-12-18 | 哈尔滨工程大学 | Social network dynamic influence maximization method based on theme |
| 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 |
| CN112818245A (en) * | 2021-02-25 | 2021-05-18 | 上海大学 | Social network influence maximization method based on Gaussian propagation model |
| CN113222774A (en) * | 2021-04-19 | 2021-08-06 | 浙江大学 | Social network seed user selection method and device, electronic equipment and storage medium |
| CN113378470A (en) * | 2021-06-22 | 2021-09-10 | 常熟理工学院 | Time sequence network-oriented influence maximization method and system |
| WO2021184367A1 (en) * | 2020-03-20 | 2021-09-23 | 清华大学 | Social network graph generation method based on degree distribution generation model |
| 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 (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104616200A (en) * | 2015-02-11 | 2015-05-13 | 中南大学 | Influence maximization initial node selecting method based on node characteristics |
| CN105913287A (en) * | 2016-05-20 | 2016-08-31 | 重庆大学 | Influence maximization method based on community structure |
-
2018
- 2018-03-29 CN CN201810269151.5A patent/CN108510115A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104616200A (en) * | 2015-02-11 | 2015-05-13 | 中南大学 | Influence maximization initial node selecting method based on node characteristics |
| CN105913287A (en) * | 2016-05-20 | 2016-08-31 | 重庆大学 | Influence maximization method based on community structure |
Non-Patent Citations (3)
| Title |
|---|
| ANDREW T. STEPHEN等: "Explaining the power-law degree distribution in a social commerce network", 《SOCIAL NETWORKS》 * |
| BO LIU等: "Influence Spreading Path and Its Application to the Time Constrained Social Influence Maximization Problem and Beyond", 《IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING》 * |
| JURE LESKOVEC等: "Cost-effective Outbreak Detection in Networks", 《PROCEEDINGS OF THE 13TH ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING》 * |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109741198A (en) * | 2018-11-28 | 2019-05-10 | 中国科学院计算技术研究所 | Spreading network information influence power measure, system and maximizing influence method |
| CN109560966A (en) * | 2018-12-03 | 2019-04-02 | 中国科学院信息工程研究所 | A kind of cascade propagation early detection method based on social sensor optimization |
| CN109560966B (en) * | 2018-12-03 | 2020-05-22 | 中国科学院信息工程研究所 | Network cascade propagation early discovery method based on social sensor optimization |
| 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 |
| 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 |
| CN110972227A (en) * | 2019-11-19 | 2020-04-07 | 三峡大学 | Seed node selection method for offloading cellular traffic through opportunistic mobile networks |
| CN111339444A (en) * | 2020-03-11 | 2020-06-26 | 嘉兴学院 | Method for solving social network influence problem based on coulomb force model |
| CN111339444B (en) * | 2020-03-11 | 2022-10-04 | 嘉兴学院 | Method for solving social network influence problem based on coulomb force model |
| WO2021184367A1 (en) * | 2020-03-20 | 2021-09-23 | 清华大学 | Social network graph generation method based on degree distribution generation model |
| CN111445291A (en) * | 2020-04-01 | 2020-07-24 | 电子科技大学 | Method for providing dynamic decision for social network influence maximization problem |
| CN111445291B (en) * | 2020-04-01 | 2022-05-13 | 电子科技大学 | Method for providing dynamic decision for social network influence maximization problem |
| CN112101996A (en) * | 2020-09-14 | 2020-12-18 | 哈尔滨工程大学 | Social network dynamic influence maximization method based on region |
| CN112100241A (en) * | 2020-09-14 | 2020-12-18 | 哈尔滨工程大学 | Social network dynamic influence maximization method based on theme |
| CN112101996B (en) * | 2020-09-14 | 2023-10-17 | 哈尔滨工程大学 | Region-based social network dynamic influence maximization method |
| CN112100241B (en) * | 2020-09-14 | 2023-07-25 | 哈尔滨工程大学 | Social network dynamic influence maximization method based on theme |
| 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 |
| CN112818245A (en) * | 2021-02-25 | 2021-05-18 | 上海大学 | Social network influence maximization method based on Gaussian propagation model |
| CN113222774A (en) * | 2021-04-19 | 2021-08-06 | 浙江大学 | Social network seed user selection method and device, electronic equipment and storage medium |
| CN113222774B (en) * | 2021-04-19 | 2023-05-23 | 浙江大学 | Social network seed user selection method and device, electronic equipment and storage medium |
| CN113378470A (en) * | 2021-06-22 | 2021-09-10 | 常熟理工学院 | Time sequence network-oriented influence maximization method and system |
| 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 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108510115A (en) | A kind of maximizing influence analysis method towards dynamic social networks | |
| CN108492201A (en) | A kind of social network influence power maximization approach based on community structure | |
| CN108737191A (en) | The topology control method based on unsupervised learning towards super-intensive wireless sensor network | |
| CN109921939B (en) | Method and system for selecting key nodes in communication network | |
| FINDIK | Bull optimization algorithm based on genetic operators for continuous optimization problems. | |
| Han et al. | Identifying top-k influential nodes based on discrete particle swarm optimization with local neighborhood degree centrality | |
| Tabatabaei | A new routing protocol for energy optimization in mobile ad hoc networks using the cuckoo optimization and the TOPSIS multi-criteria algorithm | |
| Chen et al. | Trapping on modular scale-free and small-world networks with multiple hubs | |
| Anwar et al. | Balanced influence maximization in the presence of homophily | |
| Jian-Guo et al. | Multistage random growing small-world networks with power-law degree distribution | |
| Gaye et al. | Spanning graph for maximizing the influence spread in Social Networks | |
| Türker et al. | Spreading in scale-free computer networks with improved clustering | |
| CN108415774A (en) | A kind of Method for HW/SW partitioning based on improvement fireworks algorithm | |
| Yu et al. | Collective learning and information diffusion for efficient emergence of social norms | |
| Ghonaim et al. | An improvement of chaos-based hash function in cryptanalysis approach: An experience with chaotic neural networks and semi-collision attack | |
| Wu et al. | Scale-free properties of information flux networks in genetic algorithms | |
| Karkooki et al. | A genetic algorithm to improve lifetime of wireless sensor networks by load balancing | |
| Hongyong et al. | Consensus of synchronization-preferential scale-free networks | |
| Bin et al. | Maximizing social influence in nearly optimal time: SRIS model | |
| Fan et al. | An individual-based mean-field model for fake-news spreading on PSO-based networks | |
| Aswanth et al. | Performance study of bio-inspired approach to clustering in wireless sensor networks | |
| Karczmarczyk et al. | Influencing information spreading processes in complex networks with probability spraying | |
| Ruan et al. | Effect of indirect social ties on cascading diffusion of information | |
| Feng et al. | Weights adaptation optimization of heterogeneous epidemic spreading networks: A constrained cooperative coevolution strategy | |
| Ally et al. | Effects of Network Structures and Fermi Function's Parameter ß in Promoting Information Spreading on Dynamic 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 |