CN109740024A - Solution method for large-scale timing diagram influence maximization problem - Google Patents
Solution method for large-scale timing diagram influence maximization problem Download PDFInfo
- Publication number
- CN109740024A CN109740024A CN201910014844.4A CN201910014844A CN109740024A CN 109740024 A CN109740024 A CN 109740024A CN 201910014844 A CN201910014844 A CN 201910014844A CN 109740024 A CN109740024 A CN 109740024A
- Authority
- CN
- China
- Prior art keywords
- node
- seed
- probability
- timing diagram
- act
- 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
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to a solution to the problem of maximizing influence of a large-scale time chart, which adopts a time chart GT(V,E,TE) The method is characterized in that data of each node of the social network are abstracted into a time sequence diagram, propagation probability among the nodes is initialized, the method is suitable for an ICT propagation model of the time sequence diagram, and influence is exerted on each node on the basis of the ICT propagation modelThe problem of maximizing the influence of the timing graph is solved according to the influence set of each node calculated in step 3, namely, a seed node set with the size of k is searched. By adopting the method, the problem of maximizing the influence of the timing diagram is solved, and the problem of maximizing the influence of the timing diagram can be quickly and efficiently solved.
Description
Technical field
The present invention relates to one kind towards extensive timing diagram maximizing influence way to solve the problem, belongs to database
Processing technology field.
Background technique
Nowadays, online social networks such as microblogging, wechat and blog etc. plays the part of important role in people's lives.
People are expressed their idea by social networks, share news and information.People are in social networks by propagating theirs
Idea and information influence the other users in network, design one kind by this " public praise (word-of-mouth) effect "
New grass roots marketing techniques, referred to as " virus marketing " (viral marketing).This marketing mode is honest and clean by information by the public
Valence duplication, tells and gives other audients, so that the influence for expanding rapidly oneself is compared with traditional marketing mode, audient is voluntary
The characteristics of receiving makes that cost is less, income is more.
On the other hand, the development of internet is but also the vulgar cultural prevalence of network rumour, network plays the work added fuel to the flames
With.And a good information can be not only made to spread through the internet out as soon as possible the research of maximizing influence problem
It goes, it is also possible to inhibit propagation of the rumour on network by means of the research of maximizing influence problem.
The research of general maximizing influence is all based on static map, i.e., social networks is abstracted as static map, base
Refer in the maximizing influence problem of static map by finding k node in a network as seed node, so that information exists
Other users are influenced by the way that k user is as much as possible in a network under specific propagation model (such as IC propagation model), i.e.,
It will affect power maximization problems to study as a discrete optimization problem.
But real-life many networks, and cannot be simply abstracted as static map, for example, person to person it
Between telephone network, the transmission of mutual mail, transportation network and cranial nerve network etc., in these networks, between node not
Can only exist in some period and contact to beginning to can all there is connection eventually, is i.e. connection between node is with timing
Property.Due to the timing contacted between node, the IC model based on static map is also resulted in, timing diagram can not be suitable for, because
This solves the problems, such as the first step of timing diagram maximizing influence, it is necessary first to improve, make it possible to applicable to IC model
It in timing diagram, then just can further study, and solve the problems, such as timing diagram maximizing influence.
Summary of the invention
(1) technical problems to be solved
In order to solve the above problem of the prior art, the present invention provides a kind of influence power towards extensive timing diagram most
Bigization way to solve the problem.
(2) technical solution
In order to achieve the above object, the main technical schemes that the present invention uses include:
A kind of maximizing influence way to solve the problem towards extensive timing diagram, includes the following steps:
It S1, by the data abstraction of each node of social networks is timing diagram, the slip chart is shown as GT(V,E,TE),
In, the V indicates the node set in social networks, and that the E is indicated is the set on timing side in network, the TEIt indicates
Each node contacts moment set；
S2, the probability of spreading node is initialized, later, weight is carried out to traditional independent cascade propagation model
Newly design the ICT propagation model based on the timing diagram；
S3, on the basis of the ICT propagation model, to each node carry out influence powerCalculating,
The influence power set for each node that S4, foundation calculate in step s3 solves timing diagram maximizing influence and asks
The seed node set that topic, i.e. searching size are k.
Method as described above, it is preferable that the step S2 specifically includes as follows:
S201, between the probability of spreading p each nodeu,vIt is initialized, the probability of spreading node is carried out random
Assignment；
S202, the ICT propagation model include: that all node v are assigned a value of Actv=-1 indicates that all nodes are all located
In an inactive state, its information communication process is as follows after choosing a seed node u:
S221. the active time Act of seed node uu=0, seed node u is at this time with Probability pu,vActivate its neighbor node
V, and node u has and only has an opportunity can activate node v；
S222. node u first determines whether Act when attempting to activate node vuWhether max (T is less than or equal to(u,v)), if greatly
Begin trying to activate next neighbor node in then directly skipping, if it is less than being equal to, then node u is just with Probability pu,vActivation section
Point v；
S223. no matter whether node u can activate node v, and u will not reactivation node v in later bout；
Once S224. node v is successfully activated, records it and enliven initial time Actv=t(u,v), wherein t(u,v)∈
T(u,v), and Actu≤t(u,v)≤max(T(u,v))；
S225. information is attempted to pass in entire social networks from new live-vertex in an inactive state neighbor node
Broadcast is gone, until not new node is activated.
Method as described above, it is preferable that influence power is carried out to each node in the step S3Calculating, tool
Steps are as follows for body:
S301, in timing graph data structure, to the neighbor node of each node according to max (T(u,v)) carry out by greatly to
Small sequence；
One S302, setting queue Q, and initial Q is sky, then chooses the influence power of being calculatedDestination node
Destination node u is put into queue Q by u；
S303, head of the queue node u is taken out from queue；
S304, a several p, and p ∈ [0,1] are generated at random, its neighbor node v is successively traversed, if p >=pu,vAnd
Actu≤max(T(u,v)), then illustrate that node u can activate its neighbor node v, then enables the active time Act of node vv←min
(t|t∈T(u,v)∩t≥Actu)；
S305, the node being activated in step 3.3 is put into setIn and queue Q in；
S306, in the case where queue Q is not sky, repeat step S303-S305；
S307, the influence power set for obtaining node u
Method as described above, it is preferable that the step S4 includes the following steps:
S401, setting S are seed node set, and initialization S is sky, determine that the size for finally needing obtained set S is
K, calculates the edge effect of all nodes, and is expressed as infs (u)；
S402, all non-seed nodes are ranked up according to the size delta infs (u) of edge effect, are sorted
As a result u1,u2,…,un, choose u1For seed node, i.e., by u1Into set S；
S403, the u calculated in step S4022New edge effect size delta infs (u2), if node u2It is new
Edge effect size be more than or equal to u3Edge effect size, then directly choose node u2Set is put into for seed node
In S, otherwise executing step S402；
S404, set of computations seed node set S size, if the size of set S be less than k, then follow the steps
S402-S403；
S405, result seed node set S is obtained, the node in set S is the seed node found.
Preferably, the probability of spreading p_ (u, v) is that live-vertex u passes through its neighbor node of side (u, v) successful activation v
Probability, p_ (u, the v) ∈ [0,1].
(3) beneficial effect
The beneficial effects of the present invention are:
Maximizing influence way to solve the problem provided by the invention towards extensive timing diagram, can quickly and
Efficiently solve the problems, such as timing diagram maximizing influence.And traditional maximizing influence algorithm can not all solve timing diagram shadow
Ring power maximization problems.The present invention carries out products propaganda in it can be used to timing social networks, is used by choosing important node
Family can save promotion costs, meanwhile, present invention can conceivably be used to refute a rumour to some malicious rumors on network with
And the historical data by analyzing propagation of the previous epidemic disease in each area carries out disease prevention etc..
Detailed description of the invention
Fig. 1 is the method flow diagram that the present invention is embodied；
Fig. 2 is the structural schematic diagram of the timing diagram of present invention specific implementation 2；
Fig. 3 is that seed node chooses schematic diagram in the embodiment of the present invention 2；
Fig. 4 is the operational effect comparison diagram of the embodiment of the present invention 2 and the prior art.
Specific embodiment
In order to preferably explain the present invention, in order to understand, with reference to the accompanying drawing, by specific embodiment, to this hair
It is bright to be described in detail.
Symbol used in the present invention and meaning are shown in Table 1.
The symbol of the invention of table 1 and meaning
The definition of " timing diagram " in the present invention: given network GT(V,E,TE) node contacts are expressed as with sequential relationship
The oriented timing diagram of social networks, V indicate the set of node, and E indicates the set on side, and | V |=n, | E |=m.TEIt indicates in figure
There is the set at connection moment, T between all nodes(u,v)Indicate set at the time of there is connection between node u and v,
T(u,v)∈TE。
The definition of " probability of spreading " in the present invention: live-vertex u is general by its neighbor node of side (u, v) successful activation v's
Rate, probability of spreading are expressed as pu,v∈[0,1]。
The definition of " node enlivens initial time " in the present invention: at the time of node v is enlivened father node u successful activation by it
For its initial time that jumps, it is expressed as Actv, and Actv=min t | (t ∈ T(u,v)∩ t≥Actu)}。
The definition of " node influence power " in the present invention: node influence power is in a network can be by node u successful activation
The set of node, is expressed as
The definition of " edge effect " in the present invention: S, the edge effect of non-seed node v are set by seed node set
For
Wherein Δ infs (v) indicates the size of edge effect.
The definition of " timing diagram maximizing influence " in the present invention: given timing diagram GT(V,E,TE) and specific propagation
Model finds a node set S in timing diagram, wherein set S contains node number | S |=k, and so that the influence of set S
PowerIt maximizes.Set S i.e. GTSeed node set.
Embodiment 1
A kind of maximizing influence way to solve the problem towards extensive timing diagram, as shown in Figure 1, including as follows
Step:
Step 1: using timing diagram GT(V,E,TE) come indicate by the data abstraction of each node of social networks be timing diagram,
Wherein V indicates the node set in social networks, and that E is indicated is the set on timing side in network, TEThe each node connection indicated
It is the set at moment.The data structure of timing diagram is only needed in the weight on side using the data structure as adjacency list
Save all connected moment between node and the probability of spreading between node.
Step 2: the probability of spreading node being initialized first, then to traditional independent cascade propagation model
New ICT propagation model is redesigned out to make it possible to suitable for timing diagram, it is specific as follows；
Step 2.1: initialization probability of spreading；
Firstly the need of between the probability of spreading p each nodeu,vIt is initialized, the probability of spreading of node is to timing diagram shadow
The algorithm realization for ringing power maximization problems will not serve conclusive, all be between the probability of spreading node under normal conditions
Carry out random assignment.Such as between the probability of spreading p each nodeu,vRandom assignment is 0.01,0.1,0.2,0.3,0.5.
Step 2.2: redesigning independent cascade propagation model；
The maximizing influence algorithm of static map saves in timing diagram without considering the initial time that node is activated
The initial time that point is successfully activated is in need of consideration.It is a kind of new by improving to have obtained to IC propagation model
The propagation model based on timing diagram, ICT (IC model on Temporal Graph) propagation model.
Information is made in timing diagram by the communication process of ICT propagation model now and being discussed in detail.
It, can be by the Act of all node v inside the network of most initialv=-1 indicates all nodes all in non-live
Jump state.Its information communication process is as follows after choosing a seed node u:
1. the active time Act of seed node uu=0, seed node u is at this time with Probability pu,vIts neighbor node v is activated,
And node u has and only has an opportunity can activate its neighbor node v.
2. node u first determines whether Act when attempting and activating neighbor node vuWhether max (T is less than or equal to(u,v)), such as
Fruit, which is greater than then directly to skip, begins trying to activate next neighbor node, and if it is less than being equal to, then node u is just with Probability pu,vSwash
Movable joint point v.
3. no matter whether node u can activate node v, u will not go to attempt activation node again in later bout
v。
4. once recording it node v is successfully activated and enlivening initial time Actv=t(u,v), wherein t(u,v)∈T(u,v),
And Actu≤t(u,v)≤max(T(u,v))。
5. information is propagated out in entire social networks from new live-vertex in the trial of an inactive state neighbor node
It goes, until not new node is activated.
Step 3: influence power being carried out to each node on the basis of ICT propagation modelCalculating, specific steps
It is as follows；
Step 3.1: in timing graph data structure, to the neighbor node of each node according to max (T(u,v)) carry out by big
To small sequence；
Step 3.2: one queue Q of setting, and initial Q is sky, then chooses the influence power of being calculatedTarget
Node u is put into queue Q by node u；
Step 3.3: head of the queue node u is taken out from queue；
Step 3.4: generating a several p, and p ∈ [0,1] at random, its neighbor node v is successively traversed, if p >=pu,vAnd
And Actu≤max(T(u,v)), then illustrate that node u can activate its neighbor node v, then enables the active time Act of node vv←
min(t|t∈T(u,v)∩t≥Actu)；
Step 3.5: the node being activated in step 3.3 is put into setIn and queue Q in；
Step 3.6: in the case where queue Q is not sky, repeating step 3.3-3.5；
Step 3.7: obtaining the influence power set of node u
Step 4: the influence power set according to each node calculated in step 3 solves timing diagram maximizing influence
The seed node set that problem, i.e. searching size are k；
Step 4.1: setting S is seed node set, and initialization S is sky, determines that the set S's for finally needing to obtain is big
Small is k, calculates the edge effect infs (u) of all nodes；
Step 4.2: all non-seed nodes being ranked up according to the size delta infs (u) of edge effect, are arranged
Sequence result u1,u2,…,un, choose u1For seed node, i.e., by u1Into set S；
Step 4.3: calculating the u in step 4.22New edge effect size delta infs (u2), if node u2's
The size of new edge effect is more than or equal to u3Edge effect size, then directly choose node u2Collection is put into for seed node
It closes in S, otherwise, executes step 4.2；
Step 4.4: the size of the set S of set of computations seed node executes step if the size of set S is less than k
Rapid 4.2-4.3；
Step 4.5: obtaining result seed node set S, the node in set S is the seed node found.
Embodiment 2
The present embodiment combination Fig. 2 and Fig. 3 are illustrated timing diagram and related notion.As shown in Figure 2 is timing diagram,
Node number is 7, V={ Cor, An, Cor, Nan, Li, Wu, Lyd }, and static side is 13, and timing side is 28, with side
For (Cor, An), this is a static side, indicates that node Cor and node An has connection, but when two nodes are not
Moment, which carves, all has connection, and the weight 1,3 and 5 on side indicates only two nodes ability when 1,3 and 5 moment of moment
There are associated relations, i.e. T(Cor,An)={ 1,3,5 }.And the probability of spreading p between nodeu,vRandom assignment is 0.01,0.1,0.2,
0.3,0.5.If node Cor is to the probability of spreading p of node AnCor,An=0.2, it indicates either to pass at the moment 1,3 or 5
Probability is broadcast all to be consistent.
Experimental data is using the data on the sequential network on heap exchange website Super User, including 100k node
290k item static state while and when 530k timing.In ICT propagation model, need to carry out assignment, experiment to probability of spreading node
Middle setting Making by Probability Sets P1=(0.01,0.1,0.2,0.3,0.5) randomly selects the probability in set in experiment to section respectively
Probability of spreading between point carries out assignment.It is chosen in tri- kinds of algorithms of BIMT, IMIT and AMIT in seed node, selected seed node
Set sizes are respectively 1,10,20,30,40 and 50.
For the method for the operation present embodiment of extensive timing diagram, process is as shown in Figure 1, include the following steps:
Step 1: being timing diagram by the upper sequential network data abstraction of website, wherein the node in timing diagram indicates net
There are two class weights on user in network, the side on timing diagram, and a kind of set existed at the time of connection between node is another kind of
Indicate the probability of spreading of progress random assignment；
Step 2: redesigning independent cascade propagation model is new propagation model ICT propagation model；
It, can be by the Act of all node v inside the network of most initialv=-1 indicates all nodes all in non-live
Jump state.Its information communication process is as follows after choosing a seed node u:
1. the active time Act of seed node uu=0, seed node u is at this time with Probability pu,vIts neighbor node v is activated,
And node u has and only has an opportunity can activate node v.
2. node u first determines whether Act when attempting and activating node vuWhether max (T is less than or equal to(u,v)), if greatly
Begin trying to activate next neighbor node in then directly skipping, if it is less than being equal to, then node u is just with Probability pu,vActivation section
Point v.
3. no matter whether node u can activate node v, u will not go to attempt activation node again in later bout
v。
4. once recording it node v is successfully activated and enlivening initial time Actv=t(u,v), wherein t(u,v)∈T(u,v),
And Actu≤t(u,v)≤max(T(u,v))。
5. information is propagated out in entire social networks from new live-vertex in the trial of an inactive state neighbor node
It goes, until not new node is activated.
Step 3: influence power being carried out to each node on the basis of ICT propagation modelCalculating, be with Fig. 1
Example, specific step is as follows；
Step 3.1: in Fig. 1, to the neighbor node of each node according to max (T(u,v)) carry out descending sequence；
Step 3.2: one queue Q of setting, and initial Q is sky, then chooses the influence power of being calculatedTarget
Node u is put into queue Q by node u, by taking node Cor as an example, i.e., node Cor is put into queue Q；
Step 3.3: head of the queue node Cor is taken out from queue；
Step 3.4: generate a several p, and p ∈ [0,1] at random, successively traverse its neighbor node v (node Nan, Han with
And An), if p >=pu,vAnd Actu≤max(T(u,v)), then illustrate that node u can activate its neighbor node v, then enables node v
Active time Actv←min(t|t∈ T(u,v)∩t≥Actu), it is assumed that Cor successful activation node Han of node, then ActHan
=9；
Step 3.5: the node Han being activated in step 3.3 is put into setIn and queue Q in
It goes；
Step 3.6: in the case where queue Q is not sky, repeating step 3.3-3.5；
Step 3.7: obtaining the influence power set of node u
Step 4: the influence power set according to each node calculated in step 3 solves timing diagram maximizing influence
The seed node set that problem, i.e. searching size are k, equally by taking Fig. 1 as an example, specific step is as follows；
Step 4.1: mapping relations one by one are made to all nodes and number 0-N, vertex Cor, Han ..., Lyd it is right respectively
Answer number 1,2 ..., 8.By startingIt is expressed as Wherein 0 number is equal to 8.,
Step 4.2: all non-seed nodes being ranked up according to the size delta infs (u) of edge effect, are arranged
Sequence result Cor, Nan, Han ..., Lyd, selection Cor are seed node, i.e., node Cor are put into set S, because of section
The node that point Cor can influence respectively correspond with number 1,3,5,7,8, so only need byIn the 1,3,5,7th and 8
Element becomes corresponding number from 0, i.e.,；
Step 4.3: at this point, not needing to carry out all nodes in new limit first to find next seed node
The calculating of effect.But the new edge effect of calculate node Nan, the influence node of node Nan are 2,3,4 and 6 first, now
Without by arrivingIn successively inquire to calculate the new edge effect of vertex Nan, it is only necessary to successively judge
In the 2nd, 3,4 and 6 element whether be 0.So, it is only necessary to which the new limit effect of vertex Nan can be obtained in 4 calculating
It answers.
Step 4.4: the new edge effect size delta infs (Nan) of the Nan in step 4.3 is calculated, if node
The size of the new edge effect of Nan is more than or equal to the edge effect size of Han, then directly choosing node Nan is seed node
It is put into set S, otherwise executing step 4.2；
Step 4.5: the size of the set S of set of computations seed node executes step if the size of set S is less than k
Rapid 4.2-4.4；
Step 4.6: obtaining result seed node set S, the node in set S is the seed node found.
The specific embodiment of the invention is by experiment test proposition towards extensive timing diagram maximizing influence problem
Solution.In the experiment, AMIT represents basic solution of the invention, and IMIT indicates optimization node edge effect meter
A kind of method calculated to each node in network and establishes one-to-one relationship between number, reaches quick access with this
The unknown purpose of node, BIMT indicates a kind of can solve extensive timing diagram maximizing influence on the basis of IMIT method
The method of problem reaches by reducing the computing repeatedly for edge effect of some inessential nodes and quickly finds seed section
The purpose of point.
By the method for the present invention (AMIT, IMIT, BIMT) and existing method (Random, Degree, DegreeSingle,
DegreeDiscount carry out operational effect comparison) crosses and experimental result data is automatically generated to acquisition such as Fig. 4 on Excle
Shown result.The small figure of each of Fig. 4 represents a number according to experimental results, and wherein data is provide in Stanford University
Existing timing social network data collection, the data set 1 in Fig. 4 (1) is the mail network data set of an European mechanism, Fig. 4
(2) data set 2 in is the online social network data collection of students of Stanford University, and the data set 3 in Fig. 4 (3) is
Storehouse exchanges the time Internet data set generated on website Math Overflow, and the data set 4 in Fig. 4 (4) is
Website Super User sequential network figure generated is exchanged by website storehouse.
7 curves in each small figure represent experiment effect of 7 kinds of solutions in four on data set, wherein three kinds
It is method proposed by the present invention (BIMT, IMIT and AMIT), in addition four kinds (Random, Degree, DegreeSingle,
DegreeDiscount the Rnadom method in) is the selected seed node by way of randomly selecting, in addition three kinds
(Degree, DegreeSingle, DegreeDiscount) is the heuritic approach of the degree based on node, by ignoring section
The timing of point carries out selected seed node.But the shortcomings that these four methods is not can guarantee node selection correct
Property, and the accuracy rate for being to guarantee selection result a little of the invention.Abscissa in figure indicates selected seed node
Number, ordinate indicate that the node number that can be activated compares shared ratio with user node sum.It can by Experimental comparison
With discovery, three kinds of solutions proposed by the invention are substantially better than other four kinds of methods deposited, that is, are choosing identical seed
In the case where node, method proposed by the present invention can activate more user nodes on four kinds of data sets.
The above described is only a preferred embodiment of the present invention, be not the limitation that other forms are done to the present invention,
Anyone skilled in the art can use the equivalent reality that technology contents disclosed above were changed or be modified as equivalent variations
Apply example.But without departing from the technical solutions of the present invention, to the above embodiments according to the technical essence of the invention
Any simple modification, equivalent variations and remodeling, still fall within the protection scope of technical solution of the present invention.
Claims (6)
1. a kind of maximizing influence way to solve the problem towards extensive timing diagram, which is characterized in that it includes as follows
Step:
It S1, by the data abstraction of each node of social networks is timing diagram, the slip chart is shown as GT(V, E, TE), wherein institute
The node set in V expression social networks is stated, that the E is indicated is the set on timing side in network, the TEWhat is indicated is each
The set at node contacts moment；
S2, the probability of spreading node is initialized, later, traditional independent cascade propagation model is redesigned
Out based on the ICT propagation model of the timing diagram；
S3, on the basis of the ICT propagation model, to each node carry out influence powerCalculating,
The influence power set for each node that S4, foundation calculate in step s3, solves the problems, such as timing diagram maximizing influence, i.e.,
Find the seed node set that size is k.
2. solution as described in claim 1, which is characterized in that in the step S2, the propagation between node
Probability is initialized as between the probability of spreading p each nodeU, vInitialized, between node probability of spreading carry out with
Machine assignment.
3. solution as described in claim 1, which is characterized in that in step s 2, the ICT propagation model include: by
All node v are assigned a value of Actv=-1 indicates all nodes all in an inactive state, chooses after a seed node u it
Information communication process is as follows:
S221. the active time Act of seed node uu=0, seed node u is at this time with Probability pU, vIts neighbor node v is activated, and is saved
Point u has and only has an opportunity and can activate node v；
S222. node u first determines whether Act when attempting to activate node vuWhether max (T is less than or equal to(u, v)), if it is greater than then
It directly skips and begins trying to activate next neighbor node, if it is less than being equal to, then node u is just with Probability pU, vActivate node v；
S223. no matter whether node u can activate node v, and u will not reactivation node v in later bout；
Once S224. node v is successfully activated, records it and enliven initial time Actv=t(u, v), wherein t(u, v)∈T(u, v), and
Actu≤t(u, v)≤max(T(u, v))；
S225. information is propagated out in entire social networks from new live-vertex in the trial of an inactive state neighbor node
It goes, until not new node is activated.
4. solution as described in claim 1, which is characterized in that carry out influence power to each node in the step S3Calculating, the specific steps are as follows:
S301, in timing graph data structure, to the neighbor node of each node according to max (T(u, v)) carry out descending row
Sequence；
One S302, setting queue Q, and initial Q is sky, chooses the influence power of being calculatedDestination node u, by target
Node u is put into queue Q；S303, head of the queue node u is taken out from queue；
S304, a several p, and p ∈ [0,1] are generated at random, its neighbor node v is successively traversed, if p >=pU, vAnd Actu≤
max(T(u, v)), then node u activates its neighbor node v, then enables the active time Act of node vv←min(t|t∈T(u, v)∩t≥
Actu)；
S305, the node being activated in step S303 is put into setIn and queue Q in；
S306, in the case where queue Q is not sky, repeat step S303-S305；
S307, the influence power set for obtaining node u
5. solution as described in claim 1, which is characterized in that the step S4 includes the following steps:
S401, setting S are seed node set, and initialization S is sky, determine that the size for the set S for finally needing to obtain is k, meter
Calculate the edge effect infs (u) of all nodes；
S402, all non-seed nodes are ranked up according to the size delta infs (u) of edge effect, obtain ranking results u1,
u2.., un, choose u1For seed node, i.e., by u1Into set S；
S403, the u calculated in step S4022New edge effect size delta infs (u2), if node u2New limit
The size of effect is more than or equal to u3Edge effect size, then directly choose node u2It is put into set S for seed node,
Otherwise executing step S402；
S404, set of computations seed node set S size, if the size of set S be less than k, then follow the steps S402-
S403；
S405, result seed node set S is obtained, the node in set S is the seed node found.
6. solution as described in claim 1, which is characterized in that the probability of spreading pU, vPass through side for live-vertex u
The probability of its neighbor node of (u, v) successful activation v, the pU, v∈ [0,1].
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910014844.4A CN109740024A (en) | 2019-01-08 | 2019-01-08 | Solution method for large-scale timing diagram influence maximization problem |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910014844.4A CN109740024A (en) | 2019-01-08 | 2019-01-08 | Solution method for large-scale timing diagram influence maximization problem |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109740024A true CN109740024A (en) | 2019-05-10 |
Family
ID=66363886
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910014844.4A Pending CN109740024A (en) | 2019-01-08 | 2019-01-08 | Solution method for large-scale timing diagram influence maximization problem |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109740024A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112446634A (en) * | 2020-12-03 | 2021-03-05 | 兰州大学 | Method and system for detecting influence maximization node in social network |
| 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 |
-
2019
- 2019-01-08 CN CN201910014844.4A patent/CN109740024A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112446634A (en) * | 2020-12-03 | 2021-03-05 | 兰州大学 | Method and system for detecting influence maximization node in social network |
| CN112446634B (en) * | 2020-12-03 | 2021-08-06 | 兰州大学 | Method and system for detecting influence maximization node in social network |
| 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 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kumar et al. | Identifying influential nodes in Social Networks: Neighborhood Coreness based voting approach | |
| Li et al. | Labeled influence maximization in social networks for target marketing | |
| Pathak et al. | A generalized linear threshold model for multiple cascades | |
| CN103064917B (en) | The high-impact customer group of a kind of specific tendency towards microblogging finds method | |
| CN100504856C (en) | Method and apparatus for characterizing documents based on clusters of related words | |
| CN109740024A (en) | Solution method for large-scale timing diagram influence maximization problem | |
| CN111523046A (en) | Cooperative evolution model construction method based on public opinion information and viewpoint of evolutionary game | |
| CN113254719A (en) | Online social network information propagation method based on status theory | |
| CN107807935A (en) | Using recommendation method and device | |
| Hajibagheri et al. | Social networks community detection using the shapley value | |
| Wei et al. | Cooperation dynamics on collaborative social networks of heterogeneous population | |
| Von Elm et al. | Age at first marriage in Peninsular Malaysia | |
| Jiang et al. | Dynamic community detection based on game theory in social networks | |
| CN103093049A (en) | Prediction method and prediction system for malicious code propagation facing social network | |
| Tan et al. | AIM: Activation increment minimization strategy for preventing bad information diffusion in OSNs | |
| CN107688964A (en) | A kind of motivational techniques for the diffusion of extensive mass-rent task | |
| Zheng et al. | Public opinion prediction model of food safety events network based on BP neural network | |
| Barabási | The science of networks | |
| CN111597023A (en) | Intelligent cluster scheduling method and device based on learning state | |
| Yu et al. | Analysis of competitive information dissemination in social network based on evolutionary game model | |
| Wang et al. | Diversity and efficiency | |
| CN109102061A (en) | A kind of group of subscribers behavior generation system meeting human dynamics rule, equipment and storage medium | |
| CN109472729A (en) | Online education big data technology platform | |
| Hui et al. | Agent-based Simulation of the Diffusion of Warnings | |
| Fu et al. | An evolving scale-free network with large clustering coefficient |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20220524 Address after: 100081 No. 5 South Main Street, Haidian District, Beijing, Zhongguancun Applicant after: BEIJING INSTITUTE OF TECHNOLOGY Address before: No. 195, Chuangxin Road, Hunnan District, Shenyang City, Liaoning Province Applicant before: Northeastern University |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190510 |