CN105138580B - A kind of network negative information influence minimum method based on the company of blocking side - Google Patents
A kind of network negative information influence minimum method based on the company of blocking side Download PDFInfo
- Publication number
- CN105138580B CN105138580B CN201510462971.2A CN201510462971A CN105138580B CN 105138580 B CN105138580 B CN 105138580B CN 201510462971 A CN201510462971 A CN 201510462971A CN 105138580 B CN105138580 B CN 105138580B
- Authority
- CN
- China
- Prior art keywords
- minimum
- negative information
- information
- greedy algorithm
- digraph
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a kind of network negative informations based on the company of blocking side to influence minimum method.This method uses digraph to indicate the propagation of information in social networks first, and finds k side in the digraph using greedy algorithm, so that the infection area of negative information is minimum when removing the k side, wherein k is positive integer;Then the k side is cut off so that the range minimum that negative information is propagated.The present invention is searched out by greedy algorithm can be by the smallest k side of fallacious message range of scatter, the outlying total number of edges much smaller than social network diagram of this k item.The social networks that the present invention can break out fallacious message is effectively controlled, and substantially reduces the spread scope of fallacious message, and the greedy algorithm proposed is closest to theoretical optimal solution, far better than other heuritic approaches.
Description
Technical field
The invention belongs to information technologies, social networks technical field, and in particular to a kind of network based on the company of blocking side is negative
Face informational influence minimizes method.
Background technique
In the past few decades, online social networks is the platform that information is propagated and the marketing activity provides convenience, and is allowed
Idea and the behavior cascade in social networks, which pass on from one to another, to be propagated.From the point of view of the function of social networks, it can not only propagate front and disappear
Breath, such as:Idea is innovated, hot topic etc. can also propagate negative news, such as:Malicious rumor, deceptive information etc..Just
As an example by rumour, even if most starting only seldom infected person, but due to triggering a series of cascade in a network
Structure, final infection number also can be very big.Therefore, effective method how is designed, to reduce the coverage of negative information,
The influence minimum for making it is a Scientific Research Problem urgently to be resolved.
On how to find the maximum point of influence power, propagate information more effectively in social networks, this problem quilt
Referred to as maximizing influence problem had attracted many concerns in recent years.Then in contrast how to make negative information
The smallest influence minimization problem is spread, is seldom paid close attention to, important is studied a question although this is also one.
The problem of minimizing is influenced about negative information, has had a few thing to complete.There is work sutdy before
Reduce the method for diffusion area by removal node.Document " [1] Albert, R., Jeong, H., and Barab_asi, A.-
L.:Error and attack tolerance of complex networks.In Nature,378-382,2000.[2]
Newman,M.E.J.,Forrest,S.,and Balthrop,J.:Email networks and the spread of
computer viruses.In Physical Review E,66:035101.[3]Wang,S.,Zhao,X.,Chen,Y.,
Li,Z.,Zhang,K.,and Xia,J.:Negative Influence Minimizing by Blocking Nodes in
Social Networks.In AAAI (Late-Breaking Developments) 2013. " has been proven that by by node
Go out descending arrangement, so that the node for removing front is usually very effective.Here, removal node just contains removal in fact
The case where side.So going the task of flash trimming more basic than removal node, the propagation of negative information is prevented by removal even side
Range is a very important thing.Kimura proposes a kind of method of trimming, to make the contaminated area of whole network most
It is small, referring to document " Kimura, M., Saito, K., and Motoda, H.:Minimizing the Spread of
Contamination by Blocking Links in a Network.In AAAI 2008.".However, his method does not have
In view of how the network for having infected is handled.Yu thinks (referring to document " Yu, Y., Berger-Wolf, T.Y., and
Saia,J.:Finding spread blockers in dynamic networks.In Advances in Social
Network Mining and Analysis, 55-76,2010. ") for finding most effective transmission blockage node, only
Look for those node degrees very high.Budak have studied influence minimize the problem of (referring to document " Budak, C.,
Agrawal,D.,and Abbadi,A.E.:Limiting the spread of misinformation in social
Networks.In WWW 2011. "), this problem is defined as a fallacious message in social networks from some specific by he
Node is propagated, and the influence of negative information some nodes is then selected and offset with some positive information.Different from before
Work, research of the invention focus on one by rumour infect social networks, in this case, research how to lead to
It crosses and removes sub-fraction side to make negative information influence to minimize.
Summary of the invention
The present invention in view of the above-mentioned problems, provide it is a kind of based on block even side network negative information influence minimum method,
The social networks that can have been broken out for fallacious message is effectively controlled, and drops the spread scope of fallacious message significantly
It is low.
Network negative information based on the company of blocking side of the invention influences minimum method, indicates society using digraph first
The propagation of information in network is handed over, and finds k side in the digraph using greedy algorithm, so that being born when removing the k side
The infection area of face information is minimum, and wherein k is positive integer;Then the k side is cut off so that the range minimum that negative information is propagated
(number of nodes of infection is minimum).
Key problem in technology point of the invention is:
1) social networks broken out for fallacious message, can be carried out and efficiently control;
2) greedy algorithm of accuracy guarantee is proposed, the available guarantee of the accuracy of result is made;
3) connecting side by cutting off k item blocks fallacious message to propagate, and k is far smaller than total number of edges, but fallacious message
Spread scope but substantially reduces.
Using method provided by the invention when social networks carries out fallacious message control, have the following advantages that:
The social networks that the present invention has been broken out primarily directed to fallacious message, carrying out excision, even side is maliciously believed to block
Cease a kind of method propagated.By greedy algorithm, the present invention search out can by the smallest k side of fallacious message range of scatter,
They are cut off, and the outlying total number of edges much smaller than social network diagram of this k item.This greedy algorithm has accuracy guarantee, f
(D)/f(D*) >=1-1/e, D are the solution of greedy algorithm output, D* representation theory optimal solution.For this problem, the present invention is proposed
Greedy algorithm be closest to theoretical optimal solution, far better than other heuritic approaches.
Detailed description of the invention
Fig. 1 is the step flow chart of the method for the present invention.
Fig. 2 (a)~Fig. 2 (d) be greedy algorithm of the invention from existing method in different data collection and different probability of spreading
Under negative information range of scatter comparison diagram.
Fig. 3 is the runing time comparison diagram of algorithms of different.
Specific embodiment
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, below by specific embodiment and
Attached drawing, the present invention will be further described.
The social networks that it is an object of the present invention to infect for one cuts a certain number of even sides, makes last
It infects area to minimize, as shown in Figure 1.In more detail, it when infecting since a part of start node, is widely used at one
Elementary probability model --- propagated under IC model (Independent Cascade, independent cascade model), it is contemplated that finding
Include k side inside one set, when this k side is removed, it is contemplated that infection area can be it is minimum, k is one given
Positive integer.We are referred to as this combinatorial optimization problem:Negatively affect minimization problem.For this problem, we have proposed
One has the greedy algorithm of accuracy guarantee, to efficiently find an optimal solution.By to two extensive true social networks
Data set tested (including Facebook and Diggers), we demonstrate the performances of greedy algorithm proposed by the present invention
It is better than two good heuristic trimming algorithms (centrad and out-degree in-degree algorithm) being studied.
The spread scopes such as negative information such as computer virus and malicious rumor are minimized this problem with one by the present invention
Digraph G=(V, E) is indicated.Herein, V and E respectively represents the set on all the points and side.IC model is applied to network by we
In the communication process of middle fallacious message, the problem of influencing minimum is then being explored on G.
We, which provide negative information, below influences the mathematical definition minimized.Assuming that negative information passes in figure G=(V, E)
It broadcasts, initializing infections node set isOur target is the propagation model for making negative information by the D set being breaking in E
Minimum is enclosed, has k side in this D set, k (< < | E |) it is a given constant.This can be expressed as following optimal
Change problem:
Wherein, σ (S | E D) is indicated after line set D is cut off, the final infected number of S set interior joint.
Herein, it is proposed that one is solved based on the greedy algorithm of maximum marginal benefit rule in G=(V, E) influence
The problem of minimum.Enabling k is the number on the side to be removed in this problem.In order to prove its validity, we are with two
Classical is compared based on the heuritic approach of impact evaluation.
In order to better use greedy algorithm (Greedy Algorithm), we by formula (1) replace with one it is equal
Optimization problem, it is as follows:
Herein,
f(D):=σ (S | E)-σ (S | E D) (3)
It is defined as:When primary infection collection is combined into S, after removing line set D, the diffusion of reduction.This formula (3) has
The property of submodule characteristic can prove below.
Theorem 1:Reduce spread function
By this property, the set D that set sizes are k is found with greedy algorithm, so that f (D) is maximized, Ke Yiyong
The expression of algorithm 1, the Algorithm 1 seen below.A new side e* is selected to this algorithm iteration to make the incrementss of f (D) most
Greatly, then this side is added in D set, until the size of set D is equal to k.E indicates currently processed in algorithm below
Side.We are it can be proved that this algorithm has accuracy guarantee, f (D)/f (D*) >=1-1/e, D are the solutions of greedy algorithm output,
D* representation theory optimal solution.
It is respectively illustrated in Fig. 2 (a)~Fig. 2 (d) result, greedy algorithm (Greedy) is in two datasets
(Facebook and Diggers), in the case where different probability of spreading p, the effect shown will be better than centrad algorithm
(Betweenness) and out-degree algorithm (Out-degree).Centrad algorithm is number two, and out-degree algorithm ranking is last.Make
For comparison, the performance of centrad and out-degree is very approached.Wherein, Fig. 2 (a) figure and Fig. 2 (b) figure are Diggers data set, figure
The probability of spreading p of 2 (a) figures is that the probability of spreading p of 0.1, Fig. 2 (b) figure is 0.05;Fig. 2 (c) figure and Fig. 2 (d) figure are Facebook
Data set, the probability of spreading p of Fig. 2 (c) figure are that the probability of spreading p of 0.1, Fig. 2 (d) figure is 0.05;In four width figures | S | i.e. starting sense
Contaminating number of nodes is 50.
In Fig. 2 (a), we can observe that, greedy algorithm will be felt by 50 sides of cutting in Diggers data set
The node of dye is reduced to 80 from 118.Here 50 cut off are in the while collection for meaning that excision is entirely connected with infection node
8.59% side in conjunction.By cutting off the side in the connected line set of infection node 8.59%, greedy algorithm, heuristic center
Algorithm is spent, heuristic out-degree algorithm can reduce by 32%, 19% respectively, 15% infection area.Other three lab diagrams are also similarly.
In addition, we conclude that, method proposed by the present invention shows more preferably in sparse network, and what is showed in dense network is not
It is very satisfactory.
Fig. 3 illustrates the runing time comparison of three kinds of algorithms, and abscissa indicates that different data collection, ordinate are operation in figure
Time (Running time).As can be seen that the runing time of centrad heuritic approach and out-degree heuritic approach is than this
The algorithm time of invention wants short.
The above embodiments are merely illustrative of the technical solutions of the present invention rather than is limited, the ordinary skill of this field
Personnel can be with modification or equivalent replacement of the technical solution of the present invention are made, without departing from the spirit and scope of the present invention, this
The protection scope of invention should be subject to described in claims.
Claims (3)
1. a kind of network negative information based on the company of blocking side influences minimum method, which is characterized in that use digraph first
It indicates the propagation of information in social networks, and finds k side in the digraph using greedy algorithm, so that when removing the k item
The infection area of negative information is minimum when side, and wherein k is positive integer;Then the k side is cut off so that the model that negative information is propagated
Enclose minimum;
The method that k side in digraph is wherein found using greedy algorithm is:
1) it sets negative information to propagate in digraph G=(V, E), V and E respectively represent all nodes and the set of Lian Bian, originate
Infecting node set isTarget is to keep the spread scope of negative information minimum by the D set being breaking in E, D set
In have k side, by the problem representation at following optimization problem:
Wherein, σ (S | E D) is indicated after line set D is cut off, the final infected number of S set interior joint;
2) the step 1) optimization problem is replaced with to following equal optimization problem:
Wherein, f (D):=σ (S | E)-σ (S | E D), it is defined as reducing after removal even line set D as primary infection set S
Diffusion;
3) the set D that set sizes are k is found with greedy algorithm, so that f (D) be made to maximize.
2. the method as described in claim 1, which is characterized in that the step 3) iteratively selects a new side e* to make f (D)
Incrementss it is maximum, then this side is added in set D, until the size of set D is equal to k.
3. method according to claim 1 or 2, it is characterised in that:Information in the social networks, which is propagated, uses separate stage
Gang mould type.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510462971.2A CN105138580B (en) | 2015-07-31 | 2015-07-31 | A kind of network negative information influence minimum method based on the company of blocking side |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510462971.2A CN105138580B (en) | 2015-07-31 | 2015-07-31 | A kind of network negative information influence minimum method based on the company of blocking side |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105138580A CN105138580A (en) | 2015-12-09 |
CN105138580B true CN105138580B (en) | 2018-11-23 |
Family
ID=54723929
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510462971.2A Active CN105138580B (en) | 2015-07-31 | 2015-07-31 | A kind of network negative information influence minimum method based on the company of blocking side |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105138580B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110111340A (en) * | 2019-04-28 | 2019-08-09 | 南开大学 | The Weakly supervised example dividing method cut based on multichannel |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106789962B (en) * | 2016-12-02 | 2019-07-16 | 浙江大学 | A kind of network Pollution restraint method based on the crash time |
CN107220486B (en) * | 2017-05-12 | 2021-07-20 | 上海交通大学 | Influence blocking maximization method based on local influence calculation |
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 |
CN110046224B (en) * | 2019-04-15 | 2023-05-09 | 哈尔滨工程大学 | Social network rumor inhibition method based on region |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859315A (en) * | 2010-04-30 | 2010-10-13 | 西北工业大学 | Heuristic solving method for maximizing influence of social network |
CN102571954A (en) * | 2011-12-02 | 2012-07-11 | 北京航空航天大学 | Complex network clustering method based on key influence of nodes |
CN103116611A (en) * | 2013-01-25 | 2013-05-22 | 西安市烟草专卖局 | Social network opinion leader identification method |
-
2015
- 2015-07-31 CN CN201510462971.2A patent/CN105138580B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859315A (en) * | 2010-04-30 | 2010-10-13 | 西北工业大学 | Heuristic solving method for maximizing influence of social network |
CN102571954A (en) * | 2011-12-02 | 2012-07-11 | 北京航空航天大学 | Complex network clustering method based on key influence of nodes |
CN103116611A (en) * | 2013-01-25 | 2013-05-22 | 西安市烟草专卖局 | Social network opinion leader identification method |
Non-Patent Citations (1)
Title |
---|
Minimizing the spread of contamination by blocking links in a network;Masahiro Kimura等;《Proceeding AAAI"08 Proceedings of the 23rd national conference on Artificial intelligence》;20080717;第2卷;第1175-1176页 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110111340A (en) * | 2019-04-28 | 2019-08-09 | 南开大学 | The Weakly supervised example dividing method cut based on multichannel |
CN110111340B (en) * | 2019-04-28 | 2021-05-14 | 南开大学 | Weak supervision example segmentation method based on multi-path segmentation |
Also Published As
Publication number | Publication date |
---|---|
CN105138580A (en) | 2015-12-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105138580B (en) | A kind of network negative information influence minimum method based on the company of blocking side | |
CN105468681B (en) | A kind of network negative information influence minimum method based on topic model | |
Zhu et al. | Information source detection in the SIR model: A sample-path-based approach | |
Milani Fard et al. | Neighborhood randomization for link privacy in social network analysis | |
Husari et al. | Using entropy and mutual information to extract threat actions from cyber threat intelligence | |
US20150047040A1 (en) | Cognitive information security using a behavioral recognition system | |
Yang et al. | Transportation dynamics on networks of mobile agents | |
Zhu et al. | Locating the contagion source in networks with partial timestamps | |
US20130086004A1 (en) | Updating a perfect hash data structure, such as a multi-dimensional perfect hash data structure, used for high-speed string matching | |
Ventresca et al. | Network robustness versus multi-strategy sequential attack | |
Gan et al. | Analysis of computer virus propagation behaviors over complex networks: a case study of Oregon routing network | |
Yang et al. | The effect of network topology on the spread of computer viruses: a modelling study | |
Chumachenko et al. | On intelligent agent-based simulation of network worms propagation | |
Abbasi et al. | Deep learning-based feature extraction and optimizing pattern matching for intrusion detection using finite state machine | |
Yuan et al. | Towards controlling virus propagation in information systems with point-to-group information sharing | |
Szynkiewicz et al. | Design and evaluation of a system for network threat signatures generation | |
Šimon et al. | Combined heuristic attack strategy on complex networks | |
Omić et al. | Network protection against worms and cascading failures using modularity partitioning | |
Zhong et al. | RankAOH: Context-driven similarity-based retrieval of experiences in cyber analysis | |
Wang et al. | Anonymization for multiple released social network graphs | |
Jiang et al. | Target privacy preserving for social networks | |
Enyioha et al. | Bio-inspired strategy for control of viral spreading in networks | |
Sood et al. | On the minimum node degree and k-connectivity in inhomogeneous random K-out graphs | |
Wickramaarachchi et al. | Distributed exact subgraph matching in small diameter dynamic graphs | |
Essouifi et al. | Discrete stochastic modelling of computer viruses prevalence on a reduced scale-free network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |