CN108520337A - A kind of rumour methods of risk assessment based on network risks entropy difference - Google Patents
A kind of rumour methods of risk assessment based on network risks entropy difference Download PDFInfo
- Publication number
- CN108520337A CN108520337A CN201810239750.2A CN201810239750A CN108520337A CN 108520337 A CN108520337 A CN 108520337A CN 201810239750 A CN201810239750 A CN 201810239750A CN 108520337 A CN108520337 A CN 108520337A
- Authority
- CN
- China
- Prior art keywords
- node
- network
- rumour
- entropy
- moment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- G06Q10/0635—Risk analysis of enterprise or organisation activities
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses the rumour methods of risk assessment based on network risks entropy difference, including：S1, the gossip propagation model based on SIR models is established to the network platform to be assessed；S2, the network structure based on gossip propagation model identify rumour source and propagation time using rumour source discrimination；S3, the forward-propagating process for simulating rumour in the network architecture, to obtain the probability that each node in current time network is in different conditions；S4, the probability that different conditions are according to current time each node calculate current time network greateset risk entropy and minimum risk entropy；S5, greateset risk entropy and minimum risk entropy using current time network, calculate the network risks entropy at current time；S6, according to the network risks entropy of network current time and previous moment calculate network risks entropy it is poor, and according to network risks entropy difference assess current time rumour potential risk caused by the network platform.The present invention can quantitatively, accurately in real time assessment rumour risk caused by network.
Description
Technical field
The present invention relates to internets and information dissemination technology field, more particularly, to a kind of net based on network risks entropy difference
The methods of risk assessment of network public sentiment, network rumour or network false information etc..
Background technology
With popularizing for internet, more and more people start with internet and work, do shopping, learn and entertain
Deng.According to the China Internet Network Information Center of August in 2017 4 days (CNNIC) the 40th time《China Internet network state of development statistics report
It accuses》It has been shown that, by June, 2017, Chinese netizen's scale reaches 7.51 hundred million, accounts for 1/5th of global netizen's sum.Network is
Medium and platform through obtaining information daily as people, expression viewpoint, mutually exchange.So easily the network platform is netizen
Having dredged one, each airs his own views, the free channel of exchange viewpoint, and the common people increasingly tend to deliver viewpoint public opinion on network.Net
The producer of network public sentiment content is transformed into website and the coefficient mixing of users via website making personnel
Group.This trend makes internet become the public opinion except the mass media such as newspaper, broadcast, TV and interpersonal communication
The mainstream media of propagation.The network media has broken many limitations of the traditional media to time, space as new media, widens
The extensiveness and intensiveness of propagation.By Internet, information can be traveled to the world incessantly in 24 hours by the network media
Each corner.The fast development of network has brought the convenient of life, but also gives rumour " having plugged wing " simultaneously.
The rudiment that information propagates at full speedization, netizen participates in anonymization, opinion expression is changeable in mood, the extramalization of position viewpoint is all network rumour
Hidden danger is buried.Netizen's scale for increasing rapidly, easily network access, mobile interchange intelligence and economic society transitional period are accumulated
Tired various social negative emotions be combined with each other, and condition is provided for the growth and sprawling of rumour.
Network rumour refers to the false speech having no factual evidence via network wide-scale distribution.Compared with traditional rumour,
Network rumour either spread speed or spread scope are obtained for great extension, are particularly susceptible with major event and dash forward
The generation of hair event and spread, while the popularity of internet is but also everybody is likely to become gossip propagation on network
Person, therefore network rumour is since occurring just tending to spread unchecked, and this trend for tending to spread unchecked of network rumour has seriously affected
The stabilization and harmony of society.The wide-scale distribution of network rumour easily upsets normal civil order, causes social trust crisis, invades
Violate citizen's personal right, therefore the method for studying network rumour risk assessment just seems and is even more important.
The relevant research of network rumour at present is also immature, and the rumour risk assessment scheme majority that researcher proposes is based on layer
The multiple criteria decision making (MCDM)s algorithm such as fractional analysis establishes evaluation index to rumour various aspects and carries out comprehensive marking and makes risk assessment.This
Class solution subjective factor is larger, it is difficult to collect data according to dynamic realtime is propagated, assessment result is not precisely timely enough.Therefore
There is an urgent need for a kind of schemes that can accurately carry out risk assessment to rumour in time for industry.
The disclosure of background above technology contents is only used for inventive concept and the technical solution that auxiliary understands the present invention, not
The prior art for necessarily belonging to present patent application, no tangible proof show the above present patent application the applying date
Before have disclosed in the case of, above-mentioned background technology should not be taken to evaluation the application novelty and creativeness.
Invention content
For the deficiency of existing rumour methods of risk assessment, the present invention proposes that one kind can monitor network speech in real time
(such as network rumour, public sentiment, deceptive information etc.) propagation condition, and the method for making risk quantification assessment in time.
The technical solution that the present invention is proposed in order to overcome the deficiencies of the prior art is as follows：
A kind of rumour methods of risk assessment based on network risks entropy difference, includes the following steps：
S1, to the network platform to be assessed, establish the gossip propagation model based on SIR models；
S2, the network structure based on the gossip propagation model, using rumour source discrimination identify rumour source and its
Propagation time；The network structure includes multiple network nodes, one network user of each node on behalf；
S3, the rumour source according to step S2 determinations and its propagation time, the forward direction of rumour is simulated in the network structure
Communication process, to obtain the probability that each node in network structure described in current time is in different conditions；Wherein, node
The state includes three kinds：Easy infection state, Infection Status and recovery state；
S4, the probability that different conditions are according to each node described in current time calculate network knot described in current time
The greateset risk entropy and minimum risk entropy of structure；
S5, the greateset risk entropy using network structure described in current time and the minimum risk entropy calculate current
The network risks entropy of network structure described in moment；
S6, the network risks according to the network risks entropy and the previous moment at current time at the network structure current time
Entropy, the network risks entropy for calculating the network structure is poor, and assesses current time rumour to institute according to the network risks entropy difference
Potential risk caused by stating the network platform.
The present invention is by studying the mechanism to spread rumors between the possible state of individual, individual during gossip propagation, base
In the factor of SIR models and influence gossip propagation, it is proposed that a kind of gossip propagation model can effectively be carved since rumour source
Draw the process that rumour is propagated in social networks；And can effectively it be reflected by adjusting probability of infection and recovery probability in model
Go out influence of the various factors to gossip propagation in reality, method proposed by the present invention is made to be more in line with reality, it is easily operated.So
Afterwards, by calculate network risks entropy difference analyse rumour risk, eliminate as in analytic hierarchy process (AHP) subjective factor to assessment result
It influences, keeps the result of rumour risk assessment more accurate.Meanwhile the method based on entropy can be in conjunction with the state of each node in network
The size for analyzing whole network rumour venture influence, can more comprehensively and objectively reflect the risk of network rumour.Meanwhile the present invention
The model that rumour risk is assessed based on network risks entropy difference proposed constructs a large amount of training set datas or characteristic without collecting
Library obtains different network risks entropys for different rumours, is not limited by model parameter, make the result of rumour risk assessment
It is more accurate.
Description of the drawings
Fig. 1 is the rumour methods of risk assessment flow chart proposed by the present invention based on network risks entropy difference；
Fig. 2 is the schematic network structure of the gossip propagation model of the specific embodiment of the invention；
Fig. 3 be the specific embodiment of the invention gossip propagation model in individual (node) state transformation schematic diagram；
Fig. 4 is the schematic diagram for finding rumour source in network structure shown in Fig. 2 using back propagation.
Specific implementation mode
The invention will be further described with specific embodiment below in conjunction with the accompanying drawings.
In order to quantitatively assess rumour risk, specific implementation mode of the invention proposes a kind of based on network risks entropy
The rumour methods of risk assessment of difference, initially sets up gossip propagation model, then determines rumour source, then simulate and pass from rumour source forward direction
Process is broadcast, finally the network entropy difference defined in communication process carrys out the risk of real-time quantization rumour.It is proposed by the present invention with reference to figure 1
The rumour methods of risk assessment includes the following steps S1 to S6：
Step S1, to the network platform to be assessed, the gossip propagation model based on SIR models is established.
The network platform such as can be with forward, comment on, thumb up mainstream function social network-i i-platform, and
And the network platform to be assessed is not limited to some platform in this method, can be the platform group that the platform of several interconnections is constituted
Body, it can be estimated that rumour real-time risk caused by institute's possibility in the network that these platforms are constituted.
A kind of exemplary network structure of the gossip propagation model based on SIR models is as shown in Fig. 2, the network structure
A social networks model can be regarded as, by many network nodes (blockage and small circular in figure are all network nodes,
Each one network user of node on behalf) and node between connecting relation constituted, there is connection pass between two nodes
System refer between two nodes can direct communication network information, indicated with line in the network structure.
It is that each node itself defines three according to SIR models (typical Epidemic Model) in the gossip propagation model
Kind state：Rumour is not received or receives rumour but does not believe that the easy infection state of rumour, is received rumour and is started to pass
The Infection Status for broadcasting rumour (i.e. it is believed that rumour) is heard and no longer believes that rumour (shows as stopping by believing that rumour is changed into after truth
Only spread rumors) recovery state.As shown in figure 3, the individual state based on SIR models changes schematic diagram, the individual in network
It before receiving rumour or receives rumour but does not believe that rumour all in easy infection state, any moment all may be with one
Determine probability α and is changed into Infection Status, and during being in Infection Status, also it can be changed into recovery state with certain probability β.Cause
This SIR model can effectively portray state transition process of each network user itself for rumour, in conjunction with as shown in Figure 2
Network structure, the dynamic process that rumour is propagated in a network can be portrayed.
Step S2, the network structure based on the gossip propagation model, rumour source is identified using rumour source discrimination
And its propagation time.Back propagation may be used in identification rumour source, can also use other typical rumour source discriminations
Such as Jordan center methods, effective distance method, Monte Carlo Method etc..Specific embodiments of the present invention are to use back propagation
For the process for finding rumour source illustrated.
Rumour can be propagated since any individual in network, and the rumour propagated since different individuals can pass through
Different paths generates different influences.Therefore, the source for finding gossip propagation is research gossip propagation process and understanding net
The primary work of each individual current state in network.The present invention is based on the method for backpropagation strategy find gossip propagation source,
With reference to figure 4, first, some nodes are chosen from the network structure as observer nodes, constitute observer nodes set S, and
The observer nodes being selected are indicated with " blockage " node in network structure as shown in Figure 2；Then, at a time t is carried out
Observation, the observer nodes that t moment is in Infection Status constitute setObserver nodes in easy infection state constitute setAnd set of records endsIn each observer nodes (assuming that t moment has the m observer nodes for being in Infection Status) infected by rumour
Moment T={ t1,t2,t3,…,tm}.To setIn each observer nodes, since infected node the latest, according to respectively quilt
The time sequencing of rumour infection can be aggregated simultaneously successively along the network path backpropagation rumour of the network structure
In all observer nodes infection node will be considered possible rumour source, be added in doubtful rumour source set U.Such as Fig. 4
It is shown, utilize setIn node carry out backpropagation demonstration, when initial (moment of τ=0) (such as infect from being infected the latest
Moment is in t moment) node A1 start backpropagation, the moment of τ=1 is reversed since t-1 moment infected node A2
It propagates, the backpropagation since t-2 moment infected node A3 of the moment of τ=2, and so on carry out backpropagation demonstration.
It can finally be aggregated simultaneouslyIn the node u of all observer nodes infection will be considered as possible rumour source, be added to described
In doubtful rumour source set U.Finally, to all node u in set U, the maximum likelihood value in t moment is sought respectively, wherein
The maximum node of maximum likelihood value is rumour source uf.Node u is in the maximum likelihood value of t moment
Wherein, PI(h,th；U) be node h in thThe rumour that the node u that moment is aggregated in U is transmitted infects and in t
Carve the probability in Infection Status, PS(k,t；U) the rumour infection transmitted for the node u that node k is not aggregated in t moment in U
And it is in easy infection shape probability of state, it can be calculated by aftermentioned formula (2) and (5)：
Wherein, α (h, th) indicate node h in thMoment infected probability, PS(h,th- 1) and PI(h,th- 1) section is indicated respectively
Point h is in (th- 1) moment is in the probability of easy infection state and Infection Status, βhIndicate that the node h in Infection Status is changed into
Restore shape probability of state, 0 ＜ βh＜ 1；UhIndicate credit value of the user representated by node h in the network platform, Ke Yicong
The network platform acquires；ChIndicate support of the user representated by node h in the network platform,phContain to hold in the content issued in the network platform for the user representated by node h and approves of attitude
Number of reviews, nhIt is commented containing what is held the attitude of objection in the content issued in the network platform for the user representated by node h
By quantity.And have：
LhFor liveness of the user representated by node h in a nearest predetermined amount of time, NhIt is node h in the network structure
In adjacent node j set, infjThe influence value for being the user representated by node j in the network platform (such as it is micro-
Rich bean vermicelli quantity), PI(j,th- 1) be node j in (th- 1) moment is in the probability of Infection Status；And have：
WithIndicate that the user representated by node h sends out in nearest one week and nearest January in the network platform respectively
The content quantity of cloth,WithIndicate the user representated by node h in nearest one week and nearest January in the net respectively
The quantity thumbed up on network platform,WithIndicate respectively the user representated by node h in nearest one week with nearest January in
The network platform last time reexamines the quantity of opinion.
PS(k,t；U)=(1- α (k, t)) PS(k,t-1) (5)
α (k, t) indicates node k in the infected probability of t moment, PS(k, t-1) indicates that node k is in susceptible at (t-1) moment
Contaminate shape probability of state.
Meanwhile for each may be rumour source node u, can gossip propagation corresponding by following formula to calculating when
Between：
Rumour source ufAnd its gossip propagation time tfFor：
Step S3, according to the rumour sources determined step S2 and its propagation time, rumour is simulated in the network structure
Forward-propagating process, to obtain the probability that each node in network structure described in current time is in different conditions.It is finding
Rumour source ufAnd extrapolate the propagation time t of rumourfAfterwards, we simulate the dynamic communication process of rumour using SIR models,
Initial time we by rumour source ufIt is defined as Infection Status, and other nodes in network are defined as easy infection state.This
Invention according to user characteristics individual in social networks, definition influence the infected probability of individual because be known as its nearest liveness,
The influence power of adjacent body, the factor for influencing to restore probability after individual is infected have user credit, User support degree.To in t
At the moment, the node i in easy infection state is by the probability that rumour infects in t moment in the network structure
Equally with aforementioned formula (3), in formula (8), LiIt is pre- in the network platform nearest one for the user representated by node i
The liveness fixed time in section, NiThe set for the adjacent node j for being node i in the network structure, infjFor node j institute's generations
Influence value of the user of table in the network platform, PI(j, t-1) is that node j is in Infection Status at (t-1) moment
Probability.Equally, LiCalculating use formula (4), substitute into corresponding parameter and seek.
Next, calculating the probability that each node is in different conditions in t moment.It is infected by rumour in t moment according to node i
Probability α (i, t), calculate node i is in the probability P of easy infection state in t momentS(i,t)：
PS(i, t)=(1- α (i, t)) PS(i,t-1) (9)
Node i is in the probability P of Infection Status in t momentI(i,t)：
And node i is in the probability P of recovery state in t momentR(i,t)：
βiIt indicates that the node i in Infection Status is changed into and restores shape probability of state, 0 ＜ βi＜ 1；UiIt indicates representated by node i
Credit value of the user in the network platform；CiIndicate support of the user representated by node i in the network platform,piContain to hold in the content issued in the network platform for the user representated by node i and approves of commenting for attitude
By quantity, niContain the comment held the attitude of objection in the content issued in the network platform for the user representated by node i
Quantity.
It is calculated after each node is in the probability of different conditions in network structure by aforementioned, present invention introduces network entropys
Concept carry out the size of rumour risk in quantitative analysis network.The concept of " entropy " was created in 1856 by physicist Clausius
Vertical, thermodynamics measures the degree of energy decline in material system with entropy.In subsequent research, Boltzmann and Ai Yinsi
The research of large quantities of scientists such as smooth makes people have deeper understanding to entropy.The 1940s, entropy was introduced information by Shannon
By Modern information science is produced, with entropy quantitative description information content, following formula is definition of the Shannon to comentropy：
Wherein, H indicates the average information of information source (entity for generating information), PzIndicate the probability that z-th of information source occurs.
For q performance indicators of network, network entropy may be defined as：
Hq=-log2Vq, q=1,2 ..., Q (13)
Wherein VqFor the normalization performance parameter of the q performance indicators.
The network entropy of network system should be the weighted sum of each individual event performance indicator entropy, i.e.,：
Wherein ωqFor the weight of q performance indicators.
Therefore, it can be used " entropy is poor "The attack effect that network system is subject to is described, wherein
ΔHq=-log2(Vq2/Vq1) it is the effect that q performance indicators of network are attacked, Vq1For q original property of network system
The normalization performance parameter of energy index, Vq2For network system under fire after q performance indicators normalization performance parameter.It is aobvious
So, Δ HnetBigger, then the overall performance of network system degenerates more severe, and attack effect is also more apparent.
Step S4, each node described in the current time found out according to above mentioned steps S3 is in the probability of different conditions, meter
Calculate the greateset risk entropy and minimum risk entropy of network structure described in current time.
According to the infection mechanism of SIR models：T moment is in the node x of easy infection state at the t+1 moment with PS(x,t
+ 1) probability remains easy infection state, while with PIThe probability of (x, t+1) is infected into Infection Status；T moment is in infection
The node y of state is at the t+1 moment with PRThe probability transition of (y, t+1) is recovery state, while with PIThe probability of (y, t+1) is kept
Infection Status.The node x of easy infection state keeps the probability P of easy infection state at the t+1 momentS(x, t+1) is smaller, infects simultaneously
The node y of state is changed into the probability P of recovery state at the t+1 momentR(y, t+1) is smaller, and network risk is bigger at this time.Thus
We can be defined on t moment network greateset risk entropy Hmax(t) it is：
Wherein：
GSAnd GIIt indicates respectively
It is the node set of easy infection state and Infection Status in network structure described in current time.
Similarly, the node x of t moment easy infection state is infected into the probability P of Infection Status at the t+1 momentI(x, t+1) is got over
It is small, while the node y of Infection Status remains the probability P of Infection Status at the t+1 momentI(y, t+1) is smaller, then at this time in network
Risk is smaller caused by rumour.As a result, we can obtain be in t moment network minimum risk entropy
Wherein,
To which the network risks entropy H (t) of network structure described in t moment is：
H (t)=Hmax(t)-Hmin(t) (17)
The network risks entropy difference Δ H (t) of network structure described in t moment is：
Δ H (t)=H (t)-H (t-1)=(Hmax(t)-Hmin(t))-(Hmax(t-1)-Hmin(t-1))。
Therefore, when the greateset risk entropy of t moment network is bigger and minimum risk entropy is smaller, while the t-1 moment is most
Risks entropy is smaller and when minimum risk entropy is bigger, the value of the network risks entropy difference of t moment is bigger, indicates current network ballad
Say that potential risks are bigger.It should more go into overdrive to be controlled at this time, inhibit rumour further to spread, avoid causing shadow to society
It rings.
Existing gossip propagation model needs to know the propagating source of rumour in advance, the mistake of dynamic communication since propagating source
Journey, and this is often difficult to obtain in advance in actual life.Because most of rumours are when having had propagated one section in reality
Between after just cause to pay close attention to, in social networks be difficult determine rumour propagation since where.
Meanwhile the method proposed by the present invention for assessing rumour risk based on network risks entropy difference constructs a large amount of instructions without collecting
Practice collection data or property data base, obtains different network risks entropys for different rumours, do not limited, made by model parameter
The result of rumour risk assessment is more accurate.The present invention is based on the methods of entropy can be in conjunction with each node in rumour dynamic communication network
State analysis whole network rumour risk size, can more fully hereinafter, quantitatively reflect the risk of network rumour.
The above content is a further detailed description of the present invention in conjunction with specific preferred embodiments, and it cannot be said that
The specific implementation of the present invention is confined to these explanations.For those skilled in the art to which the present invention belongs, it is not taking off
Under the premise of from present inventive concept, several equivalent substitute or obvious modifications can also be made, and performance or use is identical, all answered
When being considered as belonging to protection scope of the present invention.
Claims (7)
1. a kind of rumour methods of risk assessment based on network risks entropy difference, includes the following steps：
S1, to the network platform to be assessed, establish the gossip propagation model based on SIR models；
S2, the network structure based on the gossip propagation model, rumour source and its propagation are identified using rumour source discrimination
Time；The network structure includes multiple network nodes, one network user of each node on behalf；
S3, the rumour source according to step S2 determinations and its propagation time, the forward-propagating of rumour is simulated in the network structure
Process, to obtain the probability that each node in network structure described in current time is in different conditions；Wherein, node is described
State includes three kinds：Easy infection state, Infection Status and recovery state；
S4, the probability that different conditions are according to each node described in current time calculate network structure described in current time
Greateset risk entropy and minimum risk entropy；
S5, the greateset risk entropy using network structure described in current time and the minimum risk entropy calculate current time
The network risks entropy of the network structure；
S6, according to the network risks entropy at the network structure current time and the network risks entropy of the previous moment at current time,
The network risks entropy for calculating the network structure is poor, and assesses current time rumour to the net according to the network risks entropy difference
Potential risk caused by network platform.
2. rumour methods of risk assessment as described in claim 1, it is characterised in that：Rumour identifing source employed in step S2
Method includes back propagation, Jordan center methods, effective distance method and Monte Carlo Method.
3. rumour methods of risk assessment as described in claim 1, it is characterised in that：It is found using back propagation in step S2
Rumour source, including：
Several nodes in the network structure are chosen as observer nodes, constitute set S, and be observed in t moment, when t
It carves the observer nodes in Infection Status and constitutes setObserver nodes in easy infection state constitute set
Set of records endsIn each observer nodes at the time of infected by rumour；To setIn each observer nodes, be infected from the latest
Node start, according to respectively by rumour infect time sequencing, successively along the path backpropagation ballad of the network structure
Speech, can be aggregated simultaneouslyIn all observer nodes infection node be considered as doubtful rumour source, be added to doubtful rumour source collection
It closes in U；
To all node u in set U, the maximum likelihood value in t moment, the wherein maximum node of maximum likelihood value are sought respectively
As rumour source uf。
4. rumour methods of risk assessment as claimed in claim 3, it is characterised in that：Each node u in set U is in t moment
Maximum likelihood value isWherein, PI(h,th；U) be node h in thMoment is collected
It closes the rumour that the node u in U is transmitted to infect and be in the probability of Infection Status in t moment, PS(k,t；U) be node k in t
It carves the rumour infection that the node u not being aggregated in U is transmitted and is in easy infection shape probability of state；
α(h,th) indicate node h in thWhen
Carve infected probability, PS(h,th- 1) and PI(h,th- 1) indicate node h in (t respectivelyh- 1) moment be in easy infection state and
The probability of Infection Status, βhIt indicates that the node h in Infection Status is changed into and restores shape probability of state, 0 ＜ βh＜ 1；UhIndicate section
Credit value of the user in the network platform representated by point h；ChIndicate the user representated by node h in the network platform
On support,phContain in the content issued in the network platform for the user representated by node h and holds
Approve of the number of reviews of attitude, nhIt is anti-containing holding in the content issued in the network platform for the user representated by node h
To the number of reviews of attitude；
LhIt is predetermined nearest one for the user representated by node h
Liveness in period, NhThe set for the adjacent node j for being node h in the network structure, infjRepresentated by node j
Influence value of the user in the network platform, PI(j,th- 1) be node j in (th- 1) moment is in the general of Infection Status
Rate；
WithIndicate the user representated by node h in nearest one week and nearest January respectively
In the content quantity issued in the network platform,WithIndicate respectively user representated by node h nearest one week with
In the quantity thumbed up in the network platform in nearest January,WithIndicate the user representated by node h nearest respectively
The quantity for opinion of reexamining in the network platform last time in one week and nearest January；
PS(k,t；U)=(1- α (k, t)) PS(k, t-1), α (k, t) indicate node k in the infected probability of t moment, PS(k,
T-1) indicate that node k is in easy infection shape probability of state at (t-1) moment.
5. rumour methods of risk assessment as claimed in claim 4, it is characterised in that：The gossip propagation of each node u in set U
TimeRumour source ufAnd its gossip propagation time tfFor：
6. rumour methods of risk assessment as described in claim 1, it is characterised in that：In step s3, when defining described current
It is t moment to carve, and the node i in easy infection state is by the probability that rumour infects in t moment in the network structure
Wherein, LiIt is the user representated by node i in the liveness in the network platform in a nearest predetermined amount of time, NiFor
The set of adjacent node j of the node i in the network structure, infjIt is the user representated by node j in the network platform
On influence value, PI(j, t-1) is the probability that node j is in Infection Status at (t-1) moment；
Wherein,WithIndicate the user representated by node i at nearest one week and nearest one respectively
In the content quantity issued in the network platform in month,WithIndicate the user representated by node i nearest one respectively
In the quantity thumbed up in the network platform in all and nearest January,WithThe user representated by node i is indicated respectively
The quantity for opinion of reexamining in the network platform last time within nearest one week and nearest January；
According to the probability that node i is infected in t moment by rumour, calculate node i is in the probability P of easy infection state in t momentS(i,
T)=(1- α (i, t)) PS(i, t-1), node i are in the probability of Infection Status in t momentAnd node i is in the general of recovery state in t moment
Rate
Wherein, βiIt indicates that the node i in Infection Status is changed into and restores shape probability of state, 0 ＜ βi＜ 1；UiIndicate node i institute
Credit value of the user of representative in the network platform；CiIndicate the user representated by node i in the network platform
Support,piContain in the content issued in the network platform for the user representated by node i and holds approval
The number of reviews of attitude, niContain in the content issued in the network platform for the user representated by node i and holds opposition state
The number of reviews of degree.
7. rumour methods of risk assessment as claimed in claim 6, it is characterised in that：The maximum of network structure described in current time
Risk entropy Hmax(t) and minimum risk entropy Hmin(t) it is calculated respectively by following formula：
Wherein,
Wherein, GSAnd GIThe node set in network structure described in current time for easy infection state and Infection Status is indicated respectively；
Then the network risks entropy H (t) of network structure described in t moment is H (t)=Hmax(t)-Hmin(t)；
The network risks entropy difference Δ H (t) of network structure described in t moment is：
Δ H (t)=H (t)-H (t-1)=(Hmax(t)-Hmin(t))-(Hmax(t-1)-Hmin(t-1))；
The network risks entropy difference Δ H (t) is bigger, then potential risk caused by rumour is bigger；Conversely, then potential caused by rumour
Risk is smaller.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810239750.2A CN108520337B (en) | 2018-03-22 | 2018-03-22 | Riadry risk assessment method based on network risk entropy difference |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810239750.2A CN108520337B (en) | 2018-03-22 | 2018-03-22 | Riadry risk assessment method based on network risk entropy difference |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108520337A true CN108520337A (en) | 2018-09-11 |
| CN108520337B CN108520337B (en) | 2021-09-24 |
Family
ID=63432951
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810239750.2A Active CN108520337B (en) | 2018-03-22 | 2018-03-22 | Riadry risk assessment method based on network risk entropy difference |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108520337B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109522464A (en) * | 2018-10-22 | 2019-03-26 | 西南石油大学 | Information source detection method and system |
| CN112597699A (en) * | 2020-12-14 | 2021-04-02 | 新疆师范大学 | Social network rumor source identification method integrated with objective weighting method |
| CN112667784A (en) * | 2021-01-14 | 2021-04-16 | 浙江工商大学 | Rumor restraining method based on weighted reverse sampling |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101853261A (en) * | 2009-11-23 | 2010-10-06 | 电子科技大学 | Network public-opinion behavior analysis method based on social network |
| CN103049643A (en) * | 2012-11-22 | 2013-04-17 | 无锡南理工科技发展有限公司 | Mobile ad hoc network security risk assessment method based on risk entropy method and markoff chain method |
| CN106411904A (en) * | 2016-10-10 | 2017-02-15 | 华侨大学 | Network risk control method based on microstate prediction |
-
2018
- 2018-03-22 CN CN201810239750.2A patent/CN108520337B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101853261A (en) * | 2009-11-23 | 2010-10-06 | 电子科技大学 | Network public-opinion behavior analysis method based on social network |
| CN103049643A (en) * | 2012-11-22 | 2013-04-17 | 无锡南理工科技发展有限公司 | Mobile ad hoc network security risk assessment method based on risk entropy method and markoff chain method |
| CN106411904A (en) * | 2016-10-10 | 2017-02-15 | 华侨大学 | Network risk control method based on microstate prediction |
Non-Patent Citations (3)
| Title |
|---|
| LAIJIUN ZHAO,ET AL.: "SIR rumor spreading model in the new media age", 《PHYSICA A:STATISTICAL MECHANICS AND ITS APPLICATIONS》 * |
| 陈曦等: "基于信息熵的谣言信息度量方法", 《华中科技大学学报(自然科学版)》 * |
| 黄烨: "基于新浪微博的金融谣言识别方法探索", 《金融科技时代》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109522464A (en) * | 2018-10-22 | 2019-03-26 | 西南石油大学 | Information source detection method and system |
| CN112597699A (en) * | 2020-12-14 | 2021-04-02 | 新疆师范大学 | Social network rumor source identification method integrated with objective weighting method |
| CN112597699B (en) * | 2020-12-14 | 2023-03-14 | 新疆师范大学 | Social network rumor source identification method integrated with objective weighting method |
| CN112667784A (en) * | 2021-01-14 | 2021-04-16 | 浙江工商大学 | Rumor restraining method based on weighted reverse sampling |
| CN112667784B (en) * | 2021-01-14 | 2022-04-05 | 浙江工商大学 | Rumor restraining method based on weighted reverse sampling |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108520337B (en) | 2021-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105260474B (en) | A kind of microblog users influence power computational methods based on information exchange network | |
| CN104657425B (en) | Topic management type network public opinion evaluation management system and method | |
| Shulman et al. | Leveraging the power of a Twitter network for library promotion | |
| Tang et al. | Identifing influential users in an online healthcare social network | |
| CN106980692A (en) | A kind of influence power computational methods based on microblogging particular event | |
| Ognyanova et al. | A multitheoretical, multilevel, multidimensional network model of the media system: Production, content, and audiences | |
| CN108520337A (en) | A kind of rumour methods of risk assessment based on network risks entropy difference | |
| Zhao et al. | Simulation and modeling of microblog-based spread of public opinions on emergencies | |
| Wang et al. | Evolution and driving mechanism of tourism flow networks in the Yangtze River Delta urban agglomeration based on social network analysis and geographic information system: A double-network perspective | |
| Wei et al. | A new evaluation algorithm for the influence of user in social network | |
| CN107729455A (en) | A kind of social network opinion leader sort algorithm based on multidimensional characteristic analysis | |
| CN113850446A (en) | Information diffusion prediction method integrating space-time attention and heterogeneous graph convolution network | |
| Fang et al. | Social network public opinion research based on S-SEIR epidemic model | |
| CN113268976A (en) | Topic influence evaluation method facing microblog | |
| Zhang et al. | The relationship between Chinese college student offspring's physical activity and father physical activity during COVID-19 pandemic | |
| Cheng | Crowd‐Sourcing Information Dissemination Based on Spatial Behavior and Social Networks | |
| Xu et al. | COVID-19 vaccine sensing: Sentiment analysis from Twitter data | |
| Pan et al. | Multilevel analysis of individual and community predictors of smoking prevalence and frequency in China: 1991–2004 | |
| Shi et al. | Using social media for air pollution detection-the case of Eastern China smog | |
| Pan et al. | Hierarchical linear modelling of smoking prevalence and frequency in China between 1991 and 2004 | |
| CN115132369B (en) | Information propagation analysis method and system based on social media mimicry environment modeling | |
| Zixuan et al. | Impacts of Air Pollution on Chinese Expressions of Happiness on Social Media | |
| Zhang et al. | A study on the emotional and attitudinal behaviors of social media users under the sudden reopening policy of the Chinese government | |
| Stoimenov et al. | Using Sentiment Analysis of Twitter Data for Determining Popularity of City Locations | |
| Li et al. | Research on the Influence of Uploaders’ Behavior on the Effect of Bullet Screen Video Dissemination |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |