CN104008188B - Method for reliably revaluating credibility close to real behaviors - Google Patents
Method for reliably revaluating credibility close to real behaviors Download PDFInfo
- Publication number
- CN104008188B CN104008188B CN201410258417.8A CN201410258417A CN104008188B CN 104008188 B CN104008188 B CN 104008188B CN 201410258417 A CN201410258417 A CN 201410258417A CN 104008188 B CN104008188 B CN 104008188B
- Authority
- CN
- China
- Prior art keywords
- reliability
- rfr
- entity
- credibility
- original
- 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
- G06F16/903—Querying
- G06F16/9038—Presentation of query results
-
- 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
- G06F16/903—Querying
- G06F16/9032—Query formulation
Abstract
The invention discloses a method for reliably revaluating credibility close to real behaviors. The method includes the steps that original credibility of a target entity at the current moment is obtained through a credibility revaluation system, historical credibility at the current moment is calculated, the credibility fluctuation rate at the current moment is calculated according to the historical credibility and the original credibility at the current moment, the historical credibility fluctuation rate at the current moment is calculated, the credibility fluctuation trend at the current moment is calculated, and finally based on the original credibility, the historical credibility, the credibility fluctuation rate and the credibility fluctuation trend at the current moment, revaluated credibility of the original credibility at the current moment is calculated. According to the method, compared with the original credibility, the revaluated credibility is closer to the real behaviors of the target entity, the requirement for reliable evaluation of the credibility by a credibility management system is met, and the method is universal.
Description
Technical field
The present invention relates to a kind of reliability appraisal procedure, and in particular to a kind of reliable reliability revaluation side of approaching to reality behavior
Method.
Background technology
Recently as developing rapidly for network technology and being widely present for information system interconnection, it is large quantities of such as right to expedite the emergence of out
Deng the dynamic cooperative service system such as net, Agent system, Online e-business and cloud service.In such system service interaction both sides it
Between generally know little about it, this is caused in the case of without any interaction experiences in advance or priori, can cause a series of latent
Interactive risk.The Normal practice for evading such risk be deployment based on reliability belief system, on the one hand using reliability as
The foundation of information aggregation, filtration and sequence, interaction side can screen suitable interactive object;On the other hand using reliability as certainly
Social sanction's degree of body behavior, can encourage entity constantly to improve behavior, to provide preferably service.But above-mentioned two side
The key of face functional realiey is to develop reliable reliability assessment technology.
The dynamic and ambiguity of trust causes the reliable assessment to reliability extremely difficult.In order to ensure reliability assessment can
By property, the current main expansion in terms of Evaluation of reliability predictor selection, reliability add up to and precision improves three:
(1) selection of the Evaluation of reliability factor typically with apply strong correlation.The key for effectively choosing evaluation points is according to not
Choose different evaluation points in same application scenario.Such as:Felix et al. is proposed for P2P applications and referred to as trust ant colony system
The reliability assessment models of system TACS, use information element vestige is identifying the degree of belief between adjacent entities;Khan et al. is safeguarding cloud
The data safety of user and privacy are target, usage right control, proprietary rightss, prevention and safely four factors weighing user to cloud
The trusting degree of service.Qiao Xiu congruence people are using the familiarity and similarity between user as measurement social network user reliability
Foundation;Sojen Pradhan et al. are for the sensitive essence of medical information, it is proposed that one kind is used to creating, share or retrieving doctor
The trust model for the treatment of information, using medical information quality, medical service quality, informant's credibility as Evaluation of reliability because
Son;Deng.
(2) reliability is total generally to using weak related, and numerous versatility strategies are occurred in that for this.It is (subjective in personal experience
Reliability) total aspect, representative strategy is including simple average, Bayes, trust logic etc..Close in other people public praises (reliability recommendation)
Meter aspect, the summation method that eBay is used is the total strategy for obtaining paying close attention to earliest.To solve simple summation method easily by malice
The problem of (attacker can arbitrarily drive up itself reliability or belittle other people reliabilities) is attacked, current solution is to confer to recommend
The corresponding reliability of people recommends weight, most it is representational be Eigentrust use based on referrer's reliability and PeerTrust
Use based on both weighing computation methods of referrer's similarity.
(3) in terms of reliability correction, representing sex work has, and Wang et al. is closed for the existing reliability based on personal experience
Meter method have ignored prediction variance (accuracy of forecast), the total technology transition of the reliability based on other people public praises and rely on accurate system
Trust the problem of knowledge, it is proposed that a kind of general reliability assessment models, the model adds up to reliability prediction variance as reliability
Recommend the measurement foundation of reliability credibility in calculating, with reference to Kalman aggregating algorithms, expectation-maximization algorithm and hypothesis testing etc.
Statistical method, is effectively improved the accuracy of reliability assessment.Srivatsa et al. is proposed based on PID industrial control technologies
A kind of reliability revaluation model, the model can be on the premise of any restriction not be done to bottom reliability assessment algorithm to original reliability reality
Existing reliability correction, with certain versatility and robustness.
Although the Tremendous achievements in the research of reliable reliability assessment at present, existing research is reliable to weakening reliability assessment
Property assessment timeliness sex chromosome mosaicism lack concern.From in terms of the intension of trusting relationship, as a kind of closely related with behavior context
Dynamic process, trusting relationship can change with the dynamic change of behavior context.This dynamic characteristic of trusting relationship
Require that to the assessment of trusting relationship whole behavior context must be taken into account.But the presence of reliability assessment timeliness sex chromosome mosaicism is but caused
Completely taking whole behavior context into account becomes impossible:From for microcosmic point, the structure of assessment evidence is generally only capable of concern and works as
Before entity behavior before (assessment) time point (moment), cannot but take target entity current point in time and later entity row into account
For, and this part behavior context is particularly important for the assessment in time of the current reliability of target entity.From the meaning, reliability
Assessment timeliness sex chromosome mosaicism is not understood certainly, and reliability assessment result cannot just approach the real behavior of target entity, the reliability of reliability assessment
Property will have a greatly reduced quality.
The content of the invention
Goal of the invention:It is an object of the invention to solve the deficiencies in the prior art, there is provided a kind of approaching to reality row
For reliable reliability re-evaluation method.
Technical scheme:A kind of reliable reliability re-evaluation method of approaching to reality behavior of the present invention, comprises the following steps:
(1) original reliability of the target entity on current time t is obtained using reliability revaluation system, that is, treats revaluation reliability RR
[t], while original reliability RR [t] is saved in original reliability sequence, the history reliability engraved during for calculating future t+1
HR[t+1];
(2) using the original reliability on the front m historical juncture in original reliability sequence, i.e. RR [t-m] arrives RR [t-1], so
Normalization afterwards calculates history reliability HR [t] on current time t;Wherein, the front m historical juncture refer to moment t-m then
T-1, and the method that normalization is calculated are carved to carry out the weighted average calculation that original reliability attention rate is ρ (ρ≤1):
(3) according to history reliability HR [t] and original reliability RR [t] on the current time t for having tried to achieve, when calculating current
Reliability stability bandwidth RFR [t] on t is carved, while RFR [t] is saved in into history reliability stability bandwidth sequence, for calculating future t+
History reliability stability bandwidth HRFR [t+1] engraved when 1;
Wherein, RFR [t]=RR [t]-HR [t];
(4) using the reliability stability bandwidth on the front n historical juncture in history reliability stability bandwidth sequence, i.e. RFR [t-n] is arrived
RFR [t-1], then normalization calculates history reliability stability bandwidth HRFR [t] on current time t;Wherein, during front n history
Refer to moment t-n to moment t-1 quarters, and the method that normalization is calculated is to carry out reliability stability bandwidth attention rate adding for θ (θ≤1)
Weight average is calculated, i.e.,:
(5) by history reliability stability bandwidth HRFR [t] and reliability stability bandwidth RFR on above-mentioned required current time t
[t], calculates reliability fluctuation tendency RFT [t] on current time t, i.e.,:
RFT [t]=RFR [t]-HRFR [t]
(6) reliable revaluation reliability DR [t] of original reliability RR [t] on current time t is calculated, i.e.,:
DR [t]=α * RR [t]+β * HR [t]+γ * RFR [t]+δ * RFR [t] * | RFT [t] |
Wherein, factor alpha (0≤α≤1), β (0≤β≤1), γ (0≤γ≤1) and δ (0≤δ≤1) are component and calculate power
Weight.
Further, the m in the step (2) is set to the vibration time slot of tactful fluctuation behavior.
In order to ensure reliability stability bandwidth and reliability fluctuation tendency synchronized update, the value of n and taking for m in the step (4)
Value is identical.
Consistently to characterize the real behavior of target entity, the α and β in step (6) is proportional to m;
Meanwhile,γ1≤γ2, this makes it possible to ensure that the punishment dynamics for deteriorating behavior are not less than
The award dynamics that behavior is improved, and then being capable of the dishonest behavior that shows of incentives target entity;
Further,δ1≤δ2, it is sensitive by the assessment for lifting burst fluctuation behavior
Degree, can in time reflect the state change of target entity behavior.
Beneficial effect:Compared with prior art, the present invention has advantages below:
(1) present invention is real according to original reliability, history reliability, reliability stability bandwidth and reliability fluctuation tendency these four data
The ageing mechanism for having showed reliability assessment approaches the integrated of mechanism with existing real behavior, to make revaluation reliability various dimensions
Synchronous approaching to reality behavior is provided may.
(2) present invention passes through the fluctuation that happens suddenly of tolerance entity behavior, distinguishes the improvement of entity behavior and deteriorate, tolerate entity
Unconscious mis action, concordance reflect the global behavior of entity, and improve the ageing of reliability assessment so that revaluation is believed
The more original reliability of degree more approaches the real behavior of target entity, and realizing trust management system will to the reliable assessment of reliability
Ask.
(3) present invention when integrated reliability assesses ageing mechanism, do not destroy by the method for adding reduction time delay
Existing real behavior is approached in mechanism, and to original appraisal procedure of original reliability the criterion of any restriction is not done, and possesses certain
Versatility.
Description of the drawings
Fig. 1 is that reliability of the present invention assesses time delay origin cause of formation figure;
Fig. 2 is reliability of the present invention fluctuation atomic mode figure;
Fig. 3 is that time delay of the present invention reduction improves assessment degree of approximation design sketch;
Fig. 4 is present system functional framework figure;
Fig. 5 is present system process chart;
Burst fluctuation behaviors and view of the Fig. 6 for the embodiment of the present invention;
Sinusoidal fluctuation behaviors and view of the Fig. 7 for the embodiment of the present invention;
Fig. 8 is the reliability fluctuation tendency difference schematic diagram of the embodiment of the present invention;
Fig. 9 improves assessment degree of approximation schematic diagram for the time delay reduction of the embodiment of the present invention;
Figure 10 assesses degree of approximation schematic diagram for the time delay reduction intensity effect of the embodiment of the present invention;
Figure 11 is the asynchronous interaction success rate schematic diagram of malicious entities accounting of the embodiment of the present invention;
Figure 12 is the asynchronous interaction success rate schematic diagram of reliability threshold values of the embodiment of the present invention.
Specific embodiment
Combine accompanying drawing to technical solution of the present invention below to be described in detail.
As shown in figure 4, a kind of reliable reliability re-evaluation method of approaching to reality behavior disclosed by the invention, according to 4 dimensions
Data to original reliability implement revaluation, they are respectively original reliability (RR) itself, history reliability (HR), reliability stability bandwidth
And reliability fluctuation tendency (RFT) (RFR).Meanwhile, in order that more original reliability RR of reliability after revaluation approaches target entity
Real behavior, employs functional structure as shown in Figure 4, including using RFR so as to measure entity burst fluctuate behavior, use
HR, RFR and RFT to distinguish the improvement of entity behavior and to deteriorate, using RR and HR to tolerate the unconscious error row of entity
For, consistently reflect entity global behavior, using RFR and RFT so as to improve reliability assess it is ageing.
The concrete steps of the present invention are as shown in Figure 5:
(1) original reliability of the target entity on current time t is obtained using reliability revaluation system, that is, treats revaluation reliability RR
[t], while original reliability RR [t] is saved in original reliability sequence, the history reliability engraved during for calculating future t+1
HR[t+1];
(2) using the original reliability on the front m historical juncture in original reliability sequence, i.e. RR [t-m] arrives RR [t-1], so
Normalization afterwards calculates history reliability HR [t] on current time t;Wherein, the front m historical juncture refer to moment t-m then
T-1, and the method that normalization is calculated are carved to carry out the weighted average calculation that original reliability attention rate is ρ (ρ≤1):
(3) according to history reliability HR [t] and original reliability RR [t] on the current time t for having tried to achieve, when calculating current
Reliability stability bandwidth RFR [t] on t is carved, while RFR [t] is saved in into history reliability stability bandwidth sequence, for calculating future t+
History reliability stability bandwidth HRFR [t+1] engraved when 1;
Wherein, RFR [t]=RR [t]-HR [t];
(4) using the reliability stability bandwidth on the front n historical juncture in history reliability stability bandwidth sequence, i.e. RFR [t-n] is arrived
RFR [t-1], then normalization calculates history reliability stability bandwidth HRFR [t] on current time t;Wherein,
The front n historical juncture refers to moment t-n to moment t-1, and the method that normalization is calculated is to carry out reliability stability bandwidth
Attention rate is the weighted average calculation of θ (θ≤1), i.e.,:
(5) by history reliability stability bandwidth HRFR [t] and reliability stability bandwidth RFR on above-mentioned required current time t
[t], calculates reliability fluctuation tendency RFT [t] on current time t, i.e.,:
RFT [t]=RFR [t]-HRFR [t]
(6) reliable revaluation reliability DR [t] of original reliability RR [t] on current time t is calculated, i.e.,:
DR [t]=α * RR [t]+β * HR [t]+γ * RFR [t]+δ * RFR [t] * | RFT [t] |
Wherein, factor alpha (0≤α≤1), β (0≤β≤1), γ (0≤γ≤1) and δ (0≤δ≤1) are component and calculate power
Weight.
M in above-mentioned steps (2) is set to the vibration time slot of tactful fluctuation behavior.
In order to ensure reliability stability bandwidth and reliability fluctuation tendency synchronized update, the value of n and taking for m in the step (4)
Value is identical.
Consistently to characterize the real behavior of target entity, the α and β in step (6) is proportional to m.
Meanwhile,γ1≤ γ 2, this makes it possible to ensure that the punishment dynamics for deteriorating behavior are not little
In the award dynamics improved to behavior, and then being capable of the dishonest behavior that shows of incentives target entity.
In step (6),δ1≤δ2, it is sensitive by the assessment for lifting burst fluctuation behavior
Degree, can in time reflect the state change of target entity behavior.
The concrete operating principle of the present invention is as follows:
As shown in figure 1, there is delay problem in general reliability assessment, it is assumed that t1, t2And t3It is that target entity behavior is upper and lower
Three different time points (t in literary space1<t2<t3), wherein t1For behavior behavior1Time of origin point, t2For
behavior1Assessment time point (i.e. assessment result produce time point), while being also behavior behavior2Time of origin
Point, t3It is behavior behavior2Assessment time point, then time point t2The reliability assessment result that place obtains is only capable of characterizing (t2-
t1) behavior behavior before time span1, and behavior behavior2Assessment result then need through (t3-t2) time span
After could obtain, and target entity generally can show new row after this section of time span.Wherein time period (t2-t1) be
Reliability assesses time delay.
To weaken reliability assessment time delay, target entity short-term future behaviour can be predicted, i.e., in t2It is before right
behavior2It is predicted, and will predicts the outcome and include into behavior1Evidence to be assessed, then in t2What the moment obtained comments
Estimating result just can approx characterize t2Real behavior behavior at moment2, so as to reach the mesh for strengthening reliability assessment reliability
's.
In the present invention, the method for weakening reliability assessment time delay is as follows:
Assume known:Given reliability assessment algorithm EVA (entity, behavior), entity behavior BH (entity, x) (0
≤ BH (entity, x)≤1), belief function TV (entity, x) (0≤TV (and entity, x)≤1) (x is the time), and behavior
Prediction algorithm FC (behavior, RFR, RFT), then for physical object entity, if its real behavior at moment t is
BH (entity, t), original reliability stability bandwidth be RFR=(TV (entity, x)) ' |X=t, original reliability fluctuation tendency be RFT=
(TV(entity,x))″|X=t;(entity t) can be approximately reliability TV ' after the time delay reduction that then it is obtained at moment t
EVA(entity,FC(BH(entity,t),RFR,RFT)).As shown in Fig. 2 the four kinds of atomic modes fluctuated for reliability, i.e.,
Behavior burst deteriorates (RFR<0,RFT<0), (RFR≤0, RFT≤0), behavior burst improve (RFR to malicious act persistence>0,
RFT>0), dishonest behavior persistence (RFR >=0, RFT≤0).
FC (BH (entity, t), RFR, RFT)=BH (entity, t)+δ * RFR* | RFT |, Jin Eryou are set in the present invention
TV ' (entity, t) ≈ EVA (entity, BH (entity, t)+δ * RFR* | RFT |), while iterating to calculate for time delay during reduction
Convenience, arrange TV ' (entity, t)=TV (entity, t)+δ * RFR* | RFT |, wherein, δ>0 is time delay reduction intensity system
Number, stability bandwidth RFR is defined as time delay reduction direction coefficient (encourage or punish), and | RFT | is defined as time delay reduction granularity.
Wherein, reliability assessment reliability distance for TD (BH, TV)=| TV (entity, t)-BH (entity, t) |, BH
(entity, t) (0≤BH (entity, t)≤1) represent given real behavior, TV (entity, and t) (0≤TV (entity, t)
≤ 1) represent the reliability assessment result for being directed to real behavior.So reliability assessment degree of approximation then for TA (BH, TD)=1-TD (BH,
TV)/BH.Further, reliability assessed reliability is obtained. (entity, t), and two kinds are believed given substantive truth behavior BH
Degree assessment result TV1 (entity, t) (corresponding assessment degree of approximation is TA1) and TV2 (entity, t) (assessment degree of approximation TA2);
If have TA1≤TA2 set up, TV2 assessment BH (entity, t) on it is higher than the TV1 degree of reliability.
As shown in figure 3, set BH (entity, t) (0≤BH (and entity, t)≤1) be entity behavior, TV (entity, t) (0
≤ TV (entity, t)≤1) be time lag reduction before entity reliability, corresponding reliability apart from TD=TD (BH, TV), force by assessment
Recency is TA=TA (BH, TD);(entity, is t) reliability after time lag reduction to TV ', and has TV ' (entity, t)=TV
(entity, t)+δ * RFR* | RFT | (0≤TV ' (and entity, t)≤1), corresponding reliability distance is TD '=TD (BH, TV '),
Assessment degree of approximation is TA '=TA (BH, TD ').
In Fig. 3 (a), behavior is presented burst degradating trend, i.e. RFR<0 and RFT<0, due to δ>0, therefore δ * RFR* | RFT |<
0, that is, have | BH (entity, t)-TV ' (entity, t) |<| BH (entity, t)-TV (entity, t) | set up, i.e. TD '<TD,
So TA≤TA ', it can thus be appreciated that TV ' (entity, t) assessment BH (entity, t) on than TV, (entity, t) reliability is high;
In Fig. 3 (b), malicious act is presented persistence trend, i.e. RFR≤0 and RFT >=0, due to δ>0, therefore δ * RFR* |
RFT |≤0, that is, have | BH (entity, t)-TV ' (entity, t) |≤| BH (entity, t)-TV (entity, t) | set up, i.e.,
TD '≤TD, so as to there is TA≤TA ' establishments, it can thus be appreciated that TV ' (entity, t) assessment BH (entity, t) on compare TV
(entity, t) reliability is high;
In Fig. 3 (c), behavior is presented burst improvement trend, i.e. RFR>0 and RFT>0, due to δ>0, therefore δ * RFR* | RFT |>
0, that is, have | BH (entity, t)-TV ' (entity, t) |<| BH (entity, t)-TV (entity, t) | set up, i.e. TD '<TD,
So as to there is TA≤TA ' establishments, it can thus be appreciated that TV ' (entity, t) assessment BH (entity, t) on than TV (entity, it is t) reliable
Degree is high;
In Fig. 3 (d), dishonest behavior is presented persistence trend, i.e. RFR >=0 and RFT≤0, due to δ>0, therefore δ * RFR* |
RFT | >=0, that is, have | BH (entity, t)-TV ' (entity, t) |≤| BH (entity, t)-TV (entity, t) | set up, i.e.,
TD '≤TD, so as to there is TA≤TA ' establishments, it can thus be appreciated that TV ' (entity, t) assessment BH (entity, t) on compare TV
(entity, t) reliability is high.
In sum, the time delay method for weakening in the present invention can effectively shorten reliability distance, and reliability is apart from less, reliability
Assessment degree of approximation can be higher, and then reliability assessed reliability is also higher, i.e., the present invention can be commented with the improvement reliability of high degree
Estimate reliability.
Embodiment:A reliability revaluation model is set up using the reliable reliability re-evaluation method of the approaching to reality behavior of the present invention
GMTR。
If Fig. 6 and Fig. 7 is the happen suddenly Policy model and fluctuation status of fluctuation behavior and sinusoidal fluctuation behavior, can be with from figure
Find out, although their features on Policy model are different, but there are four kinds of states, i.e. behavior burst degradation mode (RFR<0 and
RFT<0), malicious act persistence state (RFR≤0 and RFT >=0), behavior burst improvement state (RFR>0 and RFT>0), and
Dishonest behavior persistence state (RFR >=0 and RFT≤0).
By taking the burst fluctuation behavior shown in Fig. 6 as an example, Fig. 8 is illustrating reliability stability bandwidth (original reliability attention rate ρ=1)
On the basis of, reliability fluctuation tendency (ρ=1, reliability stability bandwidth that further more different reliability stability bandwidth attention rates are brought
Attention rate θ takes respectively 1 and 0.75) difference, it can be seen that with the increase of θ, and reliability fluctuation tendency can more obvious (i.e. reliability ripple
Dynamic amplitude becomes big).
It is 0.8, γ that α is respectively provided with embodiment for 0.2, β1For 0.05 and γ2For 0.2, δ1For 0.05 | 0.1 and δ2For
0.2|0.4;The vibration time slot for arranging the fluctuation behavior of target entity strategy is 10, and it is (of interest during calculating history reliability to arrange m
Original reliability number) and n (history reliability number of interest when calculating history reliability stability bandwidth) be 10.Wherein rise for simple
See, original reliability RR computationally intends the simple summation method proposed using eBay.As shown in figure 9, going in same policy fluctuation
For under, the reliability assessment degree of approximation that existing Srivatsa is showed from the GMTR in the present invention has in different vibration time slots
Certain difference.It is concrete to can be seen that, on the one hand, in the identical (γ of time delay reduction direction coefficient1=0.05, γ2=feelings 0.2)
Under condition, the reliability revaluation result that the GMTR of the present invention is obtained more is approached than the reliability revaluation result that existing Srivatsa is obtained
The real behavior of target entity;On the other hand, in the identical (δ of time delay reduction strength factor1=0.05, δ2=0.2) in the case of, θ
Value is bigger by (1>0.75) the reliability assessment degree of approximation that, GMTR of the invention is given is also higher.
As shown in Figure 10, (ρ=1, θ=1), the bigger (δ of reduction intensity under same concerns degree1=0.1>δ1=0.05, δ2
=0.4>δ2=0.2), GMTR assessments degree of approximation is also bigger.
In general, in the case of potential interactive total amount identical, interaction success rate is lower, the restraint to strategy interaction
It is bigger.In order to verify restraint of the present invention to strategy interaction, the interaction success rate under different scenes is tested below.
Target setting entity selects interactive object according to reliability, and TV (x) is the reliability of target entity entity, then TVthr
(entity) it is the minimum reliability of the tolerable other side of entity (reliability threshold values).
If there is V (entity1)>=TVthrAnd TV (entity2) (entity2)>=TVthr(entity1) while setting up,
Then entity1 can be interacted successfully with entity2;Meanwhile, total is set as the interactive total amount number of inter-entity, succ is inter-entity
The successful number of times of interaction, then interaction success rate succ_rate=succ/total.
On above-mentioned reliability assessment degree of approximation experiment basis, further target entity is divided into into honest and two classes of malice,
Only malicious entities understand implementation strategy behavior (still by taking the fluctuation that happens suddenly as an example) to abuse reliability, while arrange target entity sum being
1025。
First, the strategy interaction restraint under different malicious entities accountings compares
Malicious entities are set and are respectively 0.3 and 0.6 with the reliability threshold values of honest entity.Figure 11 illustrates interactive total amount
Interaction success rate within 10000 times.It can be seen that when malicious entities accounting is higher (40%>20%), GMTR of the invention
It is generally equalized to the restraint of strategy interaction with existing Srivatsa, but when malicious entities accounting is relatively low (20%<40%),
The restraint of the GMTR of the present invention is even better, and as interaction total amount is more, this difference is more obvious.
Second, the strategy interaction restraint under different reliability threshold values compares
It is 20% to arrange malicious entities accounting.It can be recognized from fig. 12 that in different malicious entities reliability threshold values (0.2 Hes
0.6) under, the GMTR of the present invention shows higher strategy interaction restraint than existing Srivatsa.
Above example result and accompanying drawing show that the present invention is with higher reliability assessment degree of approximation and higher tactful ripple
Dynamic Behavior inhibition power.
Claims (1)
1. the reliable reliability re-evaluation method of a kind of approaching to reality behavior, it is characterised in that comprise the following steps:
(1) original reliability of the target entity on current time t is obtained using reliability revaluation system, that is, treats revaluation reliability RR [t],
Original reliability RR [t] is saved in original reliability sequence simultaneously, the history reliability HR [t engraved during for calculating future t+1
+1];
(2) using the original reliability on the front m historical juncture in original reliability sequence, i.e. RR [t-m] arrives RR [t-1], Ran Hougui
One changes history reliability HR [t] calculated on current time t;Wherein, the front m historical juncture refer to moment t-m to moment t-
1, and the method that normalization is calculated is to carry out the weighted average calculation that original reliability attention rate is ρ (ρ≤1):
M is set to the vibration time slot of tactful fluctuation behavior
(3) according to history reliability HR [t] and original reliability RR [t] on the current time t for having tried to achieve, current time t is calculated
On reliability stability bandwidth RFR [t], while RFR [t] is saved in into history reliability stability bandwidth sequence, for calculating during future t+1
History reliability stability bandwidth HRFR [t+1] for engraving;
Wherein, RFR [t]=RR [t]-HR [t];
(4) using the reliability stability bandwidth on the front n historical juncture in history reliability stability bandwidth sequence, i.e. RFR [t-n] arrives RFR [t-
1], then normalization calculates history reliability stability bandwidth HRFR [t] on current time t;Wherein, the front n historical juncture refer to
Moment t-n to moment t-1, and the method that normalization is calculated is to carry out the weighted average that reliability stability bandwidth attention rate is θ (θ≤1)
Calculate, i.e.,:
The value of n is identical with the value of m
(5) by history reliability stability bandwidth HRFR [t] and reliability stability bandwidth RFR [t] on above-mentioned required current time t, meter
Reliability fluctuation tendency RFT [t] on current time t is calculated, i.e.,:
RFT [t]=RFR [t]-HRFR [t]
(6) reliable revaluation reliability DR [t] of original reliability RR [t] on current time t is calculated, i.e.,:
DR [t]=α * RR [t]+β * HR [t]+γ * RFR [t]+δ * RFR [t] * | RFT [t] |
Wherein, factor alpha (0≤α≤1), β (0≤β≤1), γ (0≤γ≤1) and δ (0≤δ≤1) are component and calculate weight,
α and β are proportional to m,
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410258417.8A CN104008188B (en) | 2014-06-11 | 2014-06-11 | Method for reliably revaluating credibility close to real behaviors |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410258417.8A CN104008188B (en) | 2014-06-11 | 2014-06-11 | Method for reliably revaluating credibility close to real behaviors |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104008188A CN104008188A (en) | 2014-08-27 |
CN104008188B true CN104008188B (en) | 2017-05-10 |
Family
ID=51368845
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410258417.8A Active CN104008188B (en) | 2014-06-11 | 2014-06-11 | Method for reliably revaluating credibility close to real behaviors |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104008188B (en) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103347028A (en) * | 2013-07-15 | 2013-10-09 | 福州大学 | Peer-to-peer network trust degree measurement model based on Bayes under cloud architecture |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4596945B2 (en) * | 2005-03-24 | 2010-12-15 | 富士通株式会社 | Data center demand forecasting system, demand forecasting method and demand forecasting program |
-
2014
- 2014-06-11 CN CN201410258417.8A patent/CN104008188B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103347028A (en) * | 2013-07-15 | 2013-10-09 | 福州大学 | Peer-to-peer network trust degree measurement model based on Bayes under cloud architecture |
Non-Patent Citations (1)
Title |
---|
"用于增强信任评估可靠性的时滞性弱化策略";韩志耕 等;《计算机应用》;20140110;第34卷(第1期);第82-85页 * |
Also Published As
Publication number | Publication date |
---|---|
CN104008188A (en) | 2014-08-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10440048B1 (en) | Anti-attacking modelling for CMD systems based on GSPN and Martingale theory | |
Narasimhan et al. | Learnability of influence in networks | |
Imani et al. | Optimal gene regulatory network inference using the Boolean Kalman filter and multiple model adaptive estimation | |
CN108039987B (en) | Key infrastructure vulnerability assessment method based on multilayer coupling relation network | |
CN112989438B (en) | Detection and identification method for backdoor attack of privacy protection neural network model | |
Johnen et al. | Fitting lifetime distributions to interval censored cyclic-aging data of lithium-ion batteries | |
Harirchi et al. | Model invalidation for switched affine systems with applications to fault and anomaly detection | |
Liu et al. | Modelling complex large scale systems using object oriented Bayesian networks (OOBN) | |
Zhu et al. | Revisit dynamic arima based anomaly detection | |
Oyama et al. | Lyapunov-based economic model predictive control for detecting and handling actuator and simultaneous sensor/actuator cyberattacks on process control systems | |
CN104008188B (en) | Method for reliably revaluating credibility close to real behaviors | |
Huang et al. | Attention-aware deep reinforcement learning for detecting false data injection attacks in smart grids | |
Horvath et al. | Sensor fault diagnosis of inland navigation system using physical model and pattern recognition approach | |
Wei | Application of Bayesian algorithm in risk quantification for network security | |
Dervilis et al. | A non‐linear manifold strategy for SHM approaches | |
Canepa et al. | A dual model/artificial neural network framework for privacy analysis in traffic monitoring systems | |
Zhang et al. | Causal direction inference for network alarm analysis | |
Orchard et al. | A novel RSPF approach to prediction of high-risk, low-probability failure events | |
Nakayama et al. | Detection of false data injection attacks in cyber-physical systems using dynamic invariants | |
Herp et al. | State transition of wind turbines based on empirical inference on model residuals | |
Prayote et al. | Detecting anomalies and intruders | |
Teekaraman et al. | Abridged design with demand conventions for health care applications | |
Krönert et al. | Fdr control for online anomaly detection | |
Sales et al. | Analyzing Regression-Discontinuity Designs as Randomized Experiments | |
Alvari et al. | Causal inference for early detection of pathogenic social media accounts |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into 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: 20170407 Address after: Jiangpu street Pukou Road, Yushan District of Nanjing City, Jiangsu Province, No. 86 211815 Applicant after: Nanjing Audit University Address before: Water Simon Street in Jianye District of Nanjing City, Jiangsu province 210017 No. 207 Lake Garden 3 building 607 room Applicant before: Chen Geng Applicant before: Han Zhigeng |
|
GR01 | Patent grant | ||
GR01 | Patent grant |