CN103916392A - Entity recommendation trust calculation method based on reward and punishment factor and evaluation credibility - Google Patents

Abstract

The invention provides an entity recommendation trust calculation method based on a reward and punishment factor and evaluation credibility. The method is an access control technology recommendation trust value calculating method based on trust in an open network environment. According to the method, when an interbehavior happens to a network entity, an access object refers to the recommendation trust of other entities in interaction with an access subject previously, and a recommendation trust value and an overall recommendation trust value are comprehensively calculated through the reward and punishment factor, recommendation entity evaluation credibility and recommendation trust; the access object acquires the recommendation trust value of the access subject through the recommendation entity. Therefore, the access object considers and refers to the influence of the recommendation entity on the evaluation reliability of the access subject, and fraud access behaviors (such as joint vilifying and malicious exaggeration) of malicious entities can be effectively restrained.

Description

Based on the entity recommendation trust computational methods of the rewards and punishments factor and evaluation confidence level

Technical field

The present invention relates to the access control technology field based on trusting in computer network security, relate in particular to a kind of network entity in mutual based on the rewards and punishments factor and evaluate the entity recommendation trust computational methods of confidence level.

Background technology

Along with high speed development and the extensive use of computer networking technology and the communication technology (broadband technology, wireless technology etc.), current network becomes the sharable distributed type open formula network towards a large amount of external users from early stage main To enterprises, closed network Development in-house or fixed-line subscriber colony.In open network, safe and effective access control technology has become a study hotspot in current network safety filed in the world.In recent years, some scholars start to introduce trust means in access control research both at home and abroad, have obtained some preliminary achievements.Trust and provide new approaches for solving access control problem in open dynamic network.

But, prior art open network environment based on trust access control technology in the following shortcoming of ubiquity: the height dynamic and the ambiguity that in trust evaluation and quantizing process, do not take into full account trust, shortage can quantize to instruct foundation, can not accurately portray objective fact; In open network environment, dynamic to trusting relationship of trust evaluation and quantitative model, polymerism, multifactor property etc. lack consideration, have affected accuracy and the convergence of an assessment and quantification; Malice is accessed to the defensive attack scarce capacity of behavior, inadequate to the incentive mechanism of sincerity access behavior.

Summary of the invention

For the weak point existing in the problems referred to above, the invention provides a kind of entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level.

For achieving the above object, the entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level of the present invention comprise the following steps:

1. the entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level, is characterized in that:

Step 1, evaluates confidence level and the impact of recommendation trust in recommended entity trust value calculates according to the rewards and punishments factor, recommended entity, sets up recommendation trust computation model, and this recommendation trust computation model is:

Step 2, other and the recommendation trust of the entity of the mutual mistake of access main body in access object grid of reference, evaluate confidence level and recommendation trust COMPREHENSIVE CALCULATING is set up overall recommendation trust computation model according to the described rewards and punishments factor, recommended entity;

Step 3, calculate the described rewards and punishments factor according to Reward-Penalty Functions:

Step 4, calculate described recommended entity and evaluate confidence level:

Step 5, utilize the method for described step 3 to obtain the rewards and punishments factor, utilize the method for described step 4 to obtain recommended entity evaluation confidence level, and in the recommendation trust computation model that step 1 described in the described rewards and punishments factor and the substitution of described recommended entity evaluation confidence level is set up, solve recommendation trust;

Step 6, the described rewards and punishments factor that described step 5 is obtained and described recommended entity are evaluated in the overall recommendation trust computation model that step 2 is set up described in confidence level substitution, according to relevant parameter and computation model, solve overall recommendation trust.

Further, in described step 1, recommendation trust computation model is:

$T{R}_{e\→s}=\frac{1}{2}((\mathrm{val}\left(r_{e\→s}\right)\×{\mathrm{\η}}_{e\→s}+\frac{f\left(x_{e\→s}\right)}{{\mathrm{\Σ}}_{i\∈U_{e\→s}}f\left(x_{i\→s}\right)})\×T{D}_{e\→s}),$

Wherein f (x _{e → s}) be the rewards and punishments factor, η (x _{e → s}) be recommended entity evaluation confidence level, val (r _{e → s}) refer to institute's access rights of composing in access entity and access main body interactive access and all and access main body interworking entity and compose the ratio of authority summation, TD _{e → s}refer to network entity e to access main body s direct trust value.

Further, in described step 2, overall recommendation trust computation model is:

$\mathrm{TD}{R}_{U_{e\→s}}=\frac{1}{2}\underset{i\∈U_{e\→s}}{\mathrm{\Σ}}((\mathrm{val}\left(r_{i\→s}\right)\×{\mathrm{\η}}_{i\→s}+\frac{f\left(x_{i\→s}\right)}{{\mathrm{\Σ}}_{i\∈U_{e\→s}}f\left(x_{i\→s}\right)})\×T{D}_{i\→s}),$ Wherein val (r _{i → s}) refer to institute's access rights of composing in access entity and access main body interactive access and all and access main body interworking entity and compose the ratio of authority summation.It in step 1, is the recommendation trust of body one by one in the recommended entity of asking, step 2 refers to the recommendation trust of recommended entity set, and therefore the parameter in step 1 is to refer in fact, and in step 2, the parameter corresponding with step 1 is general reference, this is the routine of mathematical formulae, in this no longer burden discussion.

Further, in described step 3, rewards and punishments function f (x) is expressed as:

$f\left(x\right)=\left\{\begin{array}{c}\mathrm{0,0}\&le;x<{\mathrm{\&xi;}}_1\\ \frac{1}{2}\sin (\frac{x-{\mathrm{\&xi;}}_1}{{\mathrm{\&xi;}}_2-{\mathrm{\&xi;}}_1}\&times;\mathrm{\&pi;}-\frac{1}{2}\mathrm{\&pi;})+\frac{1}{2},{\mathrm{\&xi;}}_1\&le;x<{\mathrm{\&xi;}}_2\\ 1,{\mathrm{\&xi;}}_2\&le;x\&le;1\end{array}\right.$

The level of trust of feedback credibility x is ξ ₁, ξ ₂, ξ ₃respectively representative: completely distrust, critical credible, completely trust, level of trust space is designated as L={ ξ ₁, ξ ₂, ξ ₃, ξ _i∩ ξ _j= (i ≠ j) and ξ ₁< ξ ₂< ξ ₃, ξ ₁, ξ ₂, ξ ₃according to the dynamic value of different applied environments, setting level of trust space is L={0,0.15,0.85}.

Further, in described step 4, the computing formula of recommended entity evaluation confidence level is:

${\mathrm{\&eta;}}_{U_{e,o}}=\frac{1}{2\mathrm{num}\left(U_{e,o}\right)}\underset{x\&Element;U_{e,o}}{\mathrm{\&Sigma;}}|{\mathrm{TD}}_{e\&RightArrow;x}-{\mathrm{TD}}_{o\&RightArrow;x}|+p\left(x\right),$ Wherein, TD _o-xrefer to object O pair set U _e,oin the direct trust of each entity, p (x) is successful probability, set U _e,oin entity node be the entity node of object O, the mutual mistake of entity E, this set U _e,oin entity evaluation cross access object, and num (U _e,o) size for set U _e,othe number of element.

Beneficial effect of the present invention is:

When network entity generation interbehavior, other and the recommendation trust of the entity of the mutual mistake of access main body in access object grid of reference, go out recommendation trust and overall recommendation trust by the rewards and punishments factor, recommended entity evaluation confidence level and recommendation trust COMPREHENSIVE CALCULATING, excitation, the sincere access behavior of standard, eliminate the Malicious recommendation of malicious entities to the problem that affects of trust evaluation, strengthen its robustness, reach and trust the object quantizing.

Accompanying drawing explanation

Fig. 1 is the entity recommendation trust computational methods flow chart based on the rewards and punishments factor and evaluation confidence level of the present invention;

Fig. 2 be of the present invention be the mapping relations schematic diagram between feedback credibility of the present invention and the rewards and punishments factor.

Embodiment

Fig. 1 is the entity recommendation trust computational methods flow chart based on the rewards and punishments factor and evaluation confidence level of the present invention.As shown in Figure 1, the entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level of the present invention comprise:

Step 1, evaluates confidence level and the impact of recommendation trust in recommended entity trust value calculates according to the rewards and punishments factor, recommended entity, sets up recommendation trust computation model, and this recommendation trust computation model is:

$T{R}_{e\&RightArrow;s}=\frac{1}{2}((\mathrm{val}\left(r_{e\&RightArrow;s}\right)\&times;{\mathrm{\&eta;}}_{e\&RightArrow;s}+\frac{f\left(x_{e\&RightArrow;s}\right)}{{\mathrm{\&Sigma;}}_{i\&Element;U_{e\&RightArrow;s}}f\left(x_{i\&RightArrow;s}\right)})\&times;T{D}_{e\&RightArrow;s})$

The rewards and punishments factor is herein f (x _{e → s}): in recommendation trust computational process, embody the degree of reliability of the direct trust of recommended entity to access main body, access the degree of accepting and believing that object is directly trusted recommended entity.The rewards and punishments factor realizes on the one hand to the effect of the sincere access of inter-entity behavior incentive, also will play on the one hand the punishment effect to inter-entity swindle access behavior.In order to strengthen the constraint to inter-entity access behavior, introduce the concept of the rewards and punishments factor.The rewards and punishments factor realizes on the one hand to the effect of the sincere access of inter-entity behavior incentive, also will play on the one hand the punishment effect to inter-entity swindle access behavior.

Recommended entity is herein evaluated confidence level and is referred to η _{e → s}: must consider that recommended entity accesses the confidence level of object relatively with reference to other entity to the recommendation trust of access main body time, i.e. whether the evaluation confidence level of recommended entity, believe the recommendation of recommended entity to determine accessing object.Access object, by reference to the recommendation trust of recommended entity self, obtains the evaluation confidence level of recommended entity.Evaluating confidence level is the basis that overall recommendation trust calculates, and it has reflected that recommendation trust closes the credibility tying up in overall recommendation trust calculating.

Val (r _{e → s}) refer in access entity and access main body interactive access institute's access rights of composing and all and access main body interworking entity and compose the ratio of authority summation.

TD _{e → s}refer to network entity e to access main body s direct trust value.

Step 2, other and the recommendation trust of the entity of the mutual mistake of access main body in access object grid of reference, evaluate confidence level and recommendation trust COMPREHENSIVE CALCULATING is set up overall recommendation trust computation model according to the described rewards and punishments factor, recommended entity;

This overall situation recommendation trust computation model is:

$\mathrm{TD}{R}_{U_{e\&RightArrow;s}}=\frac{1}{2}\underset{i\&Element;U_{e\&RightArrow;s}}{\mathrm{\&Sigma;}}((\mathrm{val}\left(r_{i\&RightArrow;s}\right)\&times;{\mathrm{\&eta;}}_{i\&RightArrow;s}+\frac{f\left(x_{i\&RightArrow;s}\right)}{{\mathrm{\&Sigma;}}_{i\&Element;U_{e\&RightArrow;s}}f\left(x_{i\&RightArrow;s}\right)})\&times;T{D}_{i\&RightArrow;s})$

Wherein, val (r _{i → s}) refer in access entity and access main body interactive access institute's access rights of composing and all and access main body interworking entity and compose the ratio of authority summation.

It in step 1, is the recommendation trust of body one by one in the recommended entity of asking, step 2 refers to the recommendation trust of recommended entity set, and therefore the parameter in step 1 is to refer in fact, and in step 2, the parameter corresponding with step 1 is general reference, this is the routine of mathematical formulae, in this no longer burden discussion.

Overall recommendation trust herein: other and the recommendation trust of the entity of the mutual mistake of access main body in access object grid of reference, evaluate confidence level and recommendation trust COMPREHENSIVE CALCULATING goes out overall recommendation trust by the rewards and punishments factor, recommended entity.

Step 3, calculate the rewards and punishments factor according to Reward-Penalty Functions:

The level of trust of feedback credibility x is ξ ₁, ξ ₂, ξ ₃respectively representative: completely distrust, critical credible, completely trust, level of trust space is designated as L={ ξ ₁, ξ ₂, ξ ₃, ξ _i∩ ξ _j= (i ≠ j) and ξ ₁< ξ ₂< ξ ₃, Reward-Penalty Functions is sent out f (x) and is expressed as:

$f\left(x\right)=\left\{\begin{array}{c}\mathrm{0,0}\&le;x<{\mathrm{\&xi;}}_1\\ \frac{1}{2}\sin (\frac{x-{\mathrm{\&xi;}}_1}{{\mathrm{\&xi;}}_2-{\mathrm{\&xi;}}_1}\&times;\mathrm{\&pi;}-\frac{1}{2}\mathrm{\&pi;})+\frac{1}{2},{\mathrm{\&xi;}}_1\&le;x<{\mathrm{\&xi;}}_2\\ 1,{\mathrm{\&xi;}}_2\&le;x\&le;1\end{array}\right.$

The result that Reward-Penalty Functions f (x) calculates is the rewards and punishments factor f (x in step 1 _{e → s}).ξ ₁, ξ ₂, ξ ₃according to the dynamic value of different applied environments, it is L={0 that the present invention sets level of trust space, 0.15,0.85}, as shown in Figure 2, Fig. 2 is the mapping relations schematic diagram between feedback credibility of the present invention and the rewards and punishments factor to mapping function between feedback credibility and the rewards and punishments factor.

Can be seen by Fig. 2, in the time that the feedback degree of belief of inter-entity is greater than 0.5, the rewards and punishments factor embodies the subsidy to trust value, play the effect that trust value is rewarded, along with the generation of inter-entity interbehavior can improve the adopt degree of object to the direct trust of object, otherwise, in the time that the feedback degree of belief of inter-entity is less than 0.5, the rewards and punishments factor embodies the reduction to trust value, play the effect of trust value punishment, can significantly reduce the degree of adopting of object to the direct trust of entity, in the time that the rewards and punishments factor is 0, can also play the filtration that malicious entities is recommended, thereby effectively identify and resist the recommendation of malicious entities, guarantee the reliability of recommendation trust.

Step 4, calculated recommendation entity is evaluated confidence level:

Access object O obtains the trust evaluation to access main body S by entity E, and the trust evaluation that object O can provide main body S to entity E is weighed, and evaluates departure degree decide the weight to accessing main body S by correlation computations.

Each recommended entity arranges one and recommends sincere factor c, the ratio that be satisfied with number of times (successful interaction times) and recommendation number of times (be satisfied with number of times and dissatisfied number of times sum) of its value for recommending, i.e. successful Probability p; The initial value of p is made as 0.5,, in the time that recommended entity participates in recommending behavior for the first time, recommends the sincere factor to be defaulted as 0.5.

If set U _{e, o}in entity node be the entity node of object O, the mutual mistake of entity E, access object is crossed in the entity evaluation in this set, num (U _{e, o}) size for set U _{e, o}the number of element.

The set U of access object O to recommended entity E _{e, o}evaluation confidence level computation model be:

${\mathrm{\&eta;}}_{U_{e,o}}=\frac{1}{2\mathrm{num}\left(U_{e,o}\right)}\underset{x\&Element;U_{e,o}}{\mathrm{\&Sigma;}}|{\mathrm{TD}}_{e\&RightArrow;x}-{\mathrm{TD}}_{o\&RightArrow;x}|+p\left(x\right),$

Wherein TDo-x refers to object O pair set Ue, the direct trust of each entity in o, and p (x) is successful probability;

Ue, o initial value is 0.5, gathers Ue, the number of o element is 0 o'clock.

Step 5, utilize the method for described step 3 to obtain the rewards and punishments factor, utilize the method for described step 4 to obtain recommended entity evaluation confidence level, and in the recommendation trust computation model that step 1 described in the described rewards and punishments factor and the substitution of described recommended entity evaluation confidence level is set up, solve recommendation trust.

Step 6, the described rewards and punishments factor that described step 5 is obtained and described recommended entity are evaluated in the overall recommendation trust computation model that step 2 is set up described in confidence level substitution, according to relevant parameter and computation model, solve overall recommendation trust.

This programme is applicable to towards the calculating of recommendation trust in the access control technology based on trusting in open network environment, go out recommendation trust and overall recommendation trust by the rewards and punishments factor, recommended entity evaluation confidence level and recommendation trust COMPREHENSIVE CALCULATING, excitation, the sincere access behavior of standard, the affect problem of the Malicious recommendation of elimination malicious entities on trust evaluation, strengthens its robustness.In open network environment, access object considers and is referenced to the impact of evaluation confidence level of recommended entity on access main body, can effectively suppress the swindle access behavior (slander as combined, maliciously exaggerate etc.) of malicious entities.

Only as described above, be only preferred embodiment of the present invention, such as professional who are familiar with this art.After understanding technological means of the present invention, natural energy, according to actual needs, is changed under instruction of the present invention.Therefore all equal variation and modifications of doing according to the present patent application the scope of the claims, all should still remain within the scope of the patent.

Claims (5)

1. the entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level, is characterized in that:

Step 1, evaluates confidence level and the impact of recommendation trust in recommended entity trust value calculates according to the rewards and punishments factor, recommended entity, sets up recommendation trust computation model, and this recommendation trust computation model is:

Step 2, other and the recommendation trust of the entity of the mutual mistake of access main body in access object grid of reference, evaluate confidence level and recommendation trust COMPREHENSIVE CALCULATING is set up overall recommendation trust computation model according to the described rewards and punishments factor, recommended entity;

Step 3, calculate the described rewards and punishments factor according to Reward-Penalty Functions:

Step 4, calculate described recommended entity and evaluate confidence level:

Step 5, utilize the method for described step 3 to obtain the rewards and punishments factor, utilize the method for described step 4 to obtain recommended entity evaluation confidence level, and in the recommendation trust computation model that step 1 described in the described rewards and punishments factor and the substitution of described recommended entity evaluation confidence level is set up, solve recommendation trust;

Step 6, the described rewards and punishments factor that described step 5 is obtained and described recommended entity are evaluated in the overall recommendation trust computation model that step 2 is set up described in confidence level substitution, according to relevant parameter and computation model, solve overall recommendation trust.

2. the entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level as claimed in claim 1, is characterized in that, in described step 1, recommendation trust computation model is:

$T{R}_{e\&RightArrow;s}=\frac{1}{2}((\mathrm{val}\left(r_{e\&RightArrow;s}\right)\&times;{\mathrm{\&eta;}}_{e\&RightArrow;s}+\frac{f\left(x_{e\&RightArrow;s}\right)}{{\mathrm{\&Sigma;}}_{i\&Element;U_{e\&RightArrow;s}}f\left(x_{i\&RightArrow;s}\right)})\&times;T{D}_{e\&RightArrow;s}),$

Wherein f (x _{e → s}) be the rewards and punishments factor, η (x _{e → s}) be recommended entity evaluation confidence level, val (r _{e → s}) refer to institute's access rights of composing in access entity and access main body interactive access and all and access main body interworking entity and compose the ratio of authority summation, TD _{e → s}refer to network entity e to access main body s direct trust value.

3. the entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level as claimed in claim 1, is characterized in that, in described step 2, overall recommendation trust computation model is:

$\mathrm{TD}{R}_{U_{e\&RightArrow;s}}=\frac{1}{2}\underset{i\&Element;U_{e\&RightArrow;s}}{\mathrm{\&Sigma;}}((\mathrm{val}\left(r_{i\&RightArrow;s}\right)\&times;{\mathrm{\&eta;}}_{i\&RightArrow;s}+\frac{f\left(x_{i\&RightArrow;s}\right)}{{\mathrm{\&Sigma;}}_{i\&Element;U_{e\&RightArrow;s}}f\left(x_{i\&RightArrow;s}\right)})\&times;T{D}_{i\&RightArrow;s}),$ Wherein val (r _{i → s}) refer to institute's access rights of composing in access entity and access main body interactive access and all and access main body interworking entity and compose the ratio of authority summation.

4. the entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level as claimed in claim 1, is characterized in that, in described step 3, rewards and punishments function f (x) is expressed as:

$f\left(x\right)=\left\{\begin{array}{c}\mathrm{0,0}\&le;x<{\mathrm{\&xi;}}_1\\ \frac{1}{2}\sin (\frac{x-{\mathrm{\&xi;}}_1}{{\mathrm{\&xi;}}_2-{\mathrm{\&xi;}}_1}\&times;\mathrm{\&pi;}-\frac{1}{2}\mathrm{\&pi;})+\frac{1}{2},{\mathrm{\&xi;}}_1\&le;x<{\mathrm{\&xi;}}_2\\ 1,{\mathrm{\&xi;}}_2\&le;x\&le;1\end{array}\right.$

The level of trust of feedback credibility x is ξ ₁, ξ ₂, ξ ₃respectively representative: completely distrust, critical credible, completely trust, level of trust space is designated as L={ ξ ₁, ξ ₂, ξ ₃, ξ _i∩ ξ _j= (i ≠ j) and ξ ₁< ξ ₂< ξ ₃, ξ ₁, ξ ₂, ξ ₃according to the dynamic value of different applied environments, setting level of trust space is L={0,0.15,0.85}.

5. the entity recommendation trust computational methods based on the rewards and punishments factor and evaluation confidence level as claimed in claim 1, is characterized in that, in described step 4, the computing formula of recommended entity evaluation confidence level is:

${\mathrm{\&eta;}}_{U_{e,o}}=\frac{1}{2\mathrm{num}\left(U_{e,o}\right)}\underset{x\&Element;U_{e,o}}{\mathrm{\&Sigma;}}|{\mathrm{TD}}_{e\&RightArrow;x}-{\mathrm{TD}}_{o\&RightArrow;x}|+p\left(x\right),$ Wherein, TD _o-xrefer to object O pair set U _e,oin the direct trust of each entity, p (x) is successful probability, set U _e,oin entity node be the entity node of object O, the mutual mistake of entity E, this set U _e,oin entity evaluation cross access object, and num (U _e,o) size for set U _e,othe number of element.