CN116821350A - A trust-based semantic blockchain consensus method for social IoT - Google Patents

Abstract

The invention relates to a trust-based semantic blockchain consensus method for social Internet of Things, which determines the trust semantics between social Internet of Things entities through calculation of trust semantic feature values and trust semantic feature vectors extracted from interaction data of social Internet of Things entities. The relationship is formally expressed; then, the trust mechanism between the trust semantic relationships is mined through the approximate reasoning algorithm and Bayesian network, and a trust semantic consensus knowledge graph is constructed to form a semantic blockchain consensus for selective access to different scenarios and applications. protocol.

Description

A trust-based semantic blockchain consensus method for social IoT

Technical field

The present invention relates to the technical field of computer information security, and specifically to a trust-based semantic blockchain consensus method for social Internet of Things.

Background technique

Semantic blockchain refers to the use of semantic technology to convert blockchain data into a unified resource description format, so that the data has semantic characteristics, thereby providing a common data format and exchange protocol for the blockchain, and realizing different resources in the blockchain and data sharing, information interaction, resource scheduling, security management and other goals between different blockchains; the semantic blockchain builds a semantic processing layer on top of the basic blockchain architecture, and its main functions include semantic description of resources, semantics Data exchange protocols, semantic data storage, semantics of smart contracts, etc.; semantic blockchain can solve problems such as heterogeneous data sharing, heterogeneous data interoperability, data verification, and semantic search in the blockchain.

Trust management is the focus of research on the Social Internet of Things. Compared with the general Internet of Things, due to the addition of social attributes, the trust management of the Social Internet of Things faces new challenges; on the one hand, existing Internet of Things trust management technologies and methods are difficult to Directly used for trust management of the social Internet of Things. The existing trust management models of the Internet of Things are mostly based on centralized trust authentication, third-party trust management, trust domain management and other models; however, the Social Internet of Things is a freer and more flexible platform than the Internet of Things. In a more open environment, the cross-domain collaboration and cooperation of the Social Internet of Things will be more large-scale, heterogeneous, and dynamic than the general Internet of Things. It is very difficult and ineffective to try to strictly define its trust boundaries. Ensure the credibility of different users in the data collection, data transmission and data exchange process of their IoT devices; secondly, using blockchain directly for trust management in social IoT faces many insurmountable problems; due to the use of different Different platforms, different development languages, different protocols, different consensus mechanisms and privacy protection solutions have resulted in the existing blockchain being a closed ecosystem, and there is still a lack of effective interoperability between heterogeneous blockchains. Mechanisms, such as data fusion, data sharing, data mutual access issues of heterogeneous blockchains, blockchain performance and efficiency issues, semantic indexing and search issues, etc.; finally, the social Internet of Things is a very complex heterogeneous system. Different social IoT application scenarios have different trust consensus models. How to unify multiple models with different trust consensus and make flexible choices according to specific scenarios is also the core issue of trust consensus in the social IoT blockchain. The main reason for the above problems is that the social Internet of Things is a more open network, and social attributes increase the diversity of network types, device types, and data types. Therefore, in order to achieve trusted interaction and sharing between heterogeneous social IoT, it is necessary to design a trust-based semantic blockchain consensus model.

Contents of the invention

The problem solved by this invention is how to build a semantic blockchain consensus protocol for the social Internet of Things. In order to solve the above problems, the present invention provides features including:

Step 1. Collect interaction data between social IoT entities on the social IoT, classify and process the interaction data, remove non-text data, and form a standard text data set;

Step 2. Define the formal terms of the trust semantic ontology;

Step 3. Extract trust semantic feature values and trust semantic feature vectors composed of trust semantic feature values from the text data set, and perform a formal description of the trust semantic ontology;

Step 4. Calculate the similarity based on the trust semantic feature vector to determine whether there is a trust semantic relationship between social IoT entities;

Step 5: Determine the trust semantic relationship between social IoT entities according to the type of concepts represented by the trust semantic ontology;

Step 6: Use an approximate inference algorithm to obtain the marginal probability distribution between trust semantic relationships, and use Bayesian network to calculate the conditional probability of a causal relationship between trust semantic relationships;

Step 7. Use the analytic hierarchy process to conduct knowledge evaluation on any trust semantic relationship obtained through reasoning in step 6, and construct a semantic consensus knowledge graph based on the trust semantic relationship with causal relationships to form semantic blocks for selective access to different scenarios and applications. Chain consensus protocol.

The beneficial effects of the present invention are: determining the trust semantic relationship between social Internet of Things entities through calculation of trust semantic feature values and trust semantic feature vectors extracted from the interaction data of social Internet of Things entities, and performing formal representation; and then through approximation The inference algorithm and Bayesian network mine the trust mechanism between trust semantic relationships, build a trust semantic consensus knowledge graph, and form a semantic blockchain consensus protocol for selective access to different scenarios and applications.

Preferably, the interaction data collected in step 1 includes text, pictures, semantics and data transfer; the standard text data set includes user ID, device ID, and interaction trust semantic content text.

Preferably, step 2 specifically includes a set of feature vectors used to define trust semantic relationships between social Internet of Things entities; the set of feature vectors includes ownership trust, dependency trust, direct trust, indirect trust, and cooperation. Trust, distrust and deceitfulness.

Preferably, the trust semantic feature value extracted from the text data set in step 3 is:

r=(interaction trust semantic feature value);

The trust semantic feature vector extracted from the text data set is:

(interaction trust semantic feature value 2),

...

(interaction trust semantic feature value n)).

Preferably, the similarity in step 4 is calculated using the Pearson Correlation method, and the specific formula is:

In the formula, a and b represent two different social IoT entities respectively; Similarity (a, b) is the similarity value between social IoT entities i and j; V is the interaction between social IoT entities a and b The total number of trust semantic texts; i is the i-th trust semantic text of the interaction between social IoT entities a and b; (r _a ) and (r _b ) are the text data given by entities a and b respectively, and/> Represents the f-th type feature extracted from the i-th interaction of a and b respectively,/> and/> represent the mean values of the eigenvectors of a and b respectively.

Preferably, the category of the trust semantic relationship in step 5 is determined based on the type of concept represented by the trust semantic ontology. The category of the trust semantic relationship is determined by solving the maximum value of each type of semantic similarity between social Internet of Things entities. Specifically The calculation formula is:

In the formula, Max(θ _f=0,n ) represents the maximum value when f takes different values. That is, the value of f that makes θ the largest is the type of trust relationship.

Preferably, the approximate reasoning method used in step 6 to obtain the marginal probability distribution between any trust semantic relationships specifically includes:

Construct a simple distribution formula q ^* (z):

In the formula, Ω is the candidate distribution, and the mean field distribution family is used as the candidate distribution. All trust semantic relations are variables in the simple distribution formula, x is the variable of one group of trust semantic relations, and z is multiple groups of mutually independent trust semantic relations. The variable z={z ₁ ,z ₂ ,...z _m ,...z _M }, q(z) is the product of the variational probability density function of each random variable: q _m is the variational probability density of each random variable, z _m is the subset of other trust semantic relations, M is the number of trust semantic relation subsets; KL(q(z)||p(z|x) is KL Divergence is used to measure the information loss when calculating the editing probability distribution; arg represents the value of z when KL(q(z)||p(z|x) takes the minimum value;

Then, optimization calculation is performed through the expectation maximization algorithm, so that when the trust semantic relationship variables e={e ₁ , e ₂ ,... _en } are known, other trust semantic relationship variables f={f ₁ , f ₂ ,. ..f _n } conditional probability;

In step 6, a Bayesian network is used to calculate the conditional probability of a causal relationship between any trust semantic relationships. The Bayesian network formula is:

In the formula, e and f are trust semantic relationship variables, and z represents the rest of the trust semantic relationship variables except e and f; the conditional probability of causality between all trust semantic relationships is calculated through Bayesian network.

As a preferred method, using the analytic hierarchy process in step 7 to evaluate the knowledge of the trust semantic relationship specifically includes:

Step 701: Establish an evaluation index hierarchy based on the feature vector set;

Step 702: Quantify the evaluation indicators, construct a judgment matrix based on each layer of evaluation indicators, calculate the maximum eigenvalue of each judgment matrix as the weight of each layer of evaluation indicators, and then obtain the trust semantic relationship based on the weight. Evaluation result vector of relationship type.

Description of the drawings

Figure 1 is an interaction diagram between social IoT views;

Figure 2 is a schematic flow diagram of the present invention;

Detailed ways

In order to make the above objects, features and advantages of the present invention more obvious and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

A trust-based semantic blockchain consensus method for social IoT. In the social IoT scenario as shown in Figure 1, interactions between social IoT entities generate interactive data. A semantic blockchain consensus method for social IoT As shown in Figure 2, it includes:

Step 1. Collect interaction data between social IoT entities on the social IoT. The interaction data includes text, pictures, semantics, data transfer, etc., and processes the interaction data by classifying it to remove non-text data and form standard text. Class data set, the standard text class data set includes: text, pictures, semantics and data transfer; the standard text class data set includes user ID, device ID, interactive trust semantic content text;

Step 2. Based on the background knowledge of semantic ontology in the existing technology, define the formal terms of the trust semantic ontology; specifically include the set of eigenvalues and eigenvectors used to define the trust semantic relationship between social Internet of Things entities; the eigenvalue belongs to The semantic ontology concept set of includes possessive trust, subordinate trust, direct trust, indirect trust, cooperative trust, distrust and deceptiveness; thus, the trust semantic ontology in this specific embodiment is expressed as:

TrustObject[].FORALL User1,Device1,User2,Device2

User::TrustObject.

User1:User[dir-trustWith->>User2]，

User2:User[dir-trustWith->>User1]，

User1:User[indir-trustWith->>User2]，

User2:User[indir-trustWith->>User1]，

User1:User[holdWith->>Device1]，

User2:User[holdWith->>Device1]，

User1:User[dir-trustWith->>Device2]，

User2:User[dir-trustWith->>Device1],

User1:User[distrustWith->>User2]，

User2:User[distrustWith->>User1]，

User1:User[che-trustWith->>User2]，

User2:User[che-trustWith->>User1]，

Device::TrustObject.

Device1:Device[belongTo->>User1]，

Device2:Device[belongTo->>User2]，

Device1:Device[cooperatesWith->>Device2],

Device2:Device[cooperatesWith->>Device1];

Step 3. Extract trust semantic feature values and trust semantic feature vectors from the text data set, which specifically includes: Extracting trust semantic feature values from the text data set is:

r=(interaction trust semantic feature value);

The interactive trust semantic content is formally represented by the trust semantic ontology; the formal representation of the trust semantic ontology is the existing basis and will not be described in detail here; and, after the formal representation of the trust semantic ontology, the trust semantic characteristics The trust semantic relationship is determined based on the interactive trust semantic content in the value;

Then the trust semantic feature vector extracted from the text data set is:

(interaction trust semantic feature value 2),

...

(interaction trust semantic feature value n);

And the interactive trust semantic feature value 1, the interactive trust semantic feature value 2...the interactive trust semantic feature value n are used to formalize the trust semantic ontology;

Step 4. Calculate the similarity based on the trust semantic feature vector to determine whether there is a trust semantic relationship between social IoT entities; the similarity in this specific embodiment is calculated using the Pearson Correlation method, and the specific formula is:

In the formula, a and b represent two different social IoT entities respectively; Similarity (a, b) is the similarity value between social IoT entities i and j; V is the interaction between social IoT entities a and b The total number of trust semantic texts; i is the i-th trust semantic text of the interaction between social IoT entities a and b; (r _a ) and (r _b ) are the text data given by entities a and b respectively, and/> Represents the f-th type feature extracted from the i-th interaction of a and b respectively,/> and/> Represents the mean value of the eigenvectors of a and b respectively;

Step 5: Determine the trust semantic relationship between social IoT entities according to the type of concept represented by the trust semantic ontology. In this specific embodiment, the category of the trust semantic relationship is determined by solving the semantic similarity of each type between social IoT entities. To obtain the maximum value, the specific calculation formula is:

In the formula, Max(θ _f=0,n ) represents the maximum value when f takes different values, that is, the value of f that makes θ the largest is the type of trust relationship;

In this specific embodiment, the similarity range for determining the existence of a trust semantic relationship is (0,1]. Determine whether the similarity calculated in step 4 falls within the predetermined range. If so, calculate the maximum value of each type of f, corresponding to The f type is the type of trust relationship; if not, there is no trust semantic relationship between social IoT entities;

Step 6: Use the approximate inference algorithm to obtain the marginal probability distribution of any trust semantic relationship, and use Bayesian network to calculate the conditional probability of a causal relationship between any trust semantic relationships; among them, use the approximate inference method to obtain any trust semantic relationship The marginal probability distributions specifically include:

Construct a simple distribution formula q ^* (z):

In the formula, Ω is the candidate distribution, and the mean field distribution family is used as the candidate distribution. All trust semantic relations are variables in the simple distribution formula, e is one of the variables of a group of trust semantic relations, and q(z) is the variation of each random variable. Product of partial probability density functions: z _i is a subset of other trust semantic relations z except e, q _i (z _i ) is the probability distribution of the subset, m is the number of trust semantic relation subsets; KL(q(z)||p(z |e) is KL divergence, which is used to measure the information loss when calculating the editing probability distribution. z is a hidden variable; arg represents the value of z when KL(q(z)||p(z|x) takes the minimum value. value;

Then, the optimization calculation is performed through the expectation maximization algorithm to obtain q ^* (z) that meets the conditions, and q ^* (z) is used to approximate p(z|e); thus, after knowing the trust semantic relationship variable e={e ₁ , When e ₂ ,...e _n }, the conditional probability of other trust semantic relationship variables f = {f ₁ , f ₂ ,...f _n };

Bayesian network is used to calculate the conditional probability of causality between any trust semantic relationships. The Bayesian network formula is:

In the formula, e and f are trust semantic relationship variables, and z represents the rest of the trust semantic relationship variables except e and f; the conditional probability of causality between all trust semantic relationships is calculated through Bayesian network;

Step 7. Use the analytic hierarchy process to conduct knowledge evaluation on any trust semantic relationship obtained through reasoning in step 6, and construct a semantic consensus knowledge graph based on the trust semantic relationship with causal relationships to form semantic blocks for selective access to different scenarios and applications. Chain consensus protocol: wherein the instruction evaluation specifically includes:

Step 701: Establish an evaluation index hierarchy based on the feature vector set;

This specific embodiment uses the trust semantic feature value in the feature vector set as the second-level evaluation index; and adds a first-level evaluation index on the basis of the second-level evaluation index. The first-level evaluation index includes ownership trust relationship, parent-generation trust relationship, and collaboration trust. relationship, mutually exclusive trust relationship; wherein, the ownership trust and subordinate trust are subordinate to the ownership trust relationship, the direct trust is subordinate to the parent trust relationship, the indirect trust and cooperative trust are subordinate to the collaboration trust relationship, and the Distrust and deceptive affiliation belong to mutually exclusive trust relationships; as shown in Figure 1:

Table 1

Step 702: Construct a judgment matrix according to the importance of each layer's evaluation index to quantify the judgment, calculate the maximum eigenvalue of each judgment matrix as the weight of each layer's evaluation index, and then use the weight to determine the trust semantic relationship between Evaluation result vector of relationship type; this specific embodiment constructs a judgment matrix B according to the importance of the evaluation index for quantification, as shown in Table 2:

Table 2

Based on the quantitative values in Table 2, establish a judgment matrix of evaluation results and four first-level evaluation index judgment matrices; calculate the maximum eigenvalue of each judgment matrix, thereby obtaining the weight vector W of the evaluation results and the weight W _l of the first-level indicators ; And calculate the evaluation results based on the weights W and W _l , the calculation formula is:

S _l = W _l *R _l

S＝W*R

In the formula, W _l is the weight vector of the first-level evaluation index, R _l is the judgment matrix of the first-level evaluation index, S _l is the evaluation result vector of the first-level evaluation index, l is the number of the first-level evaluation index, and W is the evaluation result. The weight vector of R = [S ₁ , S ₂ , S ₃ , S ₄ ], S is the evaluation result vector; according to the value of the evaluation result vector, the relationship type between each trust semantic relationship is determined for validity judgment.

The standardized description of the trust semantic feature values and feature vectors specifically includes:

According to the value of the trust relationship degree between social IoT entities in the trust relationship degree set, the nature of the secondary trust relationship is divided according to the trust relationship degree range in the secondary trust relationship in the trust semantic relationship standard library, and then through the form of semantic ontology standardized description.

The formalized trust semantic feature values need to be standardized in order to perform relevant calculations and processing; in this specific embodiment, the feature values are standardized based on the average and standard deviation of the feature values, and the standard deviation of r of the feature values is used for standardization. To calculate, the standard deviation is defined as:

In the formula, n is the number of feature values in the text, is the average value of the f-th feature, and/> Represents the f-th eigenvalue of the i-th data;

Default Its standardized value is:

application

The scenarios and applications in the Social Internet of Things cover a wide range of areas. In different scenarios and different applications, the consensus protocols within similar scenarios and in the same application are generally the same, but the consensus protocols between different scenarios and different applications are many different. The application fields of social Internet of Things include Smart homes, Internet of Vehicles, lifestyle entertainment, etc., have different requirements for the consensus mechanism of blockchain in different application scenarios; according to these scenarios and applications, the optional rule for trust semantic blockchain is the ownership trust relationship Any combination of four trust relationships: parent-generation trust relationship, collaborative trust relationship, and mutually exclusive trust relationship. These combinations constitute the consensus protocol of the trust semantic blockchain. The rules available for combination include:

{

Combination 1 (ownership trust relationship),

Combination 2 (parental trust relationship),

Combination 3 (collaborative trust relationship),

Combination 4 (non-mutually exclusive trust relationship),

Combination 5 (ownership trust relationship, parental trust relationship),

Combination 6 (ownership trust relationship, collaboration trust relationship),

Combination 7 (ownership trust relationship, non-mutually exclusive trust relationship),

Combination 8 (parental trust relationship, collaborative trust relationship),

Combination 9 (parental trust relationship, non-mutually exclusive trust relationship),

Combination 10 (collaborative trust relationship, non-mutually exclusive trust relationship),

Combination 11 (ownership trust relationship, parent trust relationship, collaboration trust relationship),

Combination 12 (ownership trust relationship, parent trust relationship, non-mutually exclusive trust relationship),

Combination 13 (ownership trust relationship, collaboration trust relationship, non-mutually exclusive trust relationship),

Combination 14 (parental trust relationship, collaborative trust relationship, non-mutually exclusive trust relationship),

Combination 15 (ownership trust relationship, parent trust relationship, collaboration trust relationship, non-mutually exclusive trust relationship)

}

The relationships within the group are "and" relationships, that is, if this group of rules is selected, all conditions within the group need to be met at the same time; the relationships between groups are "or" relationships. After steps 1 to 7 of this application, we get A combination of trust consensus protocols that is based on trust, uses trust to describe the semantic state of entities, and is oriented to semantic blockchains, so that it can meet the same scenarios in the social Internet of Things and the same consensus protocols within the same application, and can also meet different needs. Scenarios and consensus protocols between different applications.

Although the present disclosure is disclosed as above, the protection scope of the present disclosure is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present disclosure, and these changes and modifications will fall within the protection scope of the present invention.

Claims (8)

1. A trust-based semantic blockchain consensus method for social IoT, including social IoT entities, characterized by: Step 1. Collect interaction data between social IoT entities on the social IoT, classify and process the interaction data, remove non-text data, and form a standard text data set; Step 2. Define the formal terms of the trust semantic ontology; Step 3. Extract trust semantic feature values and trust semantic feature vectors composed of trust semantic feature values from the text data set, and perform a formal description of the trust semantic ontology; Step 4. Calculate the similarity based on the trust semantic feature vector to determine whether there is a trust semantic relationship between social IoT entities; Step 5: Determine the trust semantic relationship between social IoT entities according to the type of concepts represented by the trust semantic ontology; Step 6: Use an approximate inference algorithm to obtain the marginal probability distribution between trust semantic relationships, and use Bayesian network to calculate the conditional probability of a causal relationship between trust semantic relationships; Step 7. Use the analytic hierarchy process to conduct knowledge evaluation on any trust semantic relationship obtained through reasoning in step 6, and construct a semantic consensus knowledge graph based on the trust semantic relationship with causal relationships to form semantic blocks for selective access to different scenarios and applications. Chain consensus protocol. 2. A trust-based social Internet of Things semantic blockchain consensus method according to claim 1, characterized in that the interaction data collected in step 1 includes text, pictures, semantics and data transfer; the standard The text data set includes user ID, device ID, and interactive trust semantic content text. 3. A trust-based social IoT semantic blockchain consensus method according to claim 2, characterized in that step 2 specifically includes characteristic values for defining trust semantic relationships between social IoT entities and A set of feature vectors; the set of semantic concepts described in the feature includes possessive trust, subordinate trust, direct trust, indirect trust, cooperative trust, distrust and deceptiveness. 4. A trust-based social Internet of Things semantic blockchain consensus method according to claim 3, characterized in that in step 3, the trust semantic feature value extracted from the text data set is: r=(interaction trust semantic feature value); The trust semantic feature vector extracted from the text data set is: (interaction trust semantic feature value 2), ... (interaction trust semantic feature value n)). 5. A trust-based social Internet of Things semantic blockchain consensus method according to claim 4, characterized in that the similarity in step 4 is calculated using the Pearson Correlation method, and the specific formula is: In the formula, a and b represent two different social IoT entities respectively; Similarity (a, b) is the similarity value between social IoT entities i and j; V is the interaction between social IoT entities a and b The total number of trust semantic texts; i is the i-th trust semantic text of the interaction between social IoT entities a and b; (r _a ) and (r _b ) are the text data given by entities a and b respectively, and/> Represents the f-th type feature extracted from the i-th interaction of a and b respectively,/> and/> represent the mean values of the eigenvectors of a and b respectively. 6. A trust-based social Internet of Things semantic blockchain consensus method according to claim 5, characterized in that the category of the trust semantic relationship in step 5 is determined according to the type of concept represented by the trust semantic ontology, The categories of trust semantic relationships are determined by solving the maximum value of each type of semantic similarity between social IoT entities. The specific calculation formula is: In the formula, Max(θ _f=0,n ) represents the maximum value when f takes different values. That is, the value of f that makes θ the largest is the type of trust relationship. 7. A trust-based social Internet of Things semantic blockchain consensus method according to claim 6, characterized in that in step 6, an approximate reasoning method is used to obtain the specific marginal probability distribution between any trust semantic relationships. include: Construct a simple distribution formula q ^* (z): In the formula, Ω is the candidate distribution, and the mean field distribution family is used as the candidate distribution. All trust semantic relations are variables in the simple distribution formula, x is the variable of one group of trust semantic relations, and z is multiple groups of mutually independent trust semantic relations. The variable z={z ₁ ,z ₂ ,...z _m ,...z _M }, q(z) is the product of the variational probability density function of each random variable: q _m is the variational probability density of each random variable, z _m is the subset of other trust semantic relations, M is the number of trust semantic relation subsets; KL(q(z)||p(z|x) is KL Divergence is used to measure the information loss when calculating the editing probability distribution; arg represents the value of z when KL(q(z)||p(z|x) takes the minimum value; Then, optimization calculation is performed through the expectation maximization algorithm, so that when the trust semantic relationship variables e={e ₁ , e ₂ ,... _en } are known, other trust semantic relationship variables f={f ₁ , f ₂ ,. ..f _n } conditional probability; In step 6, a Bayesian network is used to calculate the conditional probability of a causal relationship between any trust semantic relationships. The Bayesian network formula is: In the formula, e and f are trust semantic relationship variables, and z represents the rest of the trust semantic relationship variables except e and f; the conditional probability of causality between all trust semantic relationships is calculated through Bayesian network. 8. A trust-based social Internet of Things semantic blockchain consensus method according to claim 7, characterized in that in step 7, using the analytic hierarchy process to perform knowledge evaluation on the trust semantic relationship specifically includes: Step 701: Establish an evaluation index hierarchy based on the feature vector set; Step 702: Quantify the evaluation indicators, construct a judgment matrix based on each layer of evaluation indicators, calculate the maximum eigenvalue of each judgment matrix as the weight of each layer of evaluation indicators, and then obtain the trust semantic relationship based on the weight. Evaluation result vector of relationship type.