Knowledge Token dynamic value evaluation and transaction authorization method

Abstract

The invention discloses a knowledge Token dynamic value evaluation and transaction authorization method, and relates to the technical field of knowledge management. An original knowledge unit is packaged into a knowledge Token, and a standardized knowledge representation model is constructed by defining a core metadata field. According to the method, the original knowledge unit is packaged into the knowledge Token with the characteristics of unique identification, structured semantics and the like, the multi-dimensional core metadata field is defined, and the standardized knowledge representation model is constructed, so that the problem that traditional knowledge exists in an unstructured document and lacks machine readability is thoroughly solved; meanwhile, multiple dimensions such as calling frequency and business influence degree are fused, a knowledge Token value score is calculated in real time through a dynamic weighting algorithm, dynamic quantification of knowledge values is achieved, accurate recognition of high-value assets and efficient cleaning of low-value assets are achieved through an automatic grading mechanism, and the management efficiency and utilization value of the knowledge assets are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of knowledge management, in particular to a knowledge Token dynamic value evaluation and transaction authorization method. BACKGROUND

[0002] Traditional knowledge asset circulation is plagued by strong subjectivity of static evaluation, difficulty in right protection, cumbersome authorization process, and opaque income distribution, etc. The decentralized and traceable characteristics of blockchain build a solid foundation of trust, AI and big data realize multi-dimensional quantitative analysis of knowledge, dynamic Tokenization technology converts abstract knowledge into tradable digital assets, and superimposes digital policy support for intellectual property rights.

[0003] In the prior art, knowledge is mostly in the form of unstructured documents, lacking machine readability and semantic encapsulation, and cannot support multi-agent automated collaboration; the authorization mechanism mostly relies on static RBAC model, and it is difficult to realize fine-grained control combined with dynamic context, which easily leads to excessive or insufficient permissions; the value of knowledge is regarded as a static attribute, lacking a dynamic quantitative mechanism based on multi-dimensional indicators, and the pricing is rigid and lacks incentives; there is no closed-loop design for transaction and authorization, and the security calling guarantee system is not perfect, lacking full-link audit and evidence preservation mechanism, and it is difficult to prevent the risks of unauthorized access, data leakage and large model illusion.

[0004] Therefore, the present application provides a knowledge Token dynamic value evaluation and transaction authorization method to solve the above problems. SUMMARY

[0005] The main purpose of the present application is to provide a knowledge Token dynamic value evaluation and transaction authorization method to solve the problems raised in the background.

[0006] To achieve the above purpose, the technical solution adopted by the present application is as follows: a knowledge Token dynamic value evaluation and transaction authorization method, comprising the following steps:

[0007] Step 1: encapsulate the original knowledge unit into knowledge Token, and construct a standardized knowledge representation model by defining core metadata fields;

[0008] Step 2: based on knowledge Token, through the fusion of multiple dimensions of calling frequency, business impact, confidence, time decay and knowledge collaborative adaptation, through dynamic weighting algorithm, real-time calculation of the value score of each knowledge Token, and automatic classification processing; and based on the attributes of subject, object, scene, environment and collaboration, cross-verification of the calling request, and output of the differentiated decision results of direct authorization, integral payment authorization or rejection authorization;

[0009] The dynamic weighting algorithm is specifically: determining the initial weight of each dimension based on the entropy weight method, and dynamically adjusting the weight through an adaptive learning model combined with real-time feedback data, wherein the weight updating formula of each dimension is:

[0010] ;

[0011] wherein, represents the weight of the i-th dimension at time t, represents the learning rate, represents the value contribution change amount of the i-th dimension;

[0012] Step three: after obtaining the authorization, the system verifies the authenticity of the interface parameters and data sources, and confirms the validity of the associated Token. The transaction platform completes the real-time calculation and payment of knowledge points based on the dynamic value score, formally executes the calling of knowledge Token, and stores the key information of this call on the chain for evidence.

[0013] Preferably, the knowledge Token in step one is the smallest callable knowledge unit with unique identification, structured semantics, permission boundary, time limit attribute and value label.

[0014] Preferably, the definition of core metadata field in step one includes:

[0015] A unique identifier for globally unique identification of the knowledge Token; a knowledge ontology encapsulated in machine-readable format, containing structured content of title, body, logical rules and calculation formula;

[0016] A vector representation generated by embedding model and a semantic vector supporting semantic retrieval and similarity matching;

[0017] Anchoring the original data source through a hash algorithm to ensure traceability and tamper-proof source fingerprint;

[0018] An association relationship describing the logical or semantic connection between the knowledge Token and other Tokens;

[0019] A permission control strategy based on attribute model defining multi-dimensional permission rules of subject, object, scene, environment and collaboration required for access and calling;

[0020] A value score calculated based on multi-dimensional indicators;

[0021] A calling interface contract defined by standardized calling definition of input parameters, output format and function behavior;

[0022] A health status indicating whether the knowledge Token is currently valid, expired or needs to be archived;

[0023] Defining the time limit or time attribute of the value decay function of the knowledge Token, describing the collaborative dependency graph of other Token networks relied on by the knowledge Token in the cross-agent collaboration scenario;

[0024] The collaborative dependency graph of other Token networks relied on by the knowledge Token.

[0025] Preferably, in step two, the multiple dimensions of call frequency, business impact degree, confidence, time decay and knowledge collaboration adaptation degree are specifically:

[0026] The call frequency of the number of times the knowledge Token is successfully called within a preset time window;

[0027] The business impact degree obtained by associating key performance indicators and calculating the quantitative contribution value of the knowledge Token to business output using an attribution analysis model;

[0028] The confidence of the content confidence score calculated based on the authority level of the knowledge source, the content cross-validation state, and the consistency of historical call feedback;

[0029] The time decay based on the time attribute of the knowledge Token, calculated by a predefined decay function, is the value decay coefficient generated over time;

[0030] By analyzing the collaborative dependency graph of the knowledge Token and its frequency and role in historical collaboration cases, the knowledge collaboration adaptation degree is evaluated to obtain its applicability score in multi-agent collaboration tasks.

[0031] Preferably, in step two, the multiple attributes based on the subject, object, scene, environment, and collaboration include:

[0032] The subject attribute includes the identity characteristics of the caller, including at least the post role, the department to which it belongs, and the security level;

[0033] The object attribute includes the state of the requested knowledge Token, including at least its value score, health status, and sensitivity label;

[0034] The scene attribute includes the business background of the current call, including at least the project stage, task type, and use purpose;

[0035] The environment attribute includes the technical environment at the time of the call, including at least the network location, device fingerprint, and request time;

[0036] The collaboration attribute includes the context of the current call in the collaborative task, including at least the collaborative session identifier and the collaborative dependency relationship;

[0037] By calculating the value score of each knowledge Token in real time and performing automatic grading, specifically including: if the value score of the knowledge Token is greater than the set score threshold, it is marked as a high-value tradable asset, and the structured metadata of the knowledge Token is automatically embedded;

[0038] If the value score of the knowledge Token is less than the set archiving threshold, it enters the archiving or elimination process.

[0039] Preferably, the multi-element cross-validation of the calling request in step two and the differentiated decision results of direct authorization, integral payment authorization or rejection authorization specifically include:

[0040] S1: Real-time collection of the specific numerical values or states of the subject, object, scene, environment, and collaborative attributes of the attribute set;

[0041] S2: Matching and logical evaluation of the collected attribute set with the permission control strategy embedded in the knowledge Token;

[0042] S3: Perform differentiated decision:

[0043] If the attribute set meets all the mandatory security rules in the permission policy, and the real-time value score of the knowledge Token is higher than or equal to the pre-set high-value threshold, further judge the subject attribute and the scene attribute: if the calling party's job role belongs to the pre-defined priority role set, and the task type of this call is defined as a critical task type, output direct authorization, otherwise, output integral payment authorization;

[0044] If the attribute set does not meet any mandatory security rule in the permission policy, output rejection authorization.

[0045] Preferably, the step three includes the following processes:

[0046] S4: The system loads the permission policy related to the current calling context and matches it with the real-time risk profile of the calling party, completing the final security evaluation before calling;

[0047] S5: Analyze the calling interface contract defined in the knowledge Token, and strictly verify the input parameter format, type and semantics in the calling request according to the contract, to ensure the standardization of programmatic interaction;

[0048] S6: Verify the source fingerprint of the knowledge Token to ensure the authenticity and non-tampering of the data source, and confirm that the health status of all associated Tokens in its collaborative dependency graph is valid;

[0049] S7: According to the real-time dynamic value of the knowledge Token and the type of authorization achieved, the number of knowledge points to be paid is calculated in real time through a pricing model, and the transfer and settlement of the calling party's points are completed;

[0050] S8: After successful payment and settlement, the system formally grants the use right of this call, and triggers the execution environment to complete the call of the knowledge Token;

[0051] S9: The system records the full operation log of this call, including at least the agent ID, operation timestamp, knowledge Token version number and point change information;

[0052] S10: The key behavior information of authorization, settlement and call result is generated into an unalterable evidence by a hash value, and written into a blockchain.

[0053] Preferably, the pricing model of the knowledge points in step S7 of step three is: the number of knowledge points to be paid is equal to the real-time dynamic value of the knowledge Token multiplied by the scene coefficient, the number of knowledge points to be paid is the total number of knowledge points to be paid by the calling party, and the real-time dynamic value is the current dynamic value of the knowledge Token.

[0054] Preferably, in step S2 of step two, the collected attribute set is matched with the permission control strategy embedded in the knowledge Token and logically evaluated, which specifically includes the following steps:

[0055] Verify whether the identity features of the calling party match the subject whitelist defined in the knowledge Token permission strategy;

[0056] Verify whether the business background of this call meets the scene constraint conditions of the permission strategy;

[0057] Verify whether the state of the requested knowledge Token meets the object threshold requirements of the permission strategy;

[0058] Verify whether the technical environment at the time of the call matches the environment rules of the permission strategy;

[0059] Verify whether the context of this call in the collaborative task is consistent with the collaborative dependency graph of the knowledge Token.

[0060] A knowledge base system, comprising:

[0061] The original layer is used to store the original knowledge carrier based on a distributed storage protocol, and to generate a source fingerprint for each knowledge unit through a hash algorithm;

[0062] The activation layer communicates with the original layer and is used to store and manage the structured metadata of the knowledge Token, and responds to the dynamic update of the metadata;

[0063] An index layer in communication with the activation layer, comprising a vector library for supporting similarity retrieval based on semantic vectors and a graph database for maintaining association relationships between knowledge Tokens to support collaborative dependency verification;

[0064] A cross-library adaptation interface layer providing a unified access interface for the original element layer, the activation layer and the index layer, receiving external call requests, and coordinating data of each layer to complete the retrieval, value evaluation and authorization verification process of knowledge Tokens.

[0065] The present application has the following beneficial effects:

[0066] 1. In the present application, by encapsulating the original knowledge unit as a knowledge Token with unique identification, structured semantics and other characteristics, and defining multi-dimensional core metadata fields, a standardized knowledge representation model is constructed, completely solving the problem of traditional knowledge existing in unstructured documents and lacking machine readability, making knowledge directly parsed and programmable called by intelligent agents, and at the same time, integrating multiple dimensions such as call frequency and business impact, calculating knowledge Token value score in real time through dynamic weighting algorithm, replacing traditional subjective static evaluation method, realizing dynamic quantification of knowledge value, solving the pain points of knowledge value evaluation rigidity and lack of incentive, and through automatic grading mechanism, realizing accurate identification of high-value assets and efficient cleaning of low-value assets, greatly improving the management efficiency and utilization value of knowledge assets.

[0067] 2. In the present application, based on the cross-verification mechanism of subject, object, scene, environment and collaborative multi-attribute, fine-grained dynamic authorization is realized, multi-dimensional attribute data is collected in real time, and the embedded permission control strategy in the knowledge Token is accurately matched to output differentiated decisions of direct authorization, integral payment authorization or refusal authorization, which not only meets the efficient authorization demand in key task scenarios, but also builds a secure access bottom line through mandatory security rules, and can flexibly adjust the authorization strategy according to the real-time context of the calling party identity, business scenario and knowledge value, effectively avoiding the problem of excessive or insufficient permissions, while ensuring the security of knowledge assets, greatly simplifying the authorization process, and improving the convenience and adaptability of knowledge calling.

[0068] 3. In the present application, based on the pricing model of real-time dynamic value score of knowledge Token and scene coefficient, fair accounting and real-time settlement of knowledge points are realized, providing quantitative basis for knowledge asset income distribution, in the safety and traceability link, through source fingerprint verification to ensure data authenticity, full operation log record to realize traceability of calling behavior, key information on-chain storage to form an unalterable trust endorsement, through the characteristics of block chain to build a trust foundation for knowledge rights protection, and at the same time, transparent transaction and settlement process is adopted to ensure fair income distribution, while preventing risks such as unauthorized access and data leakage, providing comprehensive support for the safe and efficient circulation of knowledge assets.

[0069] 4. In the present application, the original element layer ensures high availability of original knowledge with distributed storage, and the knowledge right tracing is realized by hash and source fingerprint technology, solving the problems of single-point failure and right tracing difficulty; the activation layer dynamically manages the knowledge Token structured metadata, converts unstructured knowledge into machine-operable structured form, supports real-time metadata update, and improves knowledge management efficiency and timeliness; the index layer realizes semantic-level similarity retrieval through vector library, breaks through the limitations of keyword retrieval, and maintains knowledge association relationship through graph database to support collaborative dependency verification, solving the pain points of inaccurate knowledge retrieval and difficult dependency verification; the cross-database adaptation interface layer provides a unified access interface to coordinate each layer to complete the whole process of retrieval, evaluation and authorization, simplifies external calls, and ensures the coherence of the process. BRIEF DESCRIPTION OF DRAWINGS

[0070] Figure 1 A flowchart of a knowledge Token dynamic value evaluation and transaction authorization method of the present application. DETAILED DESCRIPTION

[0071] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0072] A knowledge Token dynamic value evaluation and transaction authorization method, comprising the following steps:

[0073] Step 1: encapsulate the original knowledge unit as a knowledge Token, and construct a standardized knowledge representation model by defining core metadata fields;

[0074] Step 2: based on the knowledge Token, through the fusion of multiple dimensions such as calling frequency, business impact, confidence, time decay and knowledge collaborative adaptation degree, the value score of each knowledge Token is calculated in real time through a dynamic weighting algorithm, and automatic classification processing is performed; and based on the attributes of subject, object, scene, environment and collaboration, the calling request is cross-verified, and the differential decision results of direct authorization, integral payment authorization or rejection authorization are output;

[0075] Step 3: after obtaining the authorization, the system verifies the authenticity of the interface parameters and data sources, and confirms the validity of the associated Token, the transaction platform completes the real-time accounting and payment of knowledge points according to the dynamic value score, and formally executes the calling of the knowledge Token, and the key information of this calling is stored on the chain for evidence.

[0076] The knowledge Token in step one is the smallest callable knowledge unit with unique identification, structured semantics, permission boundaries, time limit attributes, and value labels.

[0077] In detail, the above-mentioned scheme constructs a standardized knowledge representation model by encapsulating the original knowledge unit as a knowledge Token with the above-mentioned characteristics. The unique identification anchors the source identity of the knowledge Token, providing a reliable basis for authentication and tracking. The structured semantics enable the knowledge content to be directly parsed and processed by intelligent agents, solving the problem of non-structured knowledge being difficult to automate. The setting of permission boundaries realizes dynamic authorization based on multi-dimensional context, overcoming the limitations of traditional static authorization models. The introduction of time limit attributes enables the dynamic reflection of the influence of time factors on knowledge value assessment. The value label provides a quantitative basis for the circulation and income distribution of knowledge assets. These characteristics work together to enable the knowledge Token not only to exist independently as the smallest callable unit, but also to seamlessly connect subsequent value dynamic calculation and safe calling processes, thereby systematically improving the automation level and transaction credibility of knowledge circulation.

[0078] The core metadata fields defined in step one include:

[0079] A unique identifier for globally unique identification of the knowledge Token; a knowledge ontology encapsulated in a machine-readable format, containing structured content such as title, body, logical rules, and calculation formulas;

[0080] A vector representation generated by embedding the model and a semantic vector supporting semantic retrieval and similarity matching;

[0081] A source fingerprint anchored by a hash algorithm to ensure traceability and tamper resistance;

[0082] An association relationship describing the logical or semantic connection between the knowledge Token and other Tokens;

[0083] A permission control strategy based on attribute models defining multi-dimensional permission rules such as subject, object, scene, environment, and collaboration required for access and invocation;

[0084] A value score calculated based on multi-dimensional indicators;

[0085] An invocation interface contract defined by standardizing the invocation definition of input parameters, output formats, and function behavior;

[0086] A health status indicating whether the knowledge Token is currently valid, expired, or needs to be archived;

[0087] A time limit attribute defining the validity period or value decay function of the knowledge Token, describing the cross-agent collaboration scenario;

[0088] a collaborative dependency graph of other Token network relationships on which the knowledge Token depends.

[0089] Specifically, by encapsulating the original knowledge unit as a knowledge Token with the above-mentioned characteristics, a standardized knowledge representation model is constructed, the unique identification ensures the accurate indexing and referencing of the knowledge Token in a distributed environment, the structured encapsulation of the knowledge ontology enables the knowledge content to be automatically parsed and executed by the program, overcoming the defect that traditional unstructured documents cannot support multi-agent automated collaboration, the introduction of semantic vectors improves the accuracy and efficiency of knowledge discovery in cross-agent scenarios, provides tamper-proof traceability evidence for source fingerprints, solves the problems of unknown source and tampering risk of knowledge assets in transactions, the association relationship field constructs the knowledge network topology, enhances the overall adaptability and context awareness of the knowledge system, the permission control strategy realizes fine-grained access control, enables authorized decisions to be dynamically adjusted in combination with real-time context, the value score reflects the business contribution and time-varying changes of the knowledge Token in real time, provides an objective pricing basis for knowledge transactions, the invocation interface contract ensures the standardization of the knowledge Token invocation process, supports programmatic interaction and error prevention between multiple agents, the health status field realizes the automated management of the Token lifecycle, avoids system errors caused by calling invalid or expired knowledge, the time-varying attribute automatically handles the change of knowledge value over time, ensures the real-time and accuracy of value assessment, the collaborative dependency graph supports the verification of dependency integrity in complex tasks, and ensures the continuity and reliability of collaborative task execution. Through the above technical solutions, the knowledge Token can realize machine-readable automated circulation in a multi-agent collaborative environment, support the whole process of dynamic value assessment and transaction authorization;

[0090] In step two, the multiple dimensions of invocation frequency, business impact degree, confidence, time decay, and knowledge collaborative adaptation degree are as follows:

[0091] The invocation frequency of the number of times the knowledge Token is successfully invoked within a preset time window;

[0092] The business impact degree obtained by associating the key performance indicators and calculating the quantitative contribution value of the knowledge Token to business output using an attribution analysis model;

[0093] The confidence of the content confidence score calculated based on the authority level of the knowledge source, the content cross-validation state, and the consistency of historical invocation feedback;

[0094] The time decay based on the time-varying attribute of the knowledge Token, calculated by a predefined decay function, is the value decay coefficient generated over time;

[0095] The knowledge collaboration adaptation degree is evaluated by analyzing the collaborative dependency graph of the knowledge Token and the frequency and role of the knowledge Token in historical collaborative cases.

[0096] Specifically, the definition of call frequency ensures that the frequency statistics are strictly based on actual call data, avoiding subjective estimation or ambiguous counting, making the frequency as a value indicator more reliable and traceable. Secondly, the business impact degree introduces a data-driven attribution analysis mechanism to link the impact of the knowledge Token with specific business indicators, avoiding the bias of relying solely on experience, while ensuring that the contribution value calculation is quantifiable and closely related to the actual business. Thirdly, the confidence level improves the reliability of the credibility score by considering the synergistic effect of multi-source verification and historical feedback, preventing evaluation errors caused by unreliable sources or contradictory content, making the credibility evaluation more robust. In addition, the time decay avoids the rigidity of static evaluation by dynamically adjusting the value score, ensuring that the value score decays reasonably over time, making the timeliness evaluation more in line with the actual changes in the knowledge life cycle. Finally, the knowledge collaboration adaptation degree uses deep analysis of graph structure and historical data to make the adaptation degree evaluation more in line with actual collaborative scenarios, enhancing its applicability in multi-agent environments, ensuring that the collaborative value evaluation is based on real interaction patterns rather than simple assumptions, and achieving objective, dynamic, and quantifiable evaluation of the knowledge Token value score, effectively supporting the demand for precise transaction authorization.

[0097] In step two, the attributes based on the subject, object, scene, environment, and collaboration include:

[0098] The subject attribute includes the identity characteristics of the caller, at least including its job role, department affiliation, and security level.

[0099] The object attribute includes the state of the requested knowledge Token, at least including its value score, health status, and sensitivity label.

[0100] The scene attribute includes the business background of the current call, at least including the project phase, task type, and use purpose.

[0101] The environment attribute includes the technical environment at the time of the call, at least including the network location, device fingerprint, and request time.

[0102] The collaboration attribute includes the context of the current call in the collaborative task, at least including the collaborative session identifier and collaborative dependency relationship.

[0103] By calculating the value score of each knowledge Token in real time and performing automatic classification processing, specifically: if the value score of the knowledge Token is greater than the set score threshold, it is marked as a high-value tradable asset, and the structured metadata of the knowledge Token is automatically embedded.

[0104] If the value score of the knowledge Token is less than the set archiving threshold, the archiving or elimination process is entered.

[0105] Specifically, by defining the specific content of the multi-attribute dimension and setting the automatic grading rules of the value score, the problems of inaccurate authorization verification caused by ambiguous attribute definition and chaotic asset management caused by the lack of grading mechanism are solved. First, the introduction of the subject attribute enables the identity characteristics of the calling party to be converted from abstract description to quantifiable dimension, facilitating accurate identification of the permission boundary in cross-validation. The dynamic index design of the object attribute supports real-time decision-making based on value, ensuring that high-value assets are processed first. The structured parameter design of the scene attribute realizes the dynamic perception of the calling scene, making the authorization decision closely match the actual business needs. The explicit data of the environment attribute enhances the calling security verification capability, effectively preventing illegal access risks. The explicit atlas design of the collaborative attribute supports knowledge adaptation and dependency verification in multi-agent collaborative tasks. By setting the score threshold, when the value score of the knowledge Token is greater than the threshold, it is marked as a high-value tradable asset and embedded with metadata, enabling the identification of high-value assets to be converted from subjective judgment to objective rules, promoting the formation of a knowledge trading incentive mechanism. By setting the archiving threshold, when the value score of the knowledge Token is less than the threshold, it enters the archiving or elimination process, enabling the cleaning of low-value assets to be changed from passive response to active management, optimizing the overall health status and resource efficiency of the knowledge base; effectively solving the problems of insufficient context perception caused by non-specific attribute definition and the lack of high-value asset identification and low-value asset elimination mechanisms caused by the absence of grading thresholds.

[0106] The multi-element cross-validation of the calling request in step two and the differentiated decision-making results of direct authorization, integral payment authorization, or rejection authorization specifically include:

[0107] S1: Real-time collection of attribute sets of specific values or states of subject, object, scene, environment, and collaboration attributes;

[0108] S2: Matching and logical evaluation of the collected attribute sets with the permission control strategy embedded in the knowledge Token;

[0109] S3: Perform differentiated decision-making:

[0110] If the attribute set meets all the mandatory security rules in the permission strategy, and the real-time value score of the knowledge Token is higher than or equal to the pre-set high-value threshold, further judgment of the subject attribute and the scene attribute is performed: if the calling party's job role belongs to the pre-defined priority role set and the task type of this call is defined as a critical task type, direct authorization is output, otherwise, integral payment authorization is output.

[0111] If the attribute set does not satisfy any mandatory security rule in the permission policy, the output is a denial of authorization.

[0112] For example, the priority role set can be defined as: CEO, CTO, general manager of strategic development department, and director of key projects. When the value of the post role field in the subject attribute of the calling request matches any role in the list, it is considered to belong to the priority role set.

[0113] For example, the key task type set can be defined as: annual strategic planning, core technology research, major risk emergency response, and A-level customer bidding. When the value of the task type field in the scenario attribute of the calling request matches any type in the list, and the scenario attribute project stage is the execution stage, the current call is considered to be in the key task type scenario.

[0114] For example, the post role in the subject attribute of the attribute set in the request is product manager.

[0115] The scenario attribute is regular market analysis.

[0116] The real-time value score of the knowledge token is 85, which is higher than the high value threshold of 80.

[0117] If the calling party's post role does not belong to the predefined priority role set, and the task type of the current call is not defined as a key task type, the output is a credit payment authorization.

[0118] Specifically, by collecting the specific values or states of the attribute set of the subject, object, scene, environment, and collaborative multiple attributes in real time, the authorization decision is ensured to be based on the complete dynamic context at the time of invocation, providing a multi-dimensional real-time data basis for cross-validation. Subsequently, the collected attribute set is matched and logically evaluated with the permission control strategy embedded in the knowledge Token, the logical correlation analysis between attributes is realized through the comprehensive verification of mandatory security rules, preventing the vulnerabilities of single-dimensional verification and ensuring the rigor and consistency of the authorization decision. When executing differentiated decisions, if the attribute set meets all the mandatory security rules in the permission strategy and the real-time value score of the knowledge Token is higher than the preset threshold, further judgment is made based on the subject attributes and scene attributes: when the caller's post role belongs to the priority role set and the task type is a critical task type, direct authorization is output, and when the task type is not a critical task type, integral payment authorization is output. By combining dynamic value scoring with business scenario criticality, high-value knowledge can be accessed quickly without payment in core tasks, while introducing an integral constraint mechanism for non-critical scenarios, which not only encourages efficient use of knowledge, but also avoids resource waste. If the attribute set does not meet any mandatory security rule, a refusal authorization is output, strictly implementing the immediate interception of the security bottom line and eliminating the risk of exceeding authority, ensuring the controllability and security of the authorization process in a dynamic environment. Through real-time capture of attribute sets, logical evaluation of strategies, and conditional stratification of decisions, the authorization mechanism can intelligently respond to value changes and business needs, significantly improving the precision and adaptability of knowledge invocation in multi-agent collaboration. By integrating real-time value scoring of knowledge Token and multi-dimensional attribute verification, the problem of permission rigidity caused by the lack of real-time value correlation and scene adaptation in traditional authorization models is solved, effectively addressing the fine-grained and highly flexible access control needs in multi-agent collaboration, thereby realizing a dynamic, precise, and secure knowledge invocation authorization system.

[0119] Step three includes the following processes:

[0120] S4: The system loads the permission strategy related to the current invocation context and matches it with the real-time risk profile of the caller, completing the final security evaluation before invocation;

[0121] S5: Analyze the invocation interface contract defined in the knowledge Token and strictly verify the input parameter format, type, and semantics in the invocation request according to the contract, ensuring the standardization of programmatic interaction;

[0122] S6: Verify the source fingerprint of the knowledge Token to ensure the authenticity and integrity of the data source, and confirm that the health status of all associated Tokens in its collaborative dependency graph is valid;

[0123] S7: According to the real-time dynamic value of the knowledge Token and the type of authorization achieved, the number of knowledge points to be paid is calculated in real time through a pricing model, and the transfer and settlement of the calling party's points are completed;

[0124] S8: After successful payment and settlement, the system formally grants the use right of this call, and triggers the execution environment to complete the call of the knowledge Token;

[0125] S9: The system records the full operation log of this call, including at least the agent ID, operation timestamp, knowledge Token version number and point change information;

[0126] S10: The key behavior information of authorization, settlement and call result is generated into an unalterable evidence through a hash value, and written into a blockchain.

[0127] The blockchain adopts a consortium chain architecture, and the evidence content includes authorization decision hash, point settlement hash and call result hash. The block generation interval is 5 minutes;

[0128] Specifically, by constructing a complete execution process, first, the system loads the permission policy related to the current context before calling, and combines the real-time risk profile of the calling party for fine-grained matching, thereby eliminating security risks caused by environmental changes. Through the analysis of the calling interface contract and parameter verification, it is ensured that the calling request meets the preset rules, avoiding execution errors caused by abnormal parameters. Then, through source fingerprint verification data integrity and checking the health status of associated Tokens in the collaborative dependency graph, the reliability of the knowledge network is ensured. In the payment link, the points are calculated based on the real-time dynamic value and the type of authorization, ensuring fair and instant settlement. After payment, the system triggers the call execution and records the full operation log for traceability. Finally, the evidence is generated through a hash and written into a blockchain, ensuring that the key behavior information is unalterable and verifiable, forming a complete transaction chain. The transaction solves the gap between authorization decision and call execution, significantly improving the security, standardization and auditability of the transaction.

[0129] The pricing model of knowledge points in step S7 of step three is: the number of knowledge points to be paid is equal to the real-time dynamic value of the knowledge Token multiplied by the scenario coefficient. The number of knowledge points is the total number of knowledge points to be paid by the calling party, and the real-time dynamic value is the current dynamic value of the knowledge Token.

[0130] Specifically, by multiplying the number of knowledge points with the real-time dynamic value score and the scene coefficient, the dynamic and scenario-based pricing of knowledge points is realized. First, the introduction of the real-time dynamic value score makes the value assessment of knowledge Token no longer rely on static benchmarks, but dynamically adjust based on the latest data flow, effectively avoiding the value deviation caused by time decay. At the same time, the introduction of the scene coefficient further refines the pricing logic, which can dynamically adjust the payment points according to the demand intensity and business impact of different calling scenarios, thereby improving the flexibility of the pricing mechanism. In addition, it combines the real-time risk portrait of the calling party with the permission strategy matching to ensure the security and standardization of the authorization verification. Furthermore, by embedding the pricing model into the structured metadata of knowledge Token, a reliable data foundation is provided for subsequent blockchain evidence, thereby building a fair, dynamic and quantifiable knowledge transaction incentive mechanism.

[0131] In the S2 step of step two, the collected attribute set is matched and logically evaluated with the permission control strategy embedded in the knowledge Token, which includes the following steps:

[0132] Check whether the identity characteristics of the calling party match the subject whitelist defined in the permission policy of the knowledge Token;

[0133] Check whether the business background of this call meets the scene constraint conditions of the permission policy;

[0134] Check whether the state of the requested knowledge Token meets the object threshold requirements of the permission policy;

[0135] Check whether the technical environment at the time of calling matches the environment rules of the permission policy;

[0136] Check whether the context of this call in the collaborative task is consistent with the collaborative dependency graph of the knowledge Token.

[0137] Specifically, the identity feature verification of the calling party is based on real-time identity information such as post role, department affiliation, and security level, ensuring that the authorization decision can be accurately matched according to dynamic identity information rather than relying on static role groups, thereby avoiding the problem of excessive or insufficient permissions caused by identity generalization in traditional authorization models. Secondly, the business context verification associates dynamic scenario elements such as project phase, task type, and use purpose, enabling the authorization decision to adapt to complex business context changes and addressing the limitations of static scenario constraints in responding to diverse task requirements. The self-state verification of the requested knowledge Token combines threshold judgment of object attributes such as value score, health status, and sensitivity label to ensure that high-value assets are protected in a targeted manner, optimizing the allocation efficiency of knowledge resources. At the same time, the technical environment verification uses environmental parameters such as network location, device fingerprint, and request time to verify security, enhancing the ability to prevent potential technical risks. Finally, the collaborative task context verification compares the collaborative session identifier and dependency relationship with the collaborative dependency graph of the knowledge Token, ensuring the logical consistency of knowledge invocation in multi-agent collaboration and reducing permission conflicts in collaborative tasks.

[0138] A knowledge base system, comprising:

[0139] An original layer for storing original knowledge carriers based on a distributed storage protocol and generating source fingerprints for each knowledge unit through a hash algorithm;

[0140] An activation layer in communication with the original layer for storing and managing structured metadata of knowledge Tokens and responding to dynamic updates of the metadata;

[0141] An index layer in communication with the activation layer, including a vector library for supporting similarity retrieval based on semantic vectors and a graph database for maintaining the association relationship between knowledge Tokens to support collaborative dependency verification;

[0142] A cross-library adaptation interface layer providing a unified access interface for the original layer, the activation layer, and the index layer, receiving external call requests, and coordinating data among the layers to complete the retrieval, value assessment, and authorization verification process of knowledge Tokens.

[0143] Specifically, the original layer stores the original knowledge carrier based on a distributed storage protocol, and generates a source fingerprint for each knowledge unit through a hash algorithm, ensuring the authenticity and non-tamperability of the data source, providing a trusted basis for the knowledge Token, solving the problems of difficult rights protection and anti-tampering; the activation layer communicates with the original layer, stores and manages the structured metadata of the knowledge Token, and responds to the dynamic update of the metadata, which makes the knowledge have structured semantics and permission boundaries, supports real-time calculation of value score and dynamic adjustment of permission policy, solves the problems of strong subjectivity of static evaluation and rigidification of authorization mechanism; the index layer communicates with the activation layer, including the vector library and the graph database, the vector library supports similarity retrieval based on semantic vectors, and the semantic vectors generated by the embedding model realize semantic search and matching of knowledge, solve the problems of lack of machine readability and semantic encapsulation, the graph database maintains the association relationship between knowledge Tokens, verifies the applicability of knowledge in multi-agent tasks through collaborative dependency graph, supports fine-grained authorization and collaborative adaptation evaluation; the cross-library adaptation interface layer provides a unified access interface for the original layer, the activation layer and the index layer, receives external call requests, and coordinates the data of each layer to complete the retrieval, value evaluation and authorization verification process of the knowledge Token, realizes the closed-loop design of transaction and authorization, simplifies the calling process, ensures safe calling and full-link audit, solves the problems of complicated authorization process and imperfect security calling guarantee system.

[0144] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any such actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0145] Although the embodiments of the present application have been shown and described, it should be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made without departing from the principles and spirit of the present application, and the scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for dynamic value assessment and transaction authorization of knowledge tokens, characterized in that, Includes the following steps: Step 1: Encapsulate the original knowledge units into knowledge tokens, and build a standardized knowledge representation model by defining core metadata fields; Step 2: Based on knowledge tokens, by integrating multiple dimensions such as call frequency, business impact, confidence level, time decay, and knowledge collaboration adaptability, a value score for each knowledge token is calculated in real time using a dynamic weighted algorithm, and automatic classification is performed; and based on multiple attributes such as subject, object, scenario, environment, and collaboration, the call request is cross-validated with multiple factors, and differentiated decision results such as direct authorization, points payment authorization, or denial of authorization are output. Step 3: After obtaining authorization, the system verifies the authenticity of the interface parameters and data source, confirms the validity of the associated token, and the trading platform completes the real-time calculation and payment of knowledge points based on the dynamic value score, and officially executes the call of knowledge token, and puts the key information of this call on the blockchain for evidence storage.

2. The method for dynamic value assessment and transaction authorization of knowledge tokens according to claim 1, characterized in that, In step one, the knowledge token is the smallest callable knowledge unit that has a unique identifier, structured semantics, permission boundaries, timeliness attributes, and value tags.

3. The method for dynamic value assessment and transaction authorization of knowledge tokens according to claim 2, characterized in that, The core metadata fields defined in step one include: A unique identifier used to globally and uniquely identify this knowledge token; a knowledge ontology encapsulated in a machine-readable format, containing structured content including title, body, logical rules, and calculation formulas; Vector representations generated through embedding models and semantic vectors that support semantic retrieval and similarity matching; By anchoring the original data source through hash algorithms, a traceable and tamper-proof source fingerprint is ensured; Describe the logical or semantic connections between this knowledge token and other tokens; An access control strategy is based on an attribute model that defines multi-dimensional permission rules for access and invocation, including the subject, object, scenario, environment, and collaboration. A value score based on dynamic value scores calculated using multidimensional indicators; A standardized calling interface contract that clearly defines input parameters, output formats, and function behavior; The health status of the knowledge token, indicating whether it is currently valid, expired, or needs to be archived. Define the validity period or value decay function of knowledge tokens and describe their time-related attributes in cross-agent collaborative scenarios; A collaborative dependency graph of the network relationships of other tokens upon which the knowledge token depends.

4. The method for dynamic value assessment and transaction authorization of knowledge tokens according to claim 3, characterized in that, In step two, the multiple dimensions of call frequency, business impact, confidence level, timeliness decay, and knowledge collaboration adaptability are as follows: The frequency of successful calls to the knowledge token within a preset time window; The business impact is the quantitative contribution of the knowledge token to business output, calculated by associating it with key performance indicators and using an attribution analysis model. The confidence level of the content credibility score is calculated based on the authority level of the knowledge source, the cross-validation status of the content, and the consistency of historical call feedback. Based on the time-sensitive attribute of knowledge tokens, the time-sensitive decay is calculated using a predefined decay function, which represents the value decay coefficient that occurs over time. By analyzing the collaborative dependency graph of knowledge tokens and their frequency of occurrence and roles in historical collaborative cases, the knowledge collaboration fit of the resulting applicability score in multi-agent collaborative tasks is evaluated.

5. The method for dynamic value assessment and transaction authorization of knowledge tokens according to claim 4, characterized in that, Step two, based on multiple attributes including subject, object, scene, environment, and collaboration, specifically includes: The main attributes include the caller's identity characteristics, including at least their job role, department, and security level; The object attributes include the state of the requested knowledge token itself, including at least its value rating, health status, and sensitivity label; The scenario attributes include the business context in which this call occurs, including at least the project stage, task type, and purpose of use; Environmental attributes include the technical environment at the time of the call, including at least network location, device fingerprint, and request time; The collaboration attributes include the context of this call within the collaboration task, and at least include the collaboration session identifier and collaboration dependencies; The value score of each knowledge token is calculated in real time and automatically graded. Specifically, if the value score of a knowledge token is greater than the set score threshold, it is marked as a high-value tradable asset and automatically embedded in the structured metadata of the knowledge token. If the value score of the knowledge token is less than the set archiving threshold, it will enter the archiving or elimination process.

6. The method for dynamic value assessment and transaction authorization of knowledge tokens according to claim 5, characterized in that, Step two involves performing multi-factor cross-validation on the call request and outputting differentiated decision results for direct authorization, points-based payment authorization, or authorization denial. Specifically, this includes: S1: Real-time collection of the specific values ​​or states of multiple attributes of the subject, object, scene, environment, and collaboration; S2: Match and logically evaluate the collected attribute set with the permission control policy embedded in the knowledge token; S3: Implement differentiated decisions: If the attribute set satisfies all mandatory security rules in the permission policy, and the real-time value score of the knowledge token is higher than or equal to the preset high value threshold, then further determine the subject attribute and scenario attribute: if the caller's job role belongs to the predefined priority role set, and the task type of this call is defined as a critical task type, then output direct authorization; otherwise, output points payment authorization. If the attribute set does not satisfy any of the mandatory security rules in the permission policy, then the output "Authorization Denied" will be output.

7. The method for dynamic value assessment and transaction authorization of knowledge tokens according to claim 6, characterized in that, Step three includes the following process: S4: The system loads the permission policy related to the current call context and matches it with the caller's real-time risk profile to complete the final security assessment before the call. S5: Parse the call interface contract defined in the knowledge token, and strictly verify the format, type and semantics of the input parameters in the call request according to the contract to ensure the standardization of programmatic interaction; S6: Verify the source fingerprint of the knowledge token to ensure the authenticity of the data source and that it has not been tampered with, and confirm that the health status of all associated tokens in its collaborative dependency graph is valid; S7: Based on the real-time dynamic value of the knowledge token and the type of authorization achieved, calculate the amount of knowledge points to be paid in real time through the pricing model, and complete the transfer and settlement of points in the caller's account. S8: After successful payment settlement, the system officially grants the right to use this call and triggers the execution environment to complete the call to the knowledge token; S9: The system records the full operation log of this call, including at least the agent ID, operation timestamp, knowledge token version number, and points change information; S10: Generate immutable evidence of key actions related to authorization, settlement, and invocation results using hash values, and write it into the blockchain.

8. The method for dynamic value assessment and transaction authorization of knowledge tokens according to claim 7, characterized in that, The pricing model for knowledge points in step S7 of step three is as follows: the number of knowledge points to be paid is equal to the real-time dynamic value score of the knowledge token multiplied by the scenario coefficient. The number of knowledge points is the total number of knowledge points that the caller needs to pay, and the real-time dynamic value score is the current dynamic value score of the knowledge token.

9. The method for dynamic value assessment and transaction authorization of knowledge tokens according to claim 8, characterized in that, In step S2 of step two, the collected attribute set is matched and logically evaluated with the permission control policy embedded in the knowledge token, specifically including the following steps: Verify whether the caller's identity characteristics match the whitelist of entities defined in the knowledge token permission policy; Verify whether the business context in which this call occurred meets the scenario constraints of the permission policy; Verify whether the state of the requested knowledge token meets the object threshold requirements of the permission policy; Verify whether the technical environment at the time of the call matches the environment rules of the permission policy; Verify whether the context of this call in the collaborative task is consistent with the collaborative dependency graph of the knowledge token.

10. A knowledge base system for implementing the dynamic value assessment and transaction authorization method for knowledge tokens as described in claims 1-9, characterized in that, include: The original knowledge layer is used to store the original knowledge carriers based on a distributed storage protocol and to generate source fingerprints for each knowledge unit using a hash algorithm. The activation layer, which communicates with the component layer, is used to store and manage the structured metadata of the knowledge token and respond to dynamic updates of the metadata. An index layer, which communicates with the activation layer, includes a vector library and a graph database. The vector library is used to support similarity retrieval based on semantic vectors, and the graph database is used to maintain the association between knowledge tokens to support collaborative dependency verification. The cross-library adaptation interface layer provides a unified access interface for the original layer, activation layer and index layer, receives external call requests, and coordinates the data of each layer to complete the knowledge token retrieval, value assessment and authorization verification process.

Images