Search

CN-121998097-A - Cognitive reasoning model feature evaluation method, device, equipment, medium and system based on distributed account book

CN121998097ACN 121998097 ACN121998097 ACN 121998097ACN-121998097-A

Abstract

The application provides a cognitive reasoning model characteristic evaluation method, device, equipment, medium and system based on a distributed account book. The method comprises the steps of obtaining a target cognitive reasoning model, mapping and packaging the target cognitive reasoning model into an on-chain digital certificate entity in a distributed account book network, synchronizing execution characteristic data of the target cognitive reasoning model in an operation process to the distributed account book network by utilizing a predictor component, processing the execution characteristic data by utilizing an evaluation contract operator deployed in the distributed account book network, calling a preset dynamic evaluation algorithm to generate a characteristic evaluation value aiming at the on-chain digital certificate entity, and executing numerical rewriting on a property mapping parameter by utilizing an intelligent contract based on the characteristic evaluation value in response to a property state migration trigger signal so as to finish state migration processing aiming at the target cognitive reasoning model. The application relieves the technical problem of disjoint of the execution characteristics and the update of the asset state in the traditional evaluation scheme.

Inventors

  • Request for anonymity

Assignees

  • 北京认知涌现科技有限公司

Dates

Publication Date
20260508
Application Date
20260130

Claims (10)

  1. 1. The cognitive reasoning model characteristic evaluation method based on the distributed account book is characterized by comprising the following steps of: step 1, acquiring a target cognitive reasoning model, mapping and packaging the target cognitive reasoning model into an on-chain digital credential entity in a distributed account book network; Step 2, using an evaluation contract operator deployed in the distributed ledger network, invoking a preset dynamic evaluation algorithm to process the execution characteristic data so as to generate a characteristic evaluation score for the on-chain digital credential entity; And 3, monitoring a property state transition trigger signal in the distributed account network, responding to the property state transition trigger signal, invoking property mapping parameters associated with the on-chain digital certificate entity by utilizing an intelligent contract, and executing numerical rewriting on the property mapping parameters based on the characteristic evaluation value to complete state transition processing aiming at the target cognitive reasoning model.
  2. 2. The distributed ledger-based cognitive reasoning model feature assessment method of claim 1, wherein the execution feature data is collected from an under-chain environment by the predictor component, including behavior trace components of the target cognitive reasoning model invoked by an external agent, execution performance indicators for preset tasks, and associated environmental feedback parameters.
  3. 3. The cognitive inference model feature assessment method based on the distributed ledger of claim 1, wherein the step 2 specifically includes: Performing association mapping on the behavior track component, the execution efficacy index and the environment feedback parameter by using the evaluation contract operator based on marginal contribution attribution analysis logic to determine characteristic association contribution weight of the target cognitive reasoning model in a multi-model collaborative scene; And performing weighted calculation by combining the initial attribute scores of the target cognitive reasoning model and the characteristic association contribution weights so as to generate the characteristic evaluation scores.
  4. 4. The cognitive inference model feature assessment method based on a distributed ledger of claim 1, wherein the mapping and packaging of the feature assessment method as an on-chain digital voucher entity in a distributed ledger network specifically comprises: extracting unique identification fingerprints and model metadata of the target cognitive reasoning model; and solidifying the unique identification fingerprint and the model metadata in the distributed ledger network by using a non-homogeneous universal certification protocol to generate the corresponding on-chain digital certification entity.
  5. 5. The method for evaluating features of a cognitive reasoning model based on a distributed ledger as claimed in claim 1, wherein the step of synchronizing the execution feature data of the target cognitive reasoning model in the running process to the distributed ledger network by using a predictor component specifically comprises: acquiring operation original records aiming at the target cognitive reasoning model from a plurality of downlink data sources respectively by utilizing a plurality of independent verification nodes in the predictor component; and executing multiparty consistency check on the running original record to generate the execution characteristic data with the trusted memory card identifier.
  6. 6. The cognitive inference model feature assessment method based on a distributed ledger of claim 1, wherein the step 2 further comprises: Acquiring a history evaluation record stored in the distributed account network, and calling a preset time attenuation operator; and performing weighted attenuation mapping on the historical evaluation record by utilizing the time attenuation operator to generate an effective characteristic component representing the logical evolution state of the target cognitive reasoning model, and correcting the characteristic evaluation value based on the effective characteristic component.
  7. 7. The utility model provides a cognitive reasoning model characteristic evaluation device based on distributed account book which characterized in that includes: The voucher mapping module is used for acquiring a target cognitive reasoning model, mapping and packaging the target cognitive reasoning model into an on-chain digital voucher entity in the distributed account network; The propulsor synchronization module is used for synchronizing the execution characteristic data of the target cognitive reasoning model in the running process to the distributed ledger network by utilizing a propulsor component; the score evaluation module is used for calculating and generating a characteristic evaluation score aiming at the digital certificate entity on the chain by using an evaluation contract operator deployed in the distributed ledger network; and the state migration module is used for monitoring the ownership state migration trigger signal and responding to the ownership state migration trigger signal, and performing numerical rewriting on the ownership mapping parameters associated with the on-chain digital certificate entity by utilizing an intelligent contract based on the feature evaluation value.
  8. 8. An electronic device comprising a memory for storing a computer program and a processor for implementing the steps of the method according to any one of claims 1-6 when the program is executed.
  9. 9. A computer-readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the steps of the method according to any one of claims 1-6.
  10. 10. A cognitive reasoning model feature assessment system based on a distributed ledger, comprising: the cognitive reasoning model layer is used for operating the target cognitive reasoning model and generating an operation record; The predictor gateway is used for converting the operation record into trusted execution characteristic data by utilizing a predictor component and synchronizing the trusted execution characteristic data to the distributed ledger network; the distributed account book network is used for maintaining the on-chain digital certificate entity generated by the target cognitive reasoning model mapping and deploying the assessment contract operator and the intelligent contract; The intelligent contract is used for monitoring the ownership state transition trigger signal and executing rewriting processing on the ownership mapping parameters associated with the on-chain digital certificate entity based on the characteristic evaluation value.

Description

Cognitive reasoning model feature evaluation method, device, equipment, medium and system based on distributed account book Technical Field The application relates to the technical field of distributed account books and artificial intelligence, in particular to a cognitive reasoning model feature evaluation method, device, equipment, medium and system based on the distributed account books. Background With the popularity of artificial intelligence technology, various cognitive inference models (e.g., industry expert models, strategy generation models) have become the central digital asset. In a cross-platform and cross-mechanism intelligent agent cooperative scene, as the value of the cognitive reasoning model has high dynamic property and evolution property, how to execute higher fidelity quantitative evaluation on the contribution degree of the cognitive reasoning model in an actual task and realize the synchronous streaming of the ownership state according to the quantitative evaluation, thus the cognitive reasoning model has become a key requirement for the development of a distributed cognitive scene. In the existing cognitive model management scheme, a centralized database is generally adopted to record call logs and performance data of a model. The scheme comprises the steps of firstly distributing unique asset identifications for each model in a centralized server, establishing an associated performance statistical table, then intercepting a call request and recording an execution result by a centralized gateway when the models are called externally, and finally manually updating reputation scores or ownership information of the models in a database by a manager according to the periodically summarized statistical report. However, this centralization scheme has obvious technical drawbacks. Because the call log is stored in the private server of a single entity, the authenticity and integrity of the execution feature data are difficult to be verified by public trust of cross-institutions, and the risk of data tampering is easily faced. Meanwhile, as the contribution of the cognitive model in the cooperative task often has complex nonlinear characteristics, a simple summary report is difficult to establish real-time technical association between the execution characteristics and the asset value, so that hysteresis exists in the update of the ownership state, an automatic closed-loop verification mechanism is lacking, and the cognitive model cannot adapt to the cognitive asset assessment requirements of high-frequency and dynamic evolution. Disclosure of Invention In order to solve the technical problems, the application provides a cognitive reasoning model characteristic evaluation method, device, equipment, medium and system based on a distributed ledger, so as to at least alleviate the technical problems. A cognitive reasoning model characteristic evaluation method based on a distributed account book comprises the following steps: step 1, acquiring a target cognitive reasoning model, mapping and packaging the target cognitive reasoning model into an on-chain digital credential entity in a distributed account book network; Step 2, using an evaluation contract operator deployed in the distributed ledger network, invoking a preset dynamic evaluation algorithm to process the execution characteristic data so as to generate a characteristic evaluation score for the on-chain digital credential entity; And 3, monitoring a property state transition trigger signal in the distributed account network, responding to the property state transition trigger signal, invoking property mapping parameters associated with the on-chain digital certificate entity by utilizing an intelligent contract, and executing numerical rewriting on the property mapping parameters based on the characteristic evaluation value to complete state transition processing aiming at the target cognitive reasoning model. Optionally, the execution characteristic data is acquired from an under-chain environment through the predictor component, and the execution characteristic data comprises a behavior track component called by an external agent program by the target cognitive inference model, an execution efficiency index for a preset task and an associated environment feedback parameter. Optionally, the step 2 specifically includes performing association mapping on the behavior trace component, the execution effectiveness index and the environment feedback parameter by using the evaluation contract operator based on marginal contribution attribution analysis logic to determine feature association contribution weights of the target cognitive reasoning model in a multi-model collaboration scene, and performing weighted calculation by combining initial attribute scores of the target cognitive reasoning model and the feature association contribution weights to generate the feature evaluation scores. Optionally, the mapping and packaging of the unique