Learning process verifiable evidence storage and authentication method and system and storage medium

Abstract

The invention discloses a verifiable evidence storage and authentication method and system for a learning process and a storage medium, and relates to the technical field of computers and education, and the method comprises the specific steps: capturing the learning behavior data of a learner in real time, and generating a structured learning event record; performing multi-dimensional verification processing on the learning event; carrying out batch aggregation on the verified learning events according to a preset rule, generating an aggregation proof by using a cryptographic accumulator, and anchoring the aggregation proof to a tamper-resistant distributed account book; generating a dynamic learning resume through an event aggregation algorithm based on the stored learning event, and storing the dynamic learning resume in a distributed storage system; automatically evaluating the dynamic learning resume according to a preset multi-dimensional authentication standard to generate an authentication result; and generating a digital certificate meeting the verifiable certificate standard through an intelligent contract or a credible signing and issuing service based on the authentication result. According to the invention, a reliable learning process and result proof are provided for the learner for lifetime, and the education and occupational development opportunities are improved.

Description

Technical Field

[0001] This invention relates to the fields of computer technology and education technology, and more specifically to a method, system and storage medium for verifiable evidence storage and authentication of learning processes. Background Technology

[0002] Currently, with the deepening global digital transformation and the knowledge economy, lifelong learning has become an inevitable requirement for social and economic development. The exponential progress of artificial intelligence technology has accelerated the updating and iteration of knowledge, driving the learning paradigm to shift from centralized, phased school education to distributed, continuous, and self-directed learning. However, this shift makes it difficult for a large number of off-campus self-directed learning outcomes to obtain widely recognized third-party certification outside the traditional education system, leading to a severe challenge for learners facing a break in the "learning-certification-employment / further education" chain when pursuing further education or employment.

[0003] The industry has explored various technical solutions to address this authentication challenge, including: traditional digital credential technologies (such as PDF certificates or digital signature certificates based on public key infrastructure), which are easily forged and tampered with and rely on centralized institutions, posing a single point of failure risk; early online learning credential standards (such as Open Badges), which only provide static assertions about learning outcomes, with verification relying on the continuous online presence of the issuing server and failing to verify the learning process; blockchain-based academic / certificate storage technologies, which only store learning outcomes on the chain to ensure immutability, but completely ignore the verification of the authenticity of the learning process, failing to prevent cheating, proxy learning, and other dishonest behaviors; learning process data recording and analysis standards (such as the Experience Application Programming Interface (xAPI) and the learning analytics standard Caliper), which, while able to record learning activities in detail, lack anti-tampering mechanisms, allowing data to be generated or modified arbitrarily, resulting in insufficient credibility; and decentralized identity and verifiable credential standards (such as the W3C's DID and VC frameworks), which provide a general digital trust infrastructure but do not specify the generation logic of credential content, failing to guarantee the authenticity and credibility of the learning achievements themselves. These existing technical solutions all have significant flaws, namely, the separation of process and result, lack of continuous identity verification, data silos and inconsistent standards, and the contradiction between privacy and verifiability, thus failing to fundamentally solve the trust problem of autonomous learning outcomes.

[0004] Therefore, how to design a method and system that can deeply bind the learning process and the results, and has high credibility and can be widely recognized by society, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a method, system and storage medium for verifiable evidence storage and authentication of the learning process, overcoming the above-mentioned defects.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The first aspect discloses a verifiable evidence storage and authentication method for a learning process, the specific steps of which are as follows: S1. Capture learners' learning behavior data on the learning platform in real time through embedded acquisition modules or standard protocol interfaces, and generate structured learning event records. Learning events include at least learner identifier, behavior type, behavior object, timestamp, and context information. S2. Perform at least one verification process on the learning event, the verification process including but not limited to one or more combinations of identity verification, behavior authenticity verification, content originality verification, and environment credibility verification; S3. Aggregate multiple verified learning events in batches according to preset rules, generate an aggregation proof using a cryptographic accumulator, and anchor the aggregation proof to an immutable distributed ledger. S4. Based on the evidence-based learning events, a dynamic learning history is generated through an event aggregation algorithm, and the dynamic learning history is stored in a distributed storage system; S5. Based on preset multidimensional authentication standards, automatically evaluate the dynamic learning resume and generate authentication results; S6. Based on the authentication result, generate a digital certificate that conforms to the verifiable credential standard through a smart contract or trusted issuance service, and record the credential metadata in the distributed ledger.

[0007] The second aspect discloses a verifiable evidence storage and authentication system for the learning process, including: The learning application layer is used to interact with learners and capture learning behavior data. The learning application layer includes a general course platform adapter, which supports at least one of the following: Learning Tool Interoperability Protocol (LTI), Experience Application Programming Interface (xAPI), Learning Analytics Standard (Caliper), and Shared Content Object Reference Model (SCORM). An edge computing layer, deployed on learner client devices, is used to preprocess, de-identify, or cache the learning behavior data offline, and generate learning events. The event processing layer is used to perform multidimensional verification and batch processing on received learning events, and to generate aggregate proofs using a cryptographic accumulator, anchoring the aggregate proofs to an immutable distributed ledger; a dynamic learning history is generated based on the aggregate proofs, and the multidimensional verification includes identity verification, behavior authenticity verification, content originality verification, and environment trustworthiness verification; wherein, the aggregate proofs include formative evaluation notarization and reflective log notarization. The authentication and service layer is used to monitor the stored records on the distributed ledger, generate dynamic learning resumes based on the stored learning events, perform automated authentication and evaluation of the dynamic learning resumes according to the preset multi-dimensional authentication standards, and issue digital certificates that conform to the verifiable credential standard through smart contracts or trusted issuance services. The data and storage layer is used to store learning events and dynamic learning history; Security and privacy layer, used to provide zero-knowledge proof generation, selective disclosure control, or key management services.

[0008] The third aspect discloses a method for constructing a decentralized trust system, the specific steps of which are as follows: A decentralized identity (DID) is created for the learner, which is controlled by the learner and does not depend on any centralized institution. Associate the learner's learning events, dynamic learning history, and digital certificate with the decentralized identity identifier (DID); The credential metadata of the digital certificate is recorded in a distributed ledger, enabling the digital certificate to be verified independently.

[0009] The fourth aspect discloses a continuous authentication method, the specific steps of which are as follows: Initial identity authentication: At the start of the learning session, the learner's identity is confirmed through facial recognition, document verification, or multi-factor authentication; Periodic identity re-authentication: During the learning process, identity verification is performed periodically at unpredictable time intervals through random facial verification or challenge-response mechanisms. Continuous behavioral feature verification: Throughout the learning process, learners’ behavioral biometric data are continuously collected and compared with the learners’ historical behavioral feature baseline model to assess the consistency of identity in real time. The behavioral biometric data includes at least one of keystroke dynamics features, mouse movement trajectory features, or touch screen interaction features.

[0010] The fifth aspect discloses a formative evaluation evidence preservation method, the specific steps of which are as follows: Obtain response data related to formative assessment events; Based on the answer data, evaluation metadata is collected, which includes evaluation type, related knowledge points, difficulty level and scoring criteria. Record learners' answer process data, which includes the response content, response time, and answer order for each question; Calculate and generate evaluation result data, which includes the total score, the score for each knowledge point, and the mastery level assessment result; Record evidence data of answering behavior, including the distribution of answering time, the number of times answers were modified, and the number of times paused; A question-answering response record is constructed based on the question-answering process data, the evaluation result data, and the question-answering behavior evidence data. An incremental hash chain structure is adopted, in which each answer response record is linked with the previous answer response record to form a hash chain, and the final hash value of the incremental hash chain is anchored to an immutable distributed ledger.

[0011] The sixth aspect discloses a method for preserving reflection logs, and the specific steps for preserving reflection logs are as follows: Obtain the answer data related to the reflection log event; Based on the answer data, a reflection trigger context is recorded, which includes the trigger type and associated learning events; The structured reflection content is collected based on the answer data, and the structured reflection content includes descriptive dimension, emotional dimension, cognitive dimension, evaluation dimension and planning dimension; The structured reflection content is evaluated based on reflection quality indicators; this includes an automated evaluation of the depth level of the structured reflection content based on a preset cognitive hierarchy model. The reflection content is encrypted and stored, and the reflection quality assessment results and content summaries are also preserved as evidence.

[0012] The seventh aspect discloses a storage medium on which a computer program is stored, which, when executed by a processor, implements the above-mentioned learning process to verify the evidence storage and authentication method.

[0013] As can be seen from the above technical solutions, the present invention discloses a method, system, and storage medium for verifiable evidence storage and authentication during the learning process, which has the following advantages compared with the prior art: 1. Comprehensively enhances the credibility of the learning process: This invention expands the evidence storage object from the static "learning result" to the dynamic "learning process". Through real-time collection, multi-dimensional verification and batch evidence storage, it ensures that every key link in the learning process is traceable and verifiable, fundamentally solving the trust deficiency of "results on the chain, process black box", and making the credibility of learning credentials achieve a qualitative leap.

[0014] 2. Significantly enhances anti-counterfeiting capabilities: This invention introduces a three-layer progressive continuous identity verification mechanism (initial authentication → periodic re-authentication → continuous behavioral feature verification) and behavioral authenticity verification, ensuring the authenticity, integrity, and non-repudiation of learning data from the source. It effectively solves academic misconduct issues such as proxy learning and proxy exams in online learning scenarios. In particular, the continuous verification based on behavioral biometrics such as keystroke dynamics and mouse trajectory can achieve implicit and continuous confirmation of identity without disturbing the learner.

[0015] 3. Promotes an open and interoperable educational ecosystem: This invention designs a universal course platform adapter architecture, a multi-language SDK, and a developer sandbox, supporting mainstream learning technology standards such as xAPI, LTI, Caliper, and SCORM. It can seamlessly connect to any existing or newly developed online course platform, breaking down data silos and enabling cross-platform and cross-institutional integration and authentication of learning records, building a complete and verifiable lifelong learning portfolio for learners. Simultaneously, support for multiple digital credential standards ensures the interoperability of credentials issued by this system globally, breaking down barriers in the field of educational certification.

[0016] 4. Advanced privacy protection: This invention deeply integrates advanced privacy protection technologies such as zero-knowledge proof (ZKP) and selective disclosure (SD), allowing learners to prove to third parties that they meet specific learning achievement or ability requirements without exposing detailed learning process data; it can protect personal privacy while realizing the verifiability of learning achievements, and solve the contradiction between privacy and verifiability in the prior art.

[0017] 5. Achieves autonomous control over learner data: This invention uses distributed ledger technology as a trust anchor, combined with W3C DID and VC standards, to build a decentralized learning authentication trust network. Learners have complete autonomy over their learning data and digital credentials, no longer relying on any single centralized institution. Even if the issuing institution disappears, the learner's credentials can still be independently verified and have long-term validity.

[0018] 6. Significantly reduced evidence storage costs and improved system performance: This invention employs batch aggregation and cryptographic accumulator technologies to efficiently compress massive amounts of learning events into compact aggregated proofs before uploading them to the blockchain. This significantly reduces blockchain storage and computation costs, enabling the system to support large-scale learning evidence storage needs globally and resolving the performance bottleneck of blockchain in high-frequency scenarios such as education. Simultaneously, the six-layer architecture design and microservice deployment ensure high availability and elastic scalability of the system.

[0019] 7. A complete and traceable lifelong learning portfolio has been constructed: This invention not only preserves learning outcomes, but more importantly, records the entire learning process, including formative assessments and reflection logs. This provides a solid data foundation for achieving truly personalized education, learning path recommendations, and ability development predictions. Unlike traditional "data silos" scattered across various platforms, the dynamic learning resume constructed by this invention is a unified and reliable digital asset that can accompany learners throughout their lives.

[0020] 8. Enhanced System Resilience and Observability: By introducing advanced cloud-native architecture practices such as multi-active data centers, chaos engineering, and unified observability, the system built by this invention possesses extremely high availability and fault recovery capabilities. This ensures the continuous stability of the learning and evidence storage service, providing a guarantee for large-scale applications. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the overall architecture of the learning process verifiable evidence storage and authentication system based on distributed ledger of the present invention; Figure 2 This is a schematic diagram of the W3C verifiable credential data structure of the present invention; Figure 3 This is a schematic diagram of the general course platform adapter architecture of the present invention; Figure 4 This is a flowchart of the continuous authentication process of the present invention; Figure 5 This is a flowchart illustrating the zero-knowledge proof and selective disclosure process of this invention. Figure 6 A detailed flowchart illustrating the learning process of this invention; Figure 7 This is a flowchart of the learning resume construction and certificate issuance process of the present invention; Figure 8 This is a flowchart of the third-party verification process for the present invention; Figure 9 This is a schematic diagram of the formative evaluation incremental hash chain of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] This invention discloses a verifiable proof and authentication method, system, and storage medium for learning processes, aiming to solve the problem of lack of trusted third-party authentication in informal and self-study processes. It captures and verifies learning events in real time through a multi-dimensional verification engine, efficiently aggregates massive amounts of events using a cryptographic accumulator, and anchors the aggregation results to a distributed ledger, ensuring the immutability and traceability of the learning process. The system uses W3C DID and VC standards to construct learner digital identities and verifiable credentials, achieving seamless integration with various online courses through a universal course platform adapter. Furthermore, this embodiment introduces privacy-preserving technologies such as zero-knowledge proofs and selective disclosure, allowing learners to prove their learning achievements to third parties (such as schools and employers) while protecting their personal privacy. This invention also constructs an open, trusted, and privacy-preserving global learning authentication ecosystem, providing lifelong learners with reliable proof of their learning processes and results, greatly enhancing their educational and career development opportunities. Example 1

[0025] One embodiment of the present invention discloses a method for verifiable evidence storage and authentication of a learning process, the specific steps of which are as follows: S1. Capture learners' learning behavior data in real time through embedded acquisition modules or standard protocol interfaces, and generate structured learning event records. Learning events include at least learner identifier, behavior type, behavior object, timestamp, and context information. S2. The learning event is verified, including but not limited to one or more combinations of identity verification, behavior authenticity verification, content originality verification, and environment credibility verification. S3. Aggregate multiple verified learning events in batches according to preset rules, generate an aggregation proof using a cryptographic accumulator, and anchor the aggregation proof to an immutable distributed ledger. S4. Based on the evidence-based learning events, a dynamic learning history is generated through an event aggregation algorithm, and the dynamic learning history is stored in a distributed storage system; S5. Based on preset multidimensional authentication standards, automatically evaluate the dynamic learning resume and generate authentication results; S6. Based on the authentication result, generate a digital certificate that conforms to the verifiable credential standard through a smart contract or trusted issuance service, and record the credential metadata of the digital certificate in the distributed ledger.

[0026] In one embodiment, S1 is implemented through a general adapter architecture that supports at least one mainstream learning technology standard, including the Learning Tool Interoperability Protocol (LTI), the Experience Application Programming Interface (xAPI), the Learning Analytics Standard (Caliper), the Shared Content Object Reference Model (SCORM), or a combination thereof.

[0027] In one embodiment, the learner is identified as a decentralized identity (DID).

[0028] In one embodiment, S1 also includes a local caching mechanism for offline learning events and a batch synchronization mechanism after network recovery, ensuring that the learning process in offline learning scenarios can also be preserved.

[0029] Furthermore, the learning event recording generation step in S1 also includes offline learning events, and the processing steps are as follows: When in offline mode, the generated offline learning events are stored in a local encrypted database; Using the device key, each of the offline learning events in the encrypted database is digitally signed; Once the network is restored, the digitally signed offline learning events will be synchronized in batches via a secure communication link.

[0030] In one embodiment, authentication in S2 is continuous authentication, including at least one of the following levels: Initial authentication: An initial authentication is performed at the start of the learning session; Periodic identity re-authentication: During the learning process, periodic identity re-authentication is performed at least once at unpredictable time intervals; Continuous behavioral feature verification: Continuously collect learners' behavioral biometric data and compare it with the learner's historical behavioral feature baseline model to assess the consistency of identity in real time.

[0031] Furthermore, the identity verification is continuous identity verification, including at least one of the following levels: Initial identity verification: At the beginning of the learning session, the learner's identity is confirmed through facial recognition, document verification, or multi-factor authentication; Periodic identity re-authentication: During the learning process, identity is periodically confirmed through random facial verification or challenge-response mechanisms; Continuous behavioral feature verification: During the learning process, the continued existence of the learner's identity is confirmed by continuously analyzing the learner's behavioral biometric data, which includes at least one of keystroke dynamics features, mouse movement trajectory features, or touch screen interaction features.

[0032] In one embodiment, the authenticity verification of behavior in S2 is achieved by analyzing the temporal, statistical, or frequency characteristics of the learning behavior sequence and using an anomaly detection algorithm to identify abnormal learning behaviors that do not conform to the preset behavior pattern.

[0033] In one embodiment, the content originality verification in S2 uses an artificial intelligence model to perform originality analysis and quality assessment on the text, code, or images submitted by learners.

[0034] In one embodiment, the environment trustworthiness verification in S2 verifies the learning environment in which the learner is located through at least one of a Trusted Execution Environment (TEE), a device fingerprint, or geographic location information.

[0035] In one embodiment, the cryptographic accumulator in S3 is one of Merkle tree, Patricia tree, Wickel tree, RSA accumulator, or bilinear accumulator.

[0036] In one embodiment, the immutable distributed ledger in S3 is one of a public blockchain, a consortium blockchain, a private blockchain, or a directed acyclic graph (DAG) structure.

[0037] In one embodiment, the formative evaluation evidence generation step is as follows: Obtain response data related to formative assessment events; Based on the answer data, evaluation metadata for the formative assessment event is collected. The evaluation metadata includes evaluation type, related knowledge points, difficulty level, and scoring criteria. Based on the answer data, the learner's answer process data in the formative assessment event is recorded, and the answer process data includes the response content, response time and answer order for each question; Based on the evaluation metadata and the answer process data, the evaluation result data is calculated and generated, which includes the total score, the score of each knowledge point, and the judgment result of the mastery level of each knowledge point. Record learners’ behavioral evidence data during the question-answering process, including the distribution of answering time, the number of answer modifications, and the number of pauses; Based on the answer process data, evaluation result data, and behavioral evidence data, a formative assessment record is generated and linked to the learner's dynamic learning history.

[0038] Furthermore, learning events include formative assessment events, and the documentation of formative assessment events (i.e., formative assessment documentation) includes: Collect evaluation metadata, including evaluation type, related knowledge points, difficulty level, and scoring criteria; Record learners' answer process data, including the content of their response to each question, the response time, and the order in which they answered the questions; Calculate and record the evaluation results, including the total score, the score for each knowledge point, and the assessment of mastery level; Record evidence of answering behavior, including the distribution of answering time, the number of times answers were modified, and the number of times paused; Generate formative assessment records and link them to learners' learning resumes.

[0039] Furthermore, the verification of the authenticity of formative evaluation events includes: Periodic authentication is performed during the evaluation process; Analyze answering behavior patterns and detect abnormal behavior; Verify the reasonableness of the evaluation timestamp; Check the consistency between the answers and the learning records.

[0040] In one embodiment, the steps for generating the reflection log evidence are as follows: Obtain the answer data related to the reflection log event; Based on the answer data, a reflection trigger context is recorded, which includes the trigger type and the associated learning event; Based on the collected answer data and the structured reflection content corresponding to the reflection trigger context, the structured reflection content includes content data in the dimensions of description, emotion, cognition, evaluation, and planning. The structured reflection content is evaluated based on reflection quality indicators; Based on the structured reflection content and the reflection quality indicators, a reflection log is generated and stored, and the reflection log is associated with the learner's dynamic learning history.

[0041] Furthermore, learning events include reflection log events, and the documentation of reflection log events (i.e., reflection log documentation) includes: Record the context in which reflections are triggered, including the trigger type and associated learning events; Collect structured reflection content, including descriptive, emotional, cognitive, evaluative, and planning dimensions; Assess the quality of reflections by indicators including the depth of reflection, the scope of dimensions covered, and the specificity of the content. Generate a reflection log, store the record, and link it to the learner's learning history.

[0042] Furthermore, the verification of the authenticity of the reflection log includes: Analyze writing behavior fingerprints, including typing rhythm, editing patterns, and pauses in thought; Check the consistency between the reflection content and the learning records; Verify the temporal relationship between reflection time and learning events; Perform originality testing.

[0043] Furthermore, the assessment of the depth of reflection is based on a cognitive hierarchy model, which includes five levels: descriptive reflection, interpretive reflection, analytical reflection, critical reflection, and transformative reflection.

[0044] Furthermore, the reflection log evidence storage supports privacy protection mechanisms, including: Content encryption: The original content of the reflection log is encrypted end-to-end using the learner's public key, ensuring that only the learner can decrypt and view it; Summary notarization: Only the content summary (hash value) and quality assessment metrics of the reflex log are notarized on the chain; the original content is not stored. Selective disclosure: Learners can selectively disclose portions of their reflection logs or quality metrics to specific validators; Access control: Through attribute-based access control (ABAC) mechanism, fine-grained control is exercised over who can access which parts of the reflection log.

[0045] In one embodiment, the system supports the associated evidence storage of formative evaluation and reflective logs, including: The creation of a reflection log is automatically triggered upon completion of the formative assessment; Link the reflection log with the corresponding formative assessment record; A comprehensive learning progress report is generated based on the evaluation results and reflections.

[0046] In one embodiment, the distributed storage system in S4 is a content-addressable storage system, including one of the InterPlanetary File System (IPFS), Arweave, Filecoin, or Storj.

[0047] In one embodiment, the dynamic learning resume construction in S4 supports the aggregation of learning events from multiple learning institutions, and the association and integration of learning records across institutions are achieved through decentralized identity identifiers (DID).

[0048] In one embodiment, the dynamic learning resume in S4 supports multi-view display, including at least one of a timeline view, a skills tree view, or a project set view.

[0049] Furthermore, the timeline view displays the learner's complete learning trajectory along a timeline; the skill tree view focuses on skills, showcasing the learner's mastery of different skills and learning paths; and the project set view displays the real-world projects completed by the learner and the results achieved, organized by project.

[0050] In one embodiment, S5 employs an artificial intelligence model for assisted evaluation, including assessment of learning outcome quality, analysis of learning engagement, or prediction of ability level.

[0051] In one embodiment, the verifiable credentials generated in S6 include formative assessment achievement credentials and reflective learner credentials, wherein: Formative assessment achievement credentials demonstrate the level of mastery achieved by learners in a specific domain; Reflective learner credentials demonstrate that learners possess reflective learning abilities and a commitment to continuous improvement.

[0052] Furthermore, "reflective learner credentials" may include the following: Reflection frequency: The number of reflection log entries completed by learners within a certain period; Reflection depth: The average depth level of learners' reflection logs; Reflection Quality: Overall quality score of learners' reflection logs; Reflective growth: The growth trend of learners' reflective abilities.

[0053] Furthermore, the Verifiable Credentials standard in S6 is compatible with at least one of the W3C Verifiable Credentials (VC) standard, the OpenBadges standard, or the Integrated Learning Record (CLR) standard.

[0054] In one embodiment, after S6, a credential lifecycle management mechanism is also included, supporting credential updates, revocations, inheritance, and version control.

[0055] In one embodiment, a credential verification step is also included, which supports a zero-knowledge proof mechanism, allowing learners to prove to the verifier that they meet specific learning achievement or ability requirements without revealing their complete learning history.

[0056] In one embodiment, the document also includes a credential presentation step that supports a selective disclosure mechanism, allowing learners to selectively disclose certain attributes of the credential to different verifiers while maintaining the verifiability of the credential.

[0057] Furthermore, the credential presentation involves the following steps: based on the digital certificate, a verifiable representation is generated according to the learner's authorization strategy through a selective disclosure mechanism. The verifiable representation includes some attribute information from the digital certificate and maintains the verifiability of the attribute information.

[0058] Furthermore, the selective disclosure mechanism is implemented based on the Idemix anonymous credential scheme or CL signature scheme, and the selectively disclosed JWT (SD-JWT) or BBS+ signature scheme.

[0059] In one embodiment, using distributed ledger technology as the trust anchor, and combining W3C DID and VC standards, the execution steps of the decentralized learning trust system are as follows: A multi-dimensional verification process is used to verify the authenticity of learning events generated by learners, including continuous identity verification based at least on behavioral biometrics. Anchor the proof of the verified learning event to the cryptographic evidence storage step of the distributed ledger; The process of generating verifiable credentials (digital certificates) based on stored learning events allows for the issuance of verifiable credentials that can be selectively disclosed through a zero-knowledge proof mechanism.

[0060] In one embodiment, this method supports the preservation of evidence of collaborative learning processes, enabling the recording of the individual contributions and interactive behaviors of multiple learners in the same learning activity.

[0061] In one embodiment, collaborative learning evidence storage further includes: recording the specific contributions of each learner based on a version control system; calculating the contribution ratio of each learner using a contribution metric algorithm; and generating a multi-party signature certificate for the collaborative learning outcome.

[0062] In one embodiment, a trust assessment model is also included to comprehensively evaluate the issuing authority's reputation, the history of credential issuance, and community acceptance. Example 2

[0063] One embodiment of the present invention discloses a learning process verifiable evidence storage and authentication system, comprising: The learning application layer is used to interact with learners and capture learning behavior data. The learning application layer includes a general course platform adapter, which supports at least one of the following: Learning Tool Interoperability Protocol (LTI), Experience Application Programming Interface (xAPI), Learning Analytics Standard (Caliper), and Shared Content Object Reference Model (SCORM). An edge computing layer, deployed on learner client devices, is used to preprocess, de-identify, or cache the learning behavior data offline, and generate learning events. The event processing layer is used to perform multidimensional verification and batch processing on received learning events, and to generate aggregate proofs using a cryptographic accumulator, anchoring the aggregate proofs to an immutable distributed ledger; a dynamic learning history is generated based on the aggregate proofs, and the multidimensional verification includes identity verification, behavior authenticity verification, content originality verification, and environment trustworthiness verification; wherein, the aggregate proofs include formative evaluation notarization and reflective log notarization. The authentication and service layer is used to monitor the stored records on the distributed ledger, generate dynamic learning resumes based on the stored learning events, perform automated authentication and evaluation of the dynamic learning resumes according to the preset multi-dimensional authentication standards, and issue digital certificates that conform to the verifiable credential standard through smart contracts or trusted issuance services. The data and storage layer is used to store learning events and dynamic learning history; Security and privacy layer, used to provide zero-knowledge proof generation, selective disclosure control, or key management services.

[0064] In one embodiment, the learning application layer includes at least one of a web application, a mobile application, an embedded software development kit (SDK), or a browser plugin.

[0065] In one embodiment, the learning application layer includes a universal course platform adapter for converting data from different learning technology standards into a unified internal event format.

[0066] In one embodiment, the edge computing layer is deployed on the learner's client device.

[0067] In one embodiment, the event handling layer includes: A multi-dimensional verification engine, which can dynamically combine and configure different verification mechanisms and their weights according to different authentication requirements; A batch evidence processor is used to aggregate multiple verified learning events in batches according to a preset time window or event quantity threshold; generate the aggregated proof using the cryptographic accumulator; and anchor the aggregated proof to an immutable distributed ledger.

[0068] Furthermore, the multidimensional verification engine includes: The identity verification module is used to perform initial identity authentication, periodic identity re-authentication, and continuous behavioral feature verification; The behavior authenticity verification module is used to analyze the temporal, statistical, or frequency characteristics of learning behavior sequences to identify abnormal learning behaviors. The content originality verification module is used to perform originality analysis on the text, code, or image content submitted by learners using an artificial intelligence model; The environment trustworthiness verification module is used to verify the learning environment through a trusted execution environment, device fingerprint, or geolocation information.

[0069] In one embodiment, the security and privacy layer includes: A zero-knowledge proof engine is used to generate zero-knowledge proofs that prove learners meet specific conditions; it can verify learners' learning achievements or abilities without exposing their complete dynamic learning history. A selective disclosure controller is used to select partial attribute information from the digital certificate according to the learner's authorization policy; generate a verifiable presentation containing partial credential information, and maintain the verifiability of the partial attribute information; The key management service is used to generate and store the learner's public and private key pairs; manage the key lifecycle, including key rotation and key recovery; and provide secure signing and encryption services.

[0070] In one embodiment, the data and storage layer includes an event database, a history database, a distributed storage cluster, and a blockchain network.

[0071] Furthermore, the event database adopts a time-series database architecture, supporting time range queries and event sequence analysis.

[0072] Furthermore, the resume database adopts a graph database architecture, which supports querying the relationships between the learning events and analyzing the learning paths.

[0073] Furthermore, the blockchain network supports smart contracts for automating the execution of credential issuance, verification, and revocation logic.

[0074] In one embodiment, the system further includes a learning path recommendation engine that recommends personalized learning activities for the learner based on the learner's historical learning history, formative assessment records, and reflection log records.

[0075] In one embodiment, the system further includes a learner competency profile module, used to construct the learner's knowledge graph based on the formative assessment evidence; track the learner's growth trajectory in each competency dimension; and generate a visualized competency radar chart.

[0076] In one embodiment, the system supports a time travel query function, allowing users to query a learner's learning history status and ability level at any historical point in time.

[0077] In one embodiment, the system adopts a plug-in architecture, which supports the extension of verification algorithms, evidence storage strategies and digital certificate formats through plug-ins.

[0078] In one embodiment, the system provides a data export interface that allows learners to fully export all their personal learning data in a machine-readable format that complies with the GDPR's "right to data portability" requirement.

[0079] In one embodiment, the authentication and service layer includes: Open Validation Service is an application programming interface (API) used to provide third parties with verification of the authenticity and validity of digital certificates. Identity Federation Service is used to support the interoperability of learners' decentralized identities (DIDs) with other identity authentication systems (such as OpenID Connect and SAML).

[0080] In one embodiment, to ensure uninterrupted learning event storage services even in the event of a complete failure of a single data center or even a single geographical region, the system adopts a cross-regional disaster recovery approach, including: multi-data center deployment; real-time data synchronization; automatic failover; and disaster recovery drills.

[0081] Furthermore, specifically: Deploy independent system instances in multiple geographically dispersed data centers; Using a global traffic manager, learning event write requests are routed to the optimal data center based on the user's geographic location and data center health. Learning event data is asynchronously replicated between data centers to ensure eventual consistency. When any data center fails, traffic is automatically switched to other healthy data centers.

[0082] Furthermore, the replication of learning event data is achieved through a distributed message queue, where a message queue instance within each data center mirrors and replicates data from other data centers.

[0083] Furthermore, the Global Traffic Manager uses the Anycast routing protocol to direct user requests to the nearest data center to reduce network latency.

[0084] In one embodiment, the system further includes a method for enhancing the resilience of the learning process evidence storage system, specifically: Define the system's key business metrics as a steady-state assumption; In a production environment, events simulating real-world failures, such as network latency, server outages, or API errors, can be injected in a controlled manner through a chaos engineering platform. Continuously monitor key business metrics and verify whether the system's behavior under fault injection meets expectations; If the system does not meet expectations, an automatic alarm will be triggered and a diagnostic report will be generated, initiating the system repair process.

[0085] Furthermore, the chaos engineering platform can be integrated with CI / CD pipelines to automatically perform a series of resilience tests before software release, and changes that do not meet resilience requirements will be automatically prevented from deployment.

[0086] In one embodiment, the system includes an observability monitoring module for: collecting system operation metrics and logs; providing real-time alarms and anomaly detection; and supporting distributed tracing.

[0087] Specifically, this involves integrating a unified telemetry SDK into all services of the system to collect distributed tracking, metrics, and log data. The collected telemetry data is sent to a unified backend for storage and correlation analysis; The user request chain across multiple services can be visualized end-to-end using a unique tracking ID; Machine learning models are trained based on historical telemetry data for real-time anomaly detection and root cause analysis.

[0088] Furthermore, machine learning models can predict future resource usage trends in the system and automatically trigger alerts or elastic scaling operations when potential capacity bottlenecks are predicted.

[0089] In one embodiment, the system provides third-party integration interfaces, including: RESTful API interfaces; software development kits (SDKs); webhook callback mechanisms; and OAuth 2.0 authorization services.

[0090] Furthermore, to lower the integration threshold of third-party course platforms and provide third-party applications with the ability to store learning events, the steps are as follows: The API interface for learning event evidence storage is defined based on the OpenAPI specification. Based on the API specification, software development kits (SDKs) for multiple programming languages ​​are automatically generated. The SDK encapsulates the complexity of API calls, including authentication, signature calculation, and error retry logic. Third-party applications can submit learning event data to the learning process evidence storage system by integrating the SDK.

[0091] Furthermore, a developer sandbox environment isolated from the production environment is also provided, which can: Provide each third-party developer with independent API credentials and data storage space; Simulate various API response scenarios, including success, failure, and asynchronous callbacks; It provides visual API call logs and data dashboards to help developers debug.

[0092] Furthermore, the developer sandbox provides a "time travel" feature, allowing developers to simulate the state of the system at a specific point in the future to test time-related business logic.

[0093] In one embodiment, the authentication and service layer includes a multi-credential standard conversion engine for format conversion between different credential standards, including W3C verifiable credentials, open badges, and integrated learning records.

[0094] Furthermore, to ensure that the issued learning credentials are widely accepted and verified globally, the system supports multiple digital credential standards. The implementation steps are as follows: Define an internally unified credential data model that includes metadata about the complete learning process; It provides a credential conversion engine that can convert the internal unified credential model into a variety of mainstream digital credential standard formats in the industry, including W3C Verifiable Credentials (VC), Open Badges, and Comprehensive Learning Records (CLR). Allows users to export their learning outcomes as digital credentials in different standard formats as needed.

[0095] In one embodiment, to achieve scalability of the general course platform adapter and credential conversion engine, the system further includes a method for managing open-source community contributions, comprising: Establish a hierarchical governance model consisting of core contributors, community contributors, and a technical guidance committee, responsible for reviewing and managing community contribution code for the adapter plugin and credential conversion module; A clearly defined guide for adapter plugin development and credential standard adaptation; Use automated continuous integration and continuous deployment (CI / CD) pipelines to perform quality checks, compatibility tests, and security scans on community-contributed adapter plugins and credential conversion modules; Community-contributed code that passes testing will be merged into the product's main branch to enable support for new learning platforms or new credential standards.

[0096] Furthermore, provide economic incentives for core contributors and members of the technical steering committee, including but not limited to: payment based on the amount of code contributed, bug bounties, and new feature development contracts.

[0097] Furthermore, it allows community developers to contribute support for emerging or domain-specific credential standards. Example 3

[0098] One embodiment of the present invention discloses a method for constructing a decentralized trust system, the specific steps of which are as follows: A decentralized identity (DID) is created for the learner, which is controlled by the learner and does not depend on any centralized institution. Associate the learner's learning events, dynamic learning history, and digital certificate with the decentralized identity identifier (DID); The credential metadata of the digital certificate is recorded in a distributed ledger, enabling the digital certificate to be verified independently.

[0099] Furthermore, the decentralized learning trust system constructed in this invention includes the following construction method: Trust anchor establishment: The distributed ledger serves as an immutable root of trust, and the validity of all identities, credentials, and evidence records can ultimately be traced back to this root of trust.

[0100] Trust transfer mechanism: Trust transfer is achieved through the W3C VC standard. The issuing authority (such as a university) registers its DID with the trust registry. Learners trust the issuing authority and accept the credentials it issues. The verifier trusts the issuing authority and therefore also trusts the credentials it issues.

[0101] Trust assessment model: The system can establish a trust scoring model that comprehensively evaluates factors such as the issuing authority's reputation, the history of certificate issuance, and community recognition, providing a decision-making reference for the verification party.

[0102] Trust update and revocation: Through on-chain or off-chain revocation list mechanisms, trust relationships can be dynamically updated and revoked, ensuring the real-time nature and effectiveness of the trust system. Example 4

[0103] One embodiment of the present invention discloses a continuous authentication method, the specific steps of which are as follows: Initial identity authentication: At the start of the learning session, the learner's identity is confirmed through facial recognition, document verification, or multi-factor authentication; Periodic identity re-authentication: During the learning process, identity verification is performed periodically at unpredictable time intervals through random facial verification or challenge-response mechanisms. Continuous behavioral feature verification: Throughout the learning process, learners’ behavioral biometric data are continuously collected and compared with the learners’ historical behavioral feature baseline model to assess the consistency of identity in real time. The behavioral biometric data includes at least one of keystroke dynamics features, mouse movement trajectory features, or touch screen interaction features.

[0104] Furthermore, keystroke dynamics characteristics include at least one of keystroke duration, keystroke interval, keystroke pressure, or input error rate. Example 5

[0105] One aspect of this invention provides a formative evaluation evidence preservation method, the specific steps of which are as follows: Obtain response data related to formative assessment events; Based on the answer data, evaluation metadata is collected, which includes evaluation type, related knowledge points, difficulty level and scoring criteria. Record learners' answer process data, which includes the response content, response time, and answer order for each question; Calculate and generate evaluation result data, which includes the total score, the score for each knowledge point, and the mastery level assessment result; Record evidence data of answering behavior, including the distribution of answering time, the number of times answers were modified, and the number of times paused; A question-answering response record is constructed based on the question-answering process data, the evaluation result data, and the question-answering behavior evidence data. An incremental hash chain structure is adopted, in which each answer response record is linked with the previous answer response record to form a hash chain, and the final hash value of the incremental hash chain is anchored to an immutable distributed ledger.

[0106] In one embodiment, the authenticity verification of the formative evaluation event includes: performing periodic identity verification during the evaluation process; analyzing answer behavior patterns and detecting abnormal behavior; verifying the rationality of the evaluation timestamp; and checking the consistency between the answer content and the learning record.

[0107] In one embodiment, the formative assessment evidence generated based on the formative assessment event supports multiple assessment types, including diagnostic tests, exercises, in-class quizzes, learning checkpoints, self-assessment, and peer assessment, with each assessment type employing a corresponding data collection and evidence storage strategy.

[0108] In one embodiment, the formative assessment record includes dynamic tracking of learners’ ability level changes at each knowledge point, forming a learner’s ability growth trajectory.

[0109] In one embodiment, the formative evaluation evidence storage adopts an incremental hash chain structure, such as... Figure 9 As shown, each answer response record forms a hash chain with the previous record, ensuring the integrity and immutability of the evaluation process.

[0110] In one embodiment, it supports automatically triggering the creation of a reflection log after the formative assessment is completed, and generating a comprehensive learning progress report based on the assessment result data. Example 6

[0111] One aspect of this invention provides a method for storing reflection logs as evidence. The specific steps for storing reflection logs as evidence are as follows: Obtain the answer data related to the reflection log event; Based on the answer data, a reflection trigger context is recorded, which includes the trigger type and associated learning events; The structured reflection content is collected based on the answer data, and the structured reflection content includes descriptive dimension, emotional dimension, cognitive dimension, evaluation dimension and planning dimension; The structured reflection content is evaluated based on reflection quality indicators; this includes an automated evaluation of the depth level of the structured reflection content based on a preset cognitive hierarchy model. The reflection content is encrypted and stored, and the reflection quality assessment results and content summaries are also preserved as evidence.

[0112] In one embodiment, the authenticity verification of the reflection log includes: analyzing writing behavior fingerprints, including typing rhythm, editing patterns, and pauses in thinking; checking the consistency between structured reflection content and learning records; verifying the temporal relationship between reflection time and learning events; and performing originality detection.

[0113] In one embodiment, the assessment of the depth level is based on a cognitive hierarchy model, which divides the depth of reflection into five levels: descriptive reflection, interpretive reflection, analytical reflection, critical reflection, and transformative reflection.

[0114] In one embodiment, the reflection log notarization supports privacy protection, including: encrypting and storing the reflection content; notarizing only the content summary and reflection quality indicators to a distributed ledger; and allowing learners to control the scope of disclosure of the reflection content.

[0115] In one embodiment, the quality assessment metrics of the reflection log also include text sentiment analysis of the reflection content, used to quantify the learner's emotional state during the reflection process.

[0116] In one embodiment, the system supports associating reflection logs with corresponding formative assessment records and generating a comprehensive learning progress report based on the reflection content. Example 7

[0117] In one embodiment, when the computer program is executed by a processor, it implements the above-described learning process to verify the evidence storage and authentication method. Example 8

[0118] This embodiment discloses the core method and process for evidence preservation during the learning process, and the specific steps are as follows: S201. Learning Event Collection Steps: Learning event collection is the starting point of the entire evidence preservation process. It uses a universal adapter architecture to capture learners' learning behavior data on the learning platform in real time and convert it into a unified internal event format, specifically: S2011. Unified Learning Event Data Structure: This embodiment defines a unified learning event data structure (Unified Learning EventSchema), compatible with mainstream standards such as xAPI and Caliper, while extending validation-related metadata fields. This data structure uses JSON format, and its specific definition is as follows: { "$schema": "https: / / json-schema.org / draft / 2020-12 / schema", "title": "UnifiedLearningEvent", "type": "object", "required": ["eventId", "timestamp", "actor", "verb", "object", "context", "verification"], "properties": { "eventId": { "type": "string", "format": "uuid", "description": "A globally unique identifier for the event, in UUID v4 format" }, "timestamp": { "type": "string", "format": "date-time", "description": "The precise timestamp of the event, in ISO 8601 format, accurate to milliseconds." }, "actor": { "type": "object", "description": "Learner Information", "required": ["did"], "properties": { "did": { "type": "string", "description": "Learner's decentralized identity (DID)" }, "name": { "type": "string", "description": "Learner's Name" } } }, "verb": { "type": "object", "description": "Learning behavior", "required": ["id"], "properties": { "id": { "type": "string", "format": "uri", "description": "A unique identifier URI for the behavior, compatible with xAPI Verb ID" }, "display": { "type": "object", "description": "Multilingual display name for the behavior" } } }, "object": { "type": "object", "description": "Learning target", "required": ["id", "objectType"], "properties": { "id": { "type": "string", "format": "uri", "description": "The unique identifier URI of the learning object" }, "objectType": { "type": "string", "enum": ["Activity", "Agent", "Group", "StatementRef", "SubStatement"], "description": "Type of learning object" }, "definition": { "type": "object", "description": "Detailed definition of the learning object" } } }, "context": { "type": "object", "description": "Contextual information about the event" }, "result": { "type": "object", "description": "The result of the learning behavior" }, "verification": { "type": "object", "description": "Multidimensional validation results", "properties": { "identity": { "status": "enum[verified|unverified|failed]", "method": "string", "confidence": "number }, "behavior": { "status": "enum[normal|suspicious|fraudulent]", "score": "number" }, "originality": { "status": "enum[original|plagiarized|aiGenerated]", "score": "number" }, "environment": { "status": "enum[trusted|untrusted]", "details": "object } } } } } S2012. Offline Evidence Preservation and Synchronization: To address learning scenarios in environments with unstable mobile devices or networks, this embodiment employs an offline evidence storage and synchronization mechanism. The acquisition module maintains an event queue locally (such as in the browser's IndexedDB or the mobile application's local database). When the network connection is lost, newly generated learning events are added to this queue and signed using a key securely stored on the device. Once the network connection is restored, the acquisition module automatically sends the events in the queue in batches to the event processing layer for verification.

[0119] S202. Multidimensional Validation Steps: Multidimensional verification is a core component in ensuring the authenticity of the learning process. The event processing layer includes a pluggable, configurable multidimensional verification engine that performs real-time or near-real-time verification on each acquired learning event.

[0120] S2021. Continuous authentication; S2022. Verification of the authenticity of behavior: The behavioral authenticity verification engine detects anomalous behavior by analyzing the learning event stream. For example, by comparing the learner's average time to complete a series of tasks with historical baseline data, behaviors that are too fast (potentially cheating) or too slow (potentially distracted) can be identified. By analyzing video viewing event sequences, atypical learning behaviors such as dragging the progress bar, fast-forwarding, and background playback can be detected.

[0121] S2023. Content Originality Verification: When a learning event includes content submitted by a learner (such as code, articles, or design works), the system invokes the content originality verification service. This service uses an AI large language model and code / text comparison algorithms to compare the content with publicly available internet resources, internal databases, and the learner's historical submissions, providing an originality score and a report on potential sources of plagiarism.

[0122] S2024. Environmental Trustworthiness Verification: Environmental trustworthiness verification aims to ensure that learners study in a controlled and trustworthy environment. For example, when taking high-stakes certification exams, the system can require learners to use devices integrated with a Trusted Execution Environment (TEE) to ensure that the operating system and applications have not been tampered with. For ordinary learning scenarios, abnormal environments (such as virtual machines or automated scripts) can be identified by collecting device fingerprints (such as browser version, installed fonts, screen resolution, etc.).

[0123] S203. Batch Evidence Storage Steps: In order to minimize the cost of on-chain storage while ensuring security, this embodiment adopts a batch storage mechanism.

[0124] S2031. Event Batch Processing: The event processing layer packages learning events that have been validated through multidimensional analysis within a certain period of time (e.g., 10 minutes) or a certain number (e.g., 1000) into a batch.

[0125] S2032. Cryptographic Accumulator Aggregation: In this embodiment, a Merkle tree is preferred as the cryptographic accumulator. The hash value of each learning event in the batch is used as a leaf node of the Merkle tree, and the hash is calculated layer by layer upwards to generate a unique Merkle root. This Merkle root is the aggregate proof of all learning events in that batch.

[0126] S2033. Ledger Anchoring: The event processing layer packages metadata such as Merkle root, batch ID, and timestamp into a transaction and submits it to a distributed ledger (such as Ethereum or Hyperledger Fabric). Once the transaction is packaged into a block, the notarization and anchoring of the learning event in that batch is completed. Anyone can verify whether a specific learning event is included in the batch and has not been tampered with using the Merkle root and the corresponding Merkle Proof.

[0127] Furthermore, distributed ledgers support smart contracts, which are used to implement automated evidence storage logic. Specifically, smart contracts include the following functionalities: Evidence storage trigger conditions: When the batch submitted by the event processing layer meets the preset conditions (such as the batch size reaching the threshold or the time interval reaching the threshold), the smart contract automatically executes the evidence storage operation. The evidence verification logic is as follows: The smart contract verifies that the submitted Merkle root format is correct, the timestamp is valid, and the signature is legal. Evidence Recording: The smart contract permanently records the verified evidence information (Merkelgen, batch ID, timestamp, submitter address) on the blockchain; Event notification: The smart contract triggers an event notification after the notarization is completed, which is then listened to and processed by the authentication and service layer.

[0128] S204. Resume Construction Steps: S2041. Distributed Storage: Complete batch data of learning events (in JSON format) is stored in a distributed storage system such as IPFS. IPFS returns a Content Identifier (CID), which is also recorded in the anchor transaction. This "on-chain notarization (Merkelgen), off-chain data storage (IPFS)" model balances cost, efficiency, and data integrity.

[0129] S2042. Generating Learning Resume: The authentication and service layer continuously monitors new blocks on the distributed ledger and parses out the anchored transactions. Based on the CID in the transaction, it retrieves complete event data from IPFS and stores it in the resume database. The resume database aggregates learning events from different courses and platforms based on the learner's DID, constructing a complete and dynamically updated learning resume. This resume can be visualized as needed, using various views such as timelines, skill trees, and project portfolios.

[0130] S205. Certification Assessment Steps: The certification and service layer includes a certification assessment engine that can automatically evaluate learning resumes based on preset certification criteria. For example, a "Python Junior Programmer" certification might require: Complete at least 40 hours of Python-related coursework; Master 15 core knowledge points including variables, loops, and functions (proven through course quizzes and programming exercises); Complete at least 3 small projects independently.

[0131] The evaluation engine scans the learner's resume and automatically determines whether they meet all the criteria.

[0132] S206. Document Issuance Procedures: S2061. Verifiable credential generation: Once a learner meets a certain authentication standard, the authentication and service layer will invoke the credential issuance service to generate a digital credential that conforms to the W3C VC standard. This credential is a JSON-LD format document that contains the issuer, holder (learner DID), credential type, granting standard, issuance date, and a cryptographic signature.

[0133] S2062. Voucher Lifecycle Management: The system supports credential revocation. For example, if serious academic misconduct is discovered in the process of obtaining a credential, the issuing authority can add the learner's DID and credential ID to an on-chain or off-chain revocation list. Validators need to query this revocation list when verifying a credential to ensure that the credential remains valid. Example 9

[0134] This embodiment describes the six-layer architecture of the system in detail, such as Figure 1 As shown: Learning Application Layer (L1): This is the entry point for interaction between the system and learners. It can be a web application, mobile app, browser plugin, or SDK embedded in a third-party learning platform. It is responsible for presenting learning content and capturing raw user behavior data.

[0135] Edge computing layer (L2): Deployed on the learner's client device, it is responsible for preprocessing raw behavioral data, such as filtering out irrelevant mouse movement events, blurring private areas (such as faces and chat windows) in screen recordings, and caching data when the network is interrupted.

[0136] Event Processing Layer (L3): This is the core business logic layer of the system. It receives event data from the edge computing layer, converts the format through the general course platform adapter, sends it to the multi-dimensional verification engine for verification, and finally performs batch aggregation and evidence storage through the cryptographic accumulator.

[0137] Security and Privacy Layer (L4): Provides cryptographic services related to security and privacy. This includes managing learners' DIDs and associated keys, generating and verifying zero-knowledge proofs, and generating verifiable presentations based on user authorization policies.

[0138] Furthermore, the security and privacy layer includes a key management service, which is responsible for: Key generation: Generates DID and its associated public / private key pair for learners, supporting cryptographic algorithms such as Ed25519 and secp256k1; Key storage: Securely store the private key in the learner's device security area (such as iOS Keychain, Android Keystore) or in the hardware security module (HSM); Key rotation: Supports rotating keys periodically or on demand while keeping the DID unchanged, and publishing new public keys through the DID document update mechanism; Key revocation: When a key is leaked or a learner requests it, the key can be revoked and the DID document updated. Key recovery: Supports key recovery mechanisms based on social recovery or mnemonic phrases to prevent identity loss due to key loss.

[0139] Furthermore, the security and privacy layer also includes a reflection log privacy protection module, which is responsible for: Content encryption: The original content of the reflection log is encrypted using AES-256-GCM using the learner's public key; Key Management: Works with the key management service to manage the keys used to encrypt the reflection log; Access control: Implement attribute-based access control (ABAC) to determine access permissions based on the visitor's identity and attributes; Audit Log: Records all access requests to the reflex log, supporting post-event auditing.

[0140] Data and Storage Layer (L5): This is the system's persistence layer. It includes an event database (such as TiDB or CockroachDB) for storing real-time events and hot data, a resume database (such as Neo4j) for storing learning resumes, a distributed storage cluster (such as IPFS) for storing batch data, and a blockchain network (such as Ethereum) as the final anchor of trust.

[0141] Furthermore, the blockchain network supports the deployment and execution of smart contracts. The smart contracts deployed by the system include: The evidence storage contract is responsible for receiving and verifying aggregated proofs submitted by the event processing layer and permanently recording them on the blockchain. Contract Cancellation: Responsible for managing the list of revoked certificates, and supports issuing institutions in canceling issued certificates; Identity Registration Contract: Responsible for managing learners' DID registration and resolution.

[0142] The smart contracts are written in Solidity (Ethereum) or Chaincode (Hyperledger Fabric) and deployed after undergoing security audits.

[0143] Furthermore, the event database adopts a distributed NewSQL database architecture (such as TiDB and CockroachDB), and has the following characteristics: Horizontal scaling: Supports linear expansion of storage capacity and processing power by adding nodes to meet the storage needs of massive learning events; Strong consistency: Employs distributed consensus protocols such as Raft or Paxos to ensure strong consistency of data across nodes; High availability: Supports multi-replica deployment; single-node failure does not affect service availability. SQL Compatibility: Provides a standard SQL interface, reducing development and operational complexity; Partitioning strategy: Partition data by learner DID or time range to optimize query performance.

[0144] Furthermore, the resume database employs a graph database architecture (such as Neo4j) to store and query learners' learning resumes and their relationships. Specifically: Node types include learner nodes, course nodes, knowledge point nodes, skill nodes, voucher nodes, and institution nodes. Relationship types include "Completion" (learner-course), "Mastery" (learner-knowledge point), "Acquisition" (learner-credential), and "Prerequisite" (knowledge point-knowledge point), etc. Attribute storage: Each node and relationship can store rich attributes, such as completion time, mastery level, verification status, etc. Graph queries: Supports complex graph traversal queries, such as "finding all the knowledge points mastered by the learner and their prerequisites" or "finding learners with similar learning paths". Visualization: Supports visualizing learning resumes in various views such as timeline, skill tree, and knowledge graph.

[0145] Authentication and Services Layer (L6): Provides high-level services to learners, issuing authorities, and verifiers. This includes querying and displaying learning records, automated authentication assessment, credential issuance and management, third-party verification APIs, and identity federation services.

[0146] Furthermore, the authentication and service layer also includes a reflective learner credential issuance module, which is responsible for: Assessment Trigger: When the quantity or quality of a learner's reflection logs reaches a preset threshold, the voucher issuance assessment is automatically triggered. Comprehensive assessment: A comprehensive analysis of learners' reflection frequency, depth, quality, and growth trends; Credential Generation: Generate reflective learner credentials that conform to W3C VC standards based on the assessment results; Credential issuance: The credential is signed using the issuing authority's private key and sent to the learner's digital wallet. Example 10

[0147] This embodiment provides an example of a W3C VC data structure for a learning achievement credential, such as... Figure 2 As shown: { "@context": [ "https: / / www.w3.org / 2018 / credentials / v1", "https: / / www.w3.org / 2018 / credentials / examples / v1" ], "id": "http: / / example.edu / credentials / 3732", "type": ["VerifiableCredential", "LearningAchievementCredential"], "issuer": "did:example:123456789", "issuanceDate": "2026-01-15T10:00:00Z", "credentialSubject": { "id": "did:example:ebfeb1f712ebc6f1c276e12ec21", "achievement": { "type": "CourseCompletion", "name": "Introduction to Artificial Intelligence", "description": "Completed the introductory course on AI,covering topics from search algorithms to neural networks.", "criteria": "http: / / example.edu / courses / ai101 / criteria" } }, "proof": { "type": "Ed25519Signature2018", "created": "2026-01-15T10:00:00Z", "verificationMethod": "did:example:123456789#keys-1", "proofPurpose": "assertionMethod", "jws": "eyJhbGciOiJFZERTQSIsImI2NCI6ZmFsc2UsImNyaXQiOlsiYjY0Il19..l9dbe2d_g43ctb4s-bap-l-f9l9j_f9j" } } This credential is issued by the "issuer" to the "credentialSubject" (learner) to certify their completion of a learning achievement entitled "Introduction to Artificial Intelligence". The integrity of the credential is guaranteed by the digital signature in the "proof" field. Example 11

[0148] This embodiment discloses a universal course platform adapter, which is key to achieving cross-platform interoperability, such as... Figure 3 As shown, its core idea is "external standardization and internal unification," specifically: Input Module: Develop a dedicated input plugin for each mainstream learning technology standard (such as xAPI, LTI, Caliper, SCORM). This plugin is responsible for listening to or polling data from the specific learning platform and parsing it into a generic object model within the plugin.

[0149] The transformation engine converts the generic object model of the input plugin into the unified learning event data structure defined in this embodiment. This process includes field mapping, data formatting, and necessary information completion.

[0150] Output module: Sends the unified learning event data structure to subsequent modules in the event processing layer (such as the multidimensional verification engine).

[0151] With this pluggable adapter architecture, the system can quickly support new learning platforms or standards by simply developing a new input plugin for them, without changing the core business logic. Example 12

[0152] This embodiment discloses a continuous authentication method, which is a detailed elaboration of step S2021 in embodiment three. Continuous authentication is the key to ensuring that "the operator is the same person," such as... Figure 4 As shown.

[0153] Initial Identity Authentication (S701): Before starting an important learning session (such as a certification exam), learners are required to undergo a strong identity verification process. For example, this involves scanning their face via a mobile app and comparing it with an ID photo in a public security system or authoritative database to confirm that the person and the ID match.

[0154] Periodic Identity Re-authentication (S702): During the learning process, the system will pop up challenge-response verification requests at unpredictable time intervals. For example, a number will randomly appear on the screen, requiring the learner to read the number within 5 seconds. The system will capture the audio through the microphone and perform voiceprint recognition, or capture lip movements through the camera for verification.

[0155] Continuous behavioral feature verification (S703): This is the core verification method with the best user experience. The edge computing layer (L2) acquisition module continuously and silently collects the learner's behavioral biometric data, for example: Keystroke dynamics: key press duration (dwell time), flight time between two keys, input error rate, etc.

[0156] Mouse movement trajectory: movement speed, acceleration, curvature, and distribution of pause positions, etc.

[0157] Touchscreen interaction: swiping speed, pressure, contact area, etc.

[0158] The system establishes a personalized behavioral baseline model for each learner. During the learning process, the system collects behavioral data in real time and compares it with the baseline model. If the deviation exceeds a preset threshold, the verification status will be marked as "suspicious," and the system may increase the frequency of periodic re-authentication or record the suspicious event in the learning history. Example 13

[0159] This embodiment details how to protect learner privacy using zero-knowledge proofs (ZKP) and selective disclosure (SD) techniques, such as... Figure 5 As shown.

[0160] Scenario: A learner (Alice) wants to prove to an employer (Bob) that she "possesses an advanced Python programming certificate" and that "the certificate was awarded by a well-known university," but does not want to disclose the date she obtained the certificate, her specific grades, or other learning records.

[0161] process: 1. Credential Issuance: The university (issuing institution) issues Alice's DID a W3CVC containing all her academic achievements. This VC uses a signature scheme that supports selective disclosure, such as BBS+ signature.

[0162] 2. Generating a Verifiable Presentation (VP): Alice's digital wallet (part of the security and privacy layer) derives a new, partially informed Verifiable Presentation from the original VC, according to her instructions. This VP contains only two statements: "Certificate Name = Python Advanced Certificate" and "Issuing Authority = A Well-Known University," along with a cryptographic proof (BBS + Derivation Proof) that demonstrates these statements originate from the original valid VC.

[0163] 3. Presentation and Verification: Alice sends this VP to Bob. Bob's verification system can independently and offline verify the cryptographic proof of the VP, confirming that the statement was indeed issued to Alice by the university and has not been tampered with. Throughout the process, Bob has no access to any other information in Alice's original VC.

[0164] Applications of zero-knowledge proofs: In another scenario, Alice wants to prove that "her average score on programming exercises is over 90 points," but doesn't want to reveal the specific score of any particular exercise.

[0165] 1. Data storage: Alice's scores for each programming exercise are recorded and stored.

[0166] 2. ZKP Generation: Alice's digital wallet reads all practice scores and generates a zk-SNARK proof locally. This proof demonstrates that "there exists a set of numbers whose hash values ​​correspond to the hash values ​​stored on-chain, and the average of this set of numbers is greater than 90," but the proof itself does not contain any specific score value.

[0167] 3. Verification: Alice sends the zk-SNARK proof to Bob. Bob can verify the proof without knowing any specific scores to confirm that Alice's average score is indeed over 90. Example 14

[0168] This embodiment describes in detail a method for evidence preservation in an offline learning scenario.

[0169] 1. Offline detection: The edge computing layer (L2) acquisition module continuously monitors the network connectivity status. Once a network interruption is detected, it immediately switches to offline mode.

[0170] 2. Local Caching: In offline mode, all newly generated learning events are stored locally on the device in an encrypted database (such as SQLCipher). Each event is signed using the learner's device key to prevent local tampering.

[0171] 3. Network recovery: The acquisition module periodically attempts to connect to the server of the event handling layer (L3).

[0172] 4. Batch Synchronization: Once the network connection is restored, the acquisition module will package and compress all locally cached events and upload them in batches to the event processing layer via a secure HTTPS connection.

[0173] 5. Conflict Resolution: Upon receiving an offline event packet, the event processing layer performs timestamp and signature verification. If a conflict is found with an online event (e.g., two different events are recorded at the same time), the system will handle it according to a preset strategy (e.g., prioritizing the online event and marking the offline event as "pending review"). Example 15

[0174] This embodiment discloses a multidimensional certification standard and AI assessment, describing how to define and evaluate learning achievements.

[0175] The Certification and Services Layer (L6) provides a certification standards editor, allowing educational institutions or industry experts to define complex certification standards. A standard can consist of multiple rules, for example: Rule 1 (Course Learning): The number of events where "verb.id" is "completed" and "object.definition.type" is "course" >= 5.

[0176] Rule 2 (Mastery of Knowledge Points): The number of knowledge points for which "result.extensions['mastery']" is "true" is >= 10, and the list of knowledge points is ['A', 'B', 'C', ...].

[0177] Rule 3 (Project Experience): The number of events where "object.definition.type" is "project" and "result.success" is "true" is >= 1.

[0178] Rule 4 (AI Evaluation): For items in Rule 3, "result.extensions['ai_quality_score']" >= 0.8.

[0179] The certification assessment engine scans learners' resumes periodically or as needed, matching them against these rules. Rule 4, in particular, is assessed by a dedicated AI model that analyzes the complexity, standardization, readability, and efficiency of the project code submitted by learners, providing a comprehensive quality score. Example 16

[0180] Combination Figure 6-8 As shown, the overall workflow of the system is explained in sequence, specifically as follows: like Figure 6 As shown, the learning and evidence preservation process is as follows: Learners study on the learning platform.

[0181] The edge computing layer's acquisition module captures learning behavior and generates learning events (S201).

[0182] The event is sent to the event handling layer for multidimensional verification (S202).

[0183] The verified events are aggregated in batches to generate Merkle root (S203).

[0184] Merkle is anchored to the distributed ledger, while the raw event data is stored in distributed storage (S204).

[0185] like Figure 7 As shown, the resume creation and voucher issuance process is as follows: The authentication and service layer monitors new records on the distributed ledger.

[0186] Retrieve event data from distributed storage and update the learner's learning history (S301).

[0187] The certification assessment engine evaluates the resume based on preset standards (S302).

[0188] If the criteria are met, the credential issuance service is invoked to issue a verifiable credential (VC) to the learner's DID (S303).

[0189] like Figure 8 As shown, the third-party verification process is as follows: Learners want to prove a learning achievement to a validator (such as an employer).

[0190] Learners use their digital wallets to generate a verifiable representation (VP) containing only partial information from the original VC (S401).

[0191] The learner presents the VP to the validator (S402).

[0192] The verifier's system can independently verify the cryptographic signature of the VP and query the revocation list on the distributed ledger to confirm the validity of the credential (S403). The entire process requires no interaction with the issuing authority. Example 17

[0193] This embodiment discloses a multi-active data center and global routing, aiming to solve the technical problem of providing uninterrupted learning event storage services even in the event of a complete failure of a single data center or even a single geographical region. Specifically: Cell-Based Architecture: This architecture divides the system into multiple independent, fully functional "units," each containing complete application services and data storage. One or more units are deployed in each geographic region.

[0194] 2. Global Traffic Manager (GSLB): Uses a DNS-based global load balancer to intelligently route user requests to the optimal unit based on the user's geographic location, network latency, and the health status of each unit.

[0195] 3. Distributed Database: Employs a distributed database that supports strong consistency across regions (such as CockroachDB, Google Spanner, TiDB) to store core metadata such as user identity and certificate status.

[0196] 4. Distributed message queue: Use a message queue that supports cross-region replication (such as Apache Kafka with MirrorMaker 2) to replicate learning event data and achieve eventual consistency.

[0197] This implementation can achieve system availability of over 99.99%, recovery time objective (RTO) of less than 1 minute, and recovery point objective (RPO) close to zero. Example 18

[0198] This embodiment discloses chaos engineering and system resilience, aiming to solve the technical problem of how to proactively discover potential failure points before system deployment and during operation, and improve system resilience. Specifically: 1. Define steady-state assumptions: Determine the key business indicators for normal system operation (e.g., event reception success rate > 99.99%, certificate verification latency < 200ms).

[0199] 2. Design Chaos Experiments: Design a series of fault injection experiments, including: random container termination, network latency injection, simulating database master node failure, and simulating API interface timeout.

[0200] 3. Conduct experiments: Conduct experiments in a controlled production environment, limiting the blast radius to ensure that it does not affect all users.

[0201] 4. Verification and Remediation: Monitor business metrics to verify that the system is working as expected. If the assumption is incorrect, fix the identified issues and automate the experiment by incorporating it into the CI / CD process.

[0202] This implementation method can transform system failures from "passive response" to "active prevention," significantly reducing the risk of unexpected downtime in the production environment. Example 19

[0203] This embodiment discloses unified observability and AIOps, aiming to solve the technical problem of how to gain deep insights into the internal state of distributed systems and achieve intelligent fault location and prediction, specifically: 1. OpenTelemetry Data Collection: Integrate the OpenTelemetry SDK into all application services to automatically collect distributed tracing, metrics, and log data.

[0204] 2. Unified backend storage: Send the collected data to a unified backend storage (e.g., Prometheus / M3DB stores metrics, Loki / Elasticsearch stores logs, Jaeger / Tempo stores traces).

[0205] 3. Correlation Analysis: By using a unified Trace ID, cross-service request chains are linked together to achieve end-to-end request tracing.

[0206] 4. AIOps module: Anomaly detection: Use Isolation Forest or LSTM algorithms to perform real-time anomaly detection on the indicator data.

[0207] Root cause analysis: Using causal inference or the PageRank algorithm, automatically locate the root cause of a fault.

[0208] Trend prediction: Using ARIMA or Prophet algorithms, predict system capacity to achieve predictive expansion.

[0209] This implementation can reduce the mean time to fault location (MTTD) from hours to minutes and reduce the mean time to fault recovery (MTTR) by more than 50%. Example 20

[0210] This implementation method discloses the construction of a multi-language SDK and a developer sandbox, aiming to solve the technical problem of how to lower the integration threshold of third-party course platforms and accelerate the construction of the ecosystem, specifically as follows: 1. API-first design: Define all API interfaces based on the OpenAPI specification as the single source of truth.

[0211] 2. SDK code generation: Using tools such as OpenAPI Generator, SDK skeletons for languages ​​such as Python, JavaScript, Java, and Go are automatically generated from the specifications.

[0212] 3. Manual SDK Optimization: Manually optimize the automatically generated SDK to add language-specific best practices (e.g., asyncio support for Python, CompletableFuture for Java).

[0213] 4. Developer Sandbox: Tenant isolation: Assign an independent sandbox API key to each developer to achieve data logic isolation.

[0214] Scenario simulation: Trigger specific test scenarios through special request headers (such as "X-Sandbox-Scenario").

[0215] Webhook Testing: Provides a visual representation of the Webhook receiver, making it easier for developers to debug callbacks.

[0216] This implementation method can reduce the average integration time of third-party platforms from weeks to days, significantly improving the developer experience. Example 21

[0217] This embodiment discloses multi-credential standard support, aiming to solve the technical problem of how to ensure that learning credentials issued by this system can be widely accepted and verified globally, specifically: 1. Internal Unified Voucher Model: Design an internal unified voucher model based on the W3C VC Data Model 2.0 extension, containing detailed information on all learning processes and results.

[0218] 2. Voucher Conversion Engine: Implement a voucher conversion engine responsible for bidirectional conversion between internal models and external standard formats.

[0219] 3. Multi-format adapter: W3C VC Adapter: Maps internal models to the standard W3C VC JSON-LD format.

[0220] Open Badges 3.0 Adapter: Aggregates internal models into an achievement badge format.

[0221] CLR 2.0 Adapter: Constructs the internal model into a comprehensive learning record format.

[0222] ISO mdoc adapter: Exports the core identity attributes of the internal model to CBOR format.

[0223] Europass EDC Adapter: Converts internal models to a format compatible with the EU education system.

[0224] 4. Interoperability Testing: Establish an interoperability testing platform and conduct regular interoperability tests with mainstream digital wallets, badge platforms, and educational systems.

[0225] This implementation method enables the vision of "one-time authentication, globally applicable", allowing learners to export credentials to any platform that supports the above standards. Example 22

[0226] This embodiment discloses an application example of formative evaluation evidence storage: Scenario description: Learner Xiao Wang completes a Python basic knowledge diagnostic test in an AI-assisted programming course.

[0227] Implementation steps: 1. Evaluation Initialization: The system records evaluation metadata (type = diagnostic test, knowledge point = [variable, data type, control structure], number of questions = 10, total score = 100), generates an evaluation session ID, and begins authentication.

[0228] 2. Answer process record: The system records detailed response data for all questions. For example, for question 1 (multiple choice), the response time is 45 seconds, the answer is B, and the result is correct; for question 3 (code question), the response time is 300 seconds, and the number of modifications is 3.

[0229] 3. Behavioral evidence collection: The system collects behavioral evidence such as total answering time (25 minutes), average time per question (2.5 minutes), number of answer modifications (5 times), number of pauses (2 times), and number of focus loss (0 times).

[0230] 4. Result Calculation and Evidence Storage: The system calculates the total score (85 / 100), the score for each knowledge point, and the mastery level (proficiency), then generates an evidence hash and links it to the learner's learning history. The hash chain is as follows: Figure 9 As shown.

[0231] 5. Credential Generation: The system automatically generates a "Python Basics - Proficiency Level" formative assessment achievement credential. Implementation Method 23

[0232] This embodiment discloses an application example of reflection log evidence storage: Scenario description: After learner Xiao Wang completed the above diagnostic test, the system triggered the creation of a reflection log.

[0233] Implementation steps: 1. Reflection Trigger: The system records the trigger type (evaluation completion trigger) and the associated evaluation (Python basic diagnostic test).

[0234] 2. Structured Reflection Collection: The system guides learners to conduct structured reflection from five dimensions: description, emotion, cognition, evaluation, and planning. For example, in the cognitive dimension, learners record concepts they understand (variable scope) and concepts they are unclear about (complex nested conditions).

[0235] 3. Quality Assessment: The system automatically assesses the quality of reflection, such as the depth of reflection (Level 3: analytical), dimensional coverage (5 / 5), and specificity score (4 / 5).

[0236] 4. Evidence Generation: The system encrypts the reflection content, then stores the quality indicators and content summary as evidence, links them to the corresponding formative assessment records, and updates the learning resume.

[0237] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0238] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for verifying and authenticating a learning process, characterized in that, The specific steps are as follows: S1. Capture learners' learning behavior data in real time through embedded acquisition modules or standard protocol interfaces, and generate structured learning event records. Learning events include at least learner identifier, behavior type, behavior object, timestamp, and context information. S2. The learning event is verified, including but not limited to one or more combinations of identity verification, behavior authenticity verification, content originality verification, and environment credibility verification. S3. Aggregate multiple verified learning events in batches according to preset rules, generate an aggregation proof using a cryptographic accumulator, and anchor the aggregation proof to an immutable distributed ledger. S4. Based on the evidence-based learning events, a dynamic learning history is generated through an event aggregation algorithm, and the dynamic learning history is stored in a distributed storage system; S5. Based on preset multidimensional authentication standards, automatically evaluate the dynamic learning resume and generate authentication results; S6. Based on the authentication result, generate a digital certificate that conforms to the verifiable credential standard through a smart contract or trusted issuance service, and record the credential metadata of the digital certificate in the distributed ledger.

2. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, S1 is implemented through a general adapter architecture that supports at least one mainstream learning technology standard, including one or more combinations of Learning Tool Interoperability Protocol (LTI), Experience Application Programming Interface (xAPI), Learning Analytics Standard (Caliper), and Shared Content Object Reference Model (SCORM).

3. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The learner identifier mentioned in S1 is a decentralized identity identifier (DID).

4. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The learning event recording generation process in S1 also includes offline learning events, the processing steps of which are as follows: When in offline mode, the generated offline learning events are stored in a local encrypted database; Using the device key, each of the offline learning events in the encrypted database is digitally signed; Once the network is restored, the digitally signed offline learning events will be synchronized in batches via a secure communication link.

5. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The authentication described in S2 is continuous authentication, and the specific steps include: Initial authentication: Verifying the learner's identity at the start of a learning session; Periodic identity re-authentication: During the learning process, periodic identity authentication is performed through a random response mechanism; Continuous behavioral feature verification: During the learning process, learners’ behavioral biometric data are continuously analyzed and compared with the learners’ historical behavioral feature baseline model to evaluate the consistency of identity in real time.

6. The learning process verifiable evidence storage and authentication method according to claim 5, characterized in that, The keystroke dynamics features in the behavioral biometric data include at least one of keystroke duration, keystroke interval, keystroke pressure, or input error rate.

7. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The authenticity verification of behavior in S2 is achieved by analyzing the temporal, statistical, or frequency characteristics of the learning behavior sequence and using anomaly detection algorithms to identify abnormal learning behaviors that do not conform to the preset behavior pattern.

8. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, In step S2, the content originality verification uses an artificial intelligence model to perform originality analysis and quality assessment on the text, code, or image content submitted by the learner.

9. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The environment trustworthiness verification in S2 verifies the learning environment in which the learner is located by at least one of the Trusted Execution Environment (TEE), device fingerprint, or geographic location information.

10. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The cryptographic accumulator in S3 is one of Merkle tree, Patricia tree, Weckel tree, RSA accumulator, or bilinear accumulator.

11. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The immutable distributed ledger in S3 is one of the following: public blockchain, consortium blockchain, private blockchain, or directed acyclic graph (DAG) structure.

12. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, In S3, aggregated proofs include formative evaluation proofs and reflective journal proofs.

13. The learning process verifiable evidence storage and authentication method according to claim 12, characterized in that, The formative evaluation evidence generation steps are as follows: Obtain response data related to formative assessment events; Based on the answer data, evaluation metadata for the formative assessment event is collected. The evaluation metadata includes evaluation type, related knowledge points, difficulty level, and scoring criteria. Based on the answer data, the learner's answer process data in the formative assessment event is recorded, and the answer process data includes the response content, response time and answer order for each question; Based on the evaluation metadata and the answer process data, the evaluation result data is calculated and generated, which includes the total score, the score of each knowledge point, and the judgment result of the mastery level of each knowledge point. Record learners’ behavioral evidence data during the question-answering process, including the distribution of answering time, the number of answer modifications, and the number of pauses; Based on the answer process data, evaluation result data, and behavioral evidence data, a formative assessment record is generated and linked to the learner's dynamic learning history.

14. The learning process verifiable evidence storage and authentication method according to claim 12, characterized in that, The steps for generating the reflection log evidence are as follows: Obtain the answer data related to the reflection log event; Based on the answer data, a reflection trigger context is recorded, which includes the trigger type and the associated learning event; Based on the collected answer data and the structured reflection content corresponding to the reflection trigger context, the structured reflection content includes content data in the dimensions of description, emotion, cognition, evaluation, and planning. The structured reflection content is evaluated based on reflection quality indicators; Based on the structured reflection content and the reflection quality indicators, a reflection log is generated and stored, and the reflection log is associated with the learner's dynamic learning history.

15. The learning process verifiable evidence storage and authentication method according to claim 3, characterized in that, The dynamic learning resume generation step described in S4 supports the aggregation of learning events from multiple learning institutions, and realizes the association and integration of cross-institutional learning event records through the decentralized identity identifier (DID).

16. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The dynamic learning resume described in S4 supports multi-view display, including at least one of the following: timeline view, skill tree view, or project set view.

17. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The distributed storage system described in S4 is a content-addressable storage system, including one of the following: InterPlanetary File System (IPFS), Arweave, Filecoin, or Storj.

18. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, In S5, automated assessment uses artificial intelligence models for assisted evaluation, including assessment of learning outcome quality, analysis of learning engagement, or prediction of ability level.

19. A learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The verifiable credentials standard described in S6 is compatible with at least one of the W3C Verifiable Credentials Standard, the Open Badge Standard, or the Integrated Learning Record Standard.

20. A learning process verifiable evidence storage and authentication method according to claim 12, characterized in that, The verifiable credentials mentioned in S6 include formative assessment achievement credentials and reflective learner credentials, wherein the formative assessment achievement credentials demonstrate the learner's level of mastery in a specific domain; and the reflective learner credentials demonstrate the learner's reflective learning ability and continuous improvement awareness.

21. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, After digital certificates are issued in S6, a lifecycle management mechanism for digital certificates is also included, supporting the updating, revocation, inheritance, and version control of digital certificates.

22. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, It also includes credential verification, the specific steps of which are: based on the digital certificate, a proof is generated through a zero-knowledge proof mechanism to prove that the learner meets specific conditions, and to verify their learning achievements or abilities without exposing the complete dynamic learning history.

23. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, It also includes credential presentation, the specific steps of which are: based on the digital certificate, a verifiable representation is generated according to the learner's authorization strategy through a selective disclosure mechanism. The verifiable representation contains some attribute information from the digital certificate and maintains the verifiability of the attribute information.

24. The learning process verifiable evidence storage and authentication method according to claim 23, characterized in that, The selective disclosure mechanism is implemented based on the Idemix anonymous credential scheme or CL signature scheme, or the selectively disclosed JWT or BBS+ signature scheme.

25. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The method also supports the preservation of evidence of collaborative learning processes, enabling the recording of the individual contributions and interactions of multiple learners in the same learning activity.

26. A learning process verifiable evidence storage and authentication method according to claim 25, characterized in that, Collaborative learning evidence storage also includes: recording the specific contributions of each learner based on a version control system; calculating the contribution ratio of each learner using a contribution metric algorithm; and generating a multi-party signature certificate for the collaborative learning outcome.

27. The learning process verifiable evidence storage and authentication method according to claim 1, characterized in that, The method also includes a trust assessment model to comprehensively evaluate the issuing authority's reputation, the history of digital certificate issuance, and community acceptance.

28. A learning process verifiable evidence storage and authentication system, characterized in that, include: The learning application layer is used to interact with learners and capture learning behavior data. The learning application layer includes a general course platform adapter, which supports at least one of the following: Learning Tool Interoperability Protocol (LTI), Experience Application Programming Interface (xAPI), Learning Analytics Standard (Caliper), and Shared Content Object Reference Model (SCORM). An edge computing layer, deployed on learner client devices, is used to preprocess, de-identify, or cache the learning behavior data offline, and generate learning events. The event processing layer is used to perform multidimensional verification and batch processing on received learning events, and to generate aggregate proofs using a cryptographic accumulator, anchoring the aggregate proofs to an immutable distributed ledger; a dynamic learning history is generated based on the aggregate proofs, and the multidimensional verification includes identity verification, behavior authenticity verification, content originality verification, and environment trustworthiness verification; wherein, the aggregate proofs include formative evaluation notarization and reflective log notarization. The authentication and service layer is used to monitor the stored records on the distributed ledger, generate dynamic learning resumes based on the stored learning events, perform automated authentication and evaluation of the dynamic learning resumes according to the preset multi-dimensional authentication standards, and issue digital certificates that conform to the verifiable credential standard through smart contracts or trusted issuance services. The data and storage layer is used to store learning events and dynamic learning history; Security and privacy layer, used to provide zero-knowledge proof generation, selective disclosure control, or key management services.

29. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The event processing layer includes a multidimensional verification engine, which includes: The identity verification module is used to perform initial identity authentication, periodic identity re-authentication, and continuous behavioral feature verification; The behavior authenticity verification module is used to analyze the temporal, statistical, or frequency characteristics of learning behavior sequences to identify abnormal learning behaviors. The content originality verification module is used to perform originality analysis on the text, code, or image content submitted by learners using an artificial intelligence model; The environment trustworthiness verification module is used to verify the learning environment through a trusted execution environment, device fingerprint, or geolocation information.

30. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The event processing layer also includes a batch evidence processor, which is used to aggregate multiple verified learning events in batches according to a preset time window or event quantity threshold; generate the aggregated proof using the cryptographic accumulator; and anchor the aggregated proof to an immutable distributed ledger.

31. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The security and privacy layer includes a zero-knowledge proof engine, which generates zero-knowledge proofs that prove learners meet specific conditions; and verifies learners' learning achievements or abilities without exposing their complete dynamic learning history.

32. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The security and privacy layer includes a selective disclosure controller, which is used to select partial attribute information from the digital certificate according to the learner's authorization policy; generate a verifiable representation containing the partial attribute information; and maintain the verifiability of the partial attribute information.

33. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The security and privacy layer includes a key management service for generating and storing the learner's public and private key pairs; managing the key lifecycle, including key rotation and key recovery; and providing secure signature and encryption services.

34. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The learning application layer includes at least one of web applications, mobile applications, embedded software development kits (SDKs), or browser plugins.

35. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The edge computing layer is deployed on the learner's client device.

36. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The data and storage layer includes an event database, a history database, a distributed storage cluster, and a blockchain network.

37. A learning process verifiable evidence storage and authentication system according to claim 36, characterized in that, The event database adopts a time-series database architecture, supporting time range queries and event sequence analysis.

38. A learning process verifiable evidence storage and authentication system according to claim 36, characterized in that, The resume database adopts a graph database architecture, which supports querying the relationships between the learning events and analyzing the learning path.

39. A learning process verifiable evidence storage and authentication system according to claim 36, characterized in that, The blockchain network supports smart contracts for automating the execution of digital certificate issuance, verification, and revocation logic.

40. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The authentication and service layer includes an open verification service, which provides third parties with an application programming interface (API) for verifying the authenticity and validity of digital certificates.

41. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The learning application layer includes a general course platform adapter, which is used to convert data from different learning technology standards into a unified internal event format.

42. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The authentication and service layer includes an identity federation service to support the interoperability of the learner's decentralized identity (DID) with other identity authentication systems.

43. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The system also includes a learning path recommendation engine, which recommends personalized learning activities for the learner based on the learner's historical learning history, formative assessment records, and reflection log records.

44. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The system also includes a learner competency profile module, which is used to construct the learner's knowledge graph based on the formative assessment evidence; track the learner's growth trajectory in each competency dimension; and generate a visualized competency radar chart.

45. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The system supports a time travel query function, allowing users to query learners' learning history status and ability level at any historical point in time.

46. ​​A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The system adopts a plug-in architecture, which supports the extension of verification algorithms, evidence storage strategies and digital certificate formats through plug-ins.

47. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The system provides a data export interface that allows learners to export all their personal learning data in a machine-readable format that complies with GDPR's "data portability right".

48. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The system supports cross-regional disaster recovery mechanisms, including: multi-data center deployment; real-time data synchronization; automatic failover; and disaster recovery drills.

49. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The system includes an observability monitoring module for collecting system operation metrics and logs; providing real-time alarms and anomaly detection; and supporting distributed tracing.

50. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The system provides third-party integration interfaces, including: RESTful API interface; software development kit (SDK); webhook callback mechanism; and OAuth2.0 authorization service.

51. A learning process verifiable evidence storage and authentication system according to claim 28, characterized in that, The authentication and service layer includes a multi-credential standard conversion engine for format conversion between different credential standards, including W3C verifiable credentials, open badges, and integrated learning records.

52. A method for constructing a decentralized trust system, characterized in that, The specific steps are as follows: A decentralized identity (DID) is created for the learner, which is controlled by the learner and does not depend on any centralized institution. Associate the learner's learning events, dynamic learning history, and digital certificate with the decentralized identity identifier (DID); The credential metadata of the digital certificate is recorded in a distributed ledger, enabling the digital certificate to be verified independently.

53. A continuous authentication method, characterized in that, The specific steps are as follows: Initial identity authentication: At the start of the learning session, the learner's identity is confirmed through facial recognition, document verification, or multi-factor authentication; Periodic identity re-authentication: During the learning process, identity verification is performed periodically at unpredictable time intervals through random facial verification or challenge-response mechanisms. Continuous behavioral feature verification: Throughout the learning process, learners’ behavioral biometric data are continuously collected and compared with the learners’ historical behavioral feature baseline model to assess the consistency of identity in real time. The behavioral biometric data includes at least one of keystroke dynamics features, mouse movement trajectory features, or touch screen interaction features.

54. A continuous authentication method according to claim 53, characterized in that, The keystroke dynamics features include at least one of keystroke duration, keystroke interval, keystroke pressure, or input error rate.

55. A method for formative evaluation and evidence preservation, characterized in that, The specific steps are as follows: Obtain response data related to formative assessment events; Based on the answer data, evaluation metadata is collected, which includes evaluation type, related knowledge points, difficulty level and scoring criteria. Record learners' answer process data, which includes the response content, response time, and answer order for each question; Calculate and generate evaluation result data, which includes the total score, the score for each knowledge point, and the mastery level assessment result; Record evidence data of answering behavior, including the distribution of answering time, the number of times answers were modified, and the number of times paused; A question-answering response record is constructed based on the question-answering process data, the evaluation result data, and the question-answering behavior evidence data. An incremental hash chain structure is adopted, in which each answer response record is linked with the previous answer response record to form a hash chain, and the final hash value of the incremental hash chain is anchored to an immutable distributed ledger.

56. A formative evaluation evidence preservation method according to claim 55, characterized in that, The verification of the authenticity of the formative assessment events includes: performing periodic identity verification during the assessment process; analyzing answer behavior patterns and detecting abnormal behavior; verifying the rationality of the assessment timestamps; and checking the consistency between the answer content and the learning records.

57. A formative evaluation evidence preservation method according to claim 56, characterized in that, The formative assessment evidence generated based on the formative assessment events supports multiple assessment types, including diagnostic tests, exercises, in-class quizzes, learning checkpoints, self-assessment, and peer assessment. Each assessment type employs a corresponding data collection and evidence storage strategy.

58. A formative evaluation evidence preservation method according to claim 57, characterized in that, Formative assessment records include dynamic tracking of learners’ ability levels across various knowledge points, forming a trajectory of learners’ ability growth.

59. A formative evaluation evidence preservation method according to claim 56, characterized in that, The formative evaluation evidence storage adopts an incremental hash chain structure, where each answer response record forms a hash chain with the previous record, ensuring the integrity and immutability of the evaluation process.

60. A formative evaluation evidence preservation method according to claim 53, characterized in that, It supports automatically triggering the creation of a reflection log after formative assessment is completed, and generating a comprehensive learning progress report based on the assessment results data.

61. A method for storing reflective log evidence, characterized in that, The specific steps for documenting a reflection log are as follows: Obtain the answer data related to the reflection log event; Based on the answer data, a reflection trigger context is recorded, which includes the trigger type and associated learning events; The structured reflection content is collected based on the answer data, and the structured reflection content includes descriptive dimension, emotional dimension, cognitive dimension, evaluation dimension and planning dimension; The structured reflection content is evaluated based on reflection quality indicators; this includes an automated evaluation of the depth level of the structured reflection content based on a preset cognitive hierarchy model. The reflection content is encrypted and stored, and the reflection quality assessment results and content summaries are also preserved as evidence.

62. A method for storing reflective log evidence according to claim 61, characterized in that, The authenticity verification of the reflection log includes: analyzing writing behavior fingerprints, including typing rhythm, editing patterns, and pauses in thinking; checking the consistency between structured reflection content and learning records; verifying the chronological relationship between reflection time and learning events; and performing originality checks.

63. A method for storing reflective log evidence according to claim 61, characterized in that, The assessment of the depth level is based on the cognitive hierarchy model, which divides the depth of reflection into five levels: descriptive reflection, interpretive reflection, analytical reflection, critical reflection, and transformational reflection.

64. A method for storing reflective log evidence according to claim 61, characterized in that, The reflection log storage supports privacy protection, including: encrypting and storing the reflection content; storing only the content summary and reflection quality indicators to the distributed ledger; and allowing learners to control the scope of disclosure of the reflection content.

65. A method for storing reflective log evidence according to claim 62, characterized in that, The quality assessment metrics for the reflection log also include text sentiment analysis of the reflection content, used to quantify the learner's emotional state during the reflection process.

66. A method for storing reflective log evidence according to claim 61, characterized in that, It also supports linking reflection logs with corresponding formative assessment records and generating comprehensive learning progress reports based on the reflection content.

67. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the learning process described in claims 1-27, which can verify the evidence storage and authentication methods.

Images