Enterprise knowledge updating method based on knowledge version management and dynamic evidence checking

Abstract

The application discloses an enterprise knowledge updating method based on knowledge version management and dynamic evidence checking, and belongs to the technical field of enterprise knowledge management and data governance. The method comprises the following steps: determining a target knowledge item and a knowledge difference item; generating a candidate updating version and setting it as to be verified, and keeping the current effective version unchanged during the verification period; obtaining verification evidence from at least two enterprise internal business data sources to establish an evidence coverage matrix; determining a conflict state, and combining the effective evidence coverage of the key knowledge difference item to determine whether the candidate version is effective or not in combination with the unresolved conflict; and when new business records form new verification evidence, only the affected difference item is locally updated. The application realizes fine-grained verification, conflict control and incremental re-verification in the enterprise internal knowledge updating process by associating the knowledge difference item, the evidence coverage state, the evidence conflict state and the knowledge version state conversion, and reduces the risk of entering the effective knowledge base without sufficient verification.

Description

Technical Field

[0001] This application belongs to the field of enterprise knowledge management and data governance technology, specifically involving an enterprise knowledge update method based on knowledge version management and dynamic evidence verification, which is applicable to the trusted update, version control and traceability management of enterprise internal knowledge bases in the process of continuous change. Background Technology

[0002] As a crucial support for enterprise operational decision-making, business process execution, and employee training, internal knowledge bases require continuous updates to reflect the latest systems, rules, operating procedures, product information, and business standards. However, updating internal knowledge faces numerous challenges: newly generated knowledge content may not be fully validated, and directly updating it to the knowledge base could lead to the retrieval and application of erroneous knowledge; knowledge changes often involve multiple business processes, requiring validation evidence from different business data sources; and knowledge version evolution needs to be traceable so that errors can be rolled back to the correct version.

[0003] In existing technologies, one type of approach focuses on version management of the knowledge base, enabling version tracking of knowledge content by establishing version identifiers and version history. However, this type of approach typically lacks a verification mechanism for new knowledge content, failing to ensure the credibility of new versions. Another type of approach focuses on knowledge credibility assessment, determining whether to update the knowledge base by calculating a credibility score for the knowledge content. However, this type of approach is often based on a single-dimensional score, lacking fine-grained evidence verification for knowledge discrepancies and struggling to handle conflicts between multiple pieces of verification evidence.

[0004] For example, publicly available dynamic knowledge base update methods improve update efficiency by using preprocessing, semantic segmentation, and embedding models to locate target paragraphs for fine-grained updates, but lack a verification mechanism for new knowledge content, failing to ensure the credibility of the updated content. Publicly available dependency graph-based dynamic knowledge consistency governance and gating methods achieve knowledge consistency governance by constructing dependency graphs, identifying affected subgraphs, and generating control results through differential consistency verification, but primarily focus on the consistency verification of downstream dependent data rather than the verification of the new knowledge content itself.

[0005] Furthermore, existing technologies typically employ a one-time verification model when handling internal knowledge updates within enterprises. This means that new knowledge content is verified upon submission, and the knowledge base is updated directly after successful verification. This model has significant drawbacks: when verification evidence is insufficient, new knowledge content may be rejected or its update delayed; and when new verification evidence subsequently emerges, the system lacks a re-verification mechanism, potentially preventing valuable knowledge content from taking effect in a timely manner.

[0006] Therefore, existing technologies are insufficient to simultaneously meet the comprehensive needs of enterprises' internal knowledge bases for credible verification of new knowledge content, fine-grained evidence coverage verification, resolution of multiple evidence conflicts, and incremental re-verification during continuous changes. There is an urgent need for a knowledge base update method that can integrate knowledge version management with dynamic evidence verification. Summary of the Invention

[0007] This application provides an enterprise knowledge update method based on knowledge version management and dynamic evidence verification, aiming to solve the problems in the existing technology of lack of credible verification mechanism for enterprise internal knowledge updates, difficulty in handling fine-grained evidence coverage and multiple evidence conflicts, and lack of incremental re-verification capability, so as to realize credible updates and version traceability of enterprise internal knowledge in the process of continuous change.

[0008] The enterprise knowledge update method based on knowledge version management and dynamic evidence verification provided in this application includes: The process involves acquiring knowledge content generated within the enterprise that needs to be updated, comparing the knowledge content to be updated with the current valid knowledge version in the enterprise's internal knowledge base, and determining at least one target knowledge item and at least one knowledge difference item corresponding to each target knowledge item based on the comparison results. Based on the current valid knowledge version of the target knowledge item and the knowledge difference item, a candidate update version is generated, and an independent version identifier is assigned to the candidate update version. The candidate update version is set to a pending verification state. During the period when the candidate update version is in the pending verification state, the current valid knowledge version is maintained as the current valid version of the target knowledge item. For each knowledge difference item in the candidate update version, according to the difference type and business scope of the knowledge difference item, corresponding verification evidence is obtained from at least two internal business data sources of the enterprise, and each verification evidence is associated with at least the corresponding knowledge difference item, the source of the verification evidence, and the generation time of the verification evidence to form an evidence association set corresponding to the candidate update version. An evidence coverage matrix is ​​established based on the evidence association set. The evidence coverage matrix uses the knowledge difference item as the difference item index and the verification evidence as the evidence index, and records the coverage status of each verification evidence for the corresponding knowledge difference item. For multiple verification evidences corresponding to the same knowledge difference item, corresponding evidence assertions are generated based on the verification conclusions contained in each verification evidence, and conflict assertion relationships are established based on the exclusion relationship between different evidence assertions. The conflict state of the knowledge difference item is determined based on the conflict assertion relationships. Based on the evidence coverage matrix, determine whether each key knowledge difference item has valid evidence coverage, and based on the conflict assertion relationship, determine whether there are unresolved conflicts for each key knowledge difference item. When all key knowledge discrepancies are covered by valid evidence and there are no unresolved conflicts, the candidate update version is changed from the pending verification state to the effective state, and the candidate update version is determined as the new current effective knowledge version of the target knowledge item. At the same time, the original current effective knowledge version is converted into a historical knowledge version, and a version update record containing version identifier, state transition information and evidence association information is established. When at least one key knowledge difference item does not have valid evidence coverage, or at least one key knowledge difference item has unresolved conflicts, the candidate update version is kept in a pending verification state, and the current valid knowledge version remains unchanged. During the period when the candidate update version is in the pending verification state, the internal business data source of the enterprise is monitored. When a new business record is obtained, the new business record is associated and matched with the knowledge difference items in the candidate update version in the pending verification state to determine whether the new business record constitutes new verification evidence corresponding to at least one knowledge difference item. When the newly added business record constitutes new verification evidence, only the coverage status of the knowledge difference item corresponding to the new verification evidence in the evidence coverage matrix is ​​updated, and only the evidence validity check and conflict assertion relationship check are re-executed for the knowledge difference item, thereby determining whether the candidate update version containing the knowledge difference item meets the effective conditions, and performing the state transition of the candidate update version according to the effective conditions.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: The enterprise knowledge update method based on knowledge version management and dynamic evidence verification provided in this application decomposes the knowledge update process into specific knowledge difference items and establishes a correspondence between knowledge difference items and verification evidence. At the same time, it can identify conflicting evidence corresponding to the same knowledge difference item, and only reprocess the affected candidate versions and knowledge difference items when new evidence appears. This achieves reliable updates and version traceability of enterprise internal knowledge in the process of continuous change, and reduces the risk of unverified knowledge polluting the knowledge base. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A schematic diagram of an enterprise knowledge update method based on knowledge version management and dynamic evidence verification provided in an embodiment of this application; Detailed Implementation This application provides an enterprise knowledge update method based on knowledge version management and dynamic evidence verification, which addresses the technical problems in existing technologies such as the lack of a reliable verification mechanism for internal enterprise knowledge updates, difficulty in handling fine-grained evidence coverage and multiple evidence conflicts, and lack of incremental re-verification capabilities.

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0013] It should be noted that the terms "first," "second," etc., used in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such numerals can be used interchangeably where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, apparatus, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0014] It should be noted that the "knowledge item" mentioned in this application is used to characterize a knowledge unit with independent business meaning in the enterprise's internal knowledge base, including at least categories such as rules and regulations, operating procedures, product standards, business parameters, and permission configurations; the "knowledge version" is used to characterize the content status of the same knowledge item at different points in time, and each knowledge version has an independent version identifier; the "knowledge difference item" is used to characterize the specific differences between the knowledge content to be updated and the currently valid knowledge version, and is the smallest knowledge change unit that can be independently verified; the "verification evidence" is used to support the validity of the knowledge difference item through internal business data records of the enterprise, including at least categories such as policy documents, approval records, business system records, operation logs, and financial records.

[0015] It should be noted that the "key knowledge difference items" mentioned in this application are categories of knowledge difference items classified according to their degree of impact on the enterprise's business execution results. Among them, differences involving business rules, operation steps, parameter thresholds, permission conditions, effective conditions, or security constraints are identified as key knowledge difference items. The "effective evidence coverage" is used to characterize that there is effective verification evidence sufficient to support the difference for the knowledge difference item. The "unresolved conflict" is used to characterize that there is a content contradiction between multiple verification evidences for the same knowledge difference item that cannot be resolved by preset conflict resolution rules.

[0016] Example 1, as Figure 1 As shown, this application provides an enterprise knowledge update method based on knowledge version management and dynamic evidence verification, the method including: S100: Obtain knowledge content to be updated generated within the enterprise, compare the knowledge content to be updated with the current valid knowledge version in the enterprise's internal knowledge base, and determine at least one target knowledge item and at least one knowledge difference item corresponding to each target knowledge item based on the comparison result. In this embodiment, the knowledge content to be updated includes at least: newly released policy documents, updated operation manuals, revised business rules, adjusted product parameters, and changed permission configurations. This content may originate from the enterprise's internal content management system, document collaboration platform, business system interfaces, or manual upload channels.

[0017] In this embodiment, the difference comparison process includes: performing structured parsing on the knowledge content to be updated and extracting the set of knowledge elements therein; comparing the set of knowledge elements with the corresponding knowledge elements in the current valid knowledge version to identify at least one of the following: content differences, attribute differences, parameter differences, operation step differences, applicable condition differences, business rule differences, permission condition differences, or effective time differences; and determining the target knowledge item and knowledge difference item based on the identified difference type and difference range.

[0018] Specifically, taking the update of a company's expense reimbursement policy as an example: the original, currently valid knowledge version stipulated that "reimbursement amounts exceeding 5,000 yuan require approval from the department head," while the updated knowledge version stipulates that "reimbursement amounts exceeding 3,000 yuan require approval from the department head." The system, through structured parsing, identified the target knowledge item as "reimbursement approval rules," and the knowledge difference item as "the approval amount threshold has changed from 5,000 yuan to 3,000 yuan." This knowledge difference item is the smallest unit of knowledge change that can be independently verified.

[0019] In this embodiment, the difference comparison can employ at least one of the following: a text comparison algorithm based on semantic similarity, a structure comparison algorithm based on abstract syntax trees, or a parameter comparison algorithm based on rule templates. For highly structured knowledge content, such as business rules and parameter configurations, the parameter comparison algorithm based on rule templates is preferred; for unstructured or semi-structured knowledge content, such as policy documents and operation manuals, the text comparison algorithm based on semantic similarity is preferred.

[0020] S101: Obtain the knowledge content to be updated and determine the corresponding currently valid knowledge version; Specifically, the system acquires knowledge content generated within the enterprise that needs to be updated, and retrieves knowledge items corresponding to the knowledge content in the enterprise's internal knowledge base based on at least one of the business object identifier, business item identifier, knowledge category identifier, business field identifier, or source data identifier contained in the knowledge content to be updated.

[0021] In this embodiment, the knowledge content to be updated can originate from internal enterprise documents, business rule change notices, operation process adjustment records, product parameter change records, business system configuration change records, approval rule adjustment records, enterprise announcements, or other internal business data. When acquiring the knowledge content to be updated, metadata such as the source identifier, generation time, publication time, submitting entity, and business system identifier corresponding to the knowledge content to be updated are also acquired.

[0022] Specifically, for the acquired knowledge content to be updated, information such as preprocessed business objects, business matters, business fields, and applicable scope is read, and the information is matched with knowledge items in the knowledge base to determine the target knowledge item candidate set corresponding to the knowledge content to be updated.

[0023] When there are multiple candidate knowledge items in the target knowledge item candidate set, the knowledge item corresponding to the knowledge content to be updated is determined according to at least one of the following: business object matching degree, business matter matching degree, scope of application matching degree, and knowledge category matching degree.

[0024] After identifying the corresponding knowledge item, read the version index information of the knowledge item, obtain the knowledge version associated with the current valid version of the knowledge item, and determine the knowledge version associated with the current valid version of the knowledge item as the current valid knowledge version corresponding to the knowledge content to be updated.

[0025] In this embodiment, the currently valid knowledge version includes at least a version identifier, a knowledge item identifier, version content, version generation time, effective time, applicable scope, and status identifier. By establishing a correspondence between knowledge item identifiers and currently valid knowledge versions, only one currently valid knowledge version exists for knowledge retrieval and business applications at any given time for the same knowledge item.

[0026] S102: Establish the correspondence between the knowledge content to be updated and the currently valid knowledge version; Specifically, the set of knowledge elements formed by preprocessing the knowledge content to be updated is matched with the set of knowledge elements in the current valid knowledge version according to the knowledge item identifier, business object, business field and applicable scope, so as to form an element correspondence relationship between the knowledge elements to be updated and the current valid knowledge elements.

[0027] The knowledge elements include at least one or more of the following: business object, business matter, business field, field value, rule condition, execution action, parameter, parameter threshold, scope of application, permission condition, and time attribute.

[0028] For knowledge elements that can be directly matched by a unique identifier, a direct matching relationship is established based on the unique identifier; for knowledge elements that do not have a unique identifier, a matching relationship is established based on at least one of the following: business object, business matter, field name, field value type, and scope of application.

[0029] For knowledge elements that exist in the knowledge content to be updated but not in the current valid knowledge version, mark the knowledge element as a new element; for knowledge elements that exist in the current valid knowledge version but not in the knowledge content to be updated, mark the knowledge element as an element to be deleted; for knowledge elements that exist in both but have different corresponding attributes or content, mark them as elements to be compared.

[0030] Through the above processing, the knowledge content to be updated and the current valid knowledge version are converted into a set of knowledge elements with corresponding relationships, providing a comparative basis for subsequent difference identification for the same business meaning.

[0031] S103: Compare the differences between the knowledge content to be updated and the current valid knowledge version; Specifically, a comparison of differences is performed item by item between the knowledge elements to be updated and the currently valid knowledge elements for establishing corresponding relationships, and the corresponding comparison method is selected according to the type of knowledge element.

[0032] For business fields, compare the values ​​of the corresponding business fields in the knowledge content to be updated with those in the current valid knowledge version to determine if the business field values ​​have changed. For parameters, compare the parameter name, parameter unit, parameter value range, and parameter threshold to determine if the parameters have changed. For rule conditions, compare the condition objects, comparison relationships, condition values, and logical relationships in the rule conditions to determine if the rule conditions have changed. For execution actions, compare the execution subject, execution operation, and operation sequence to determine if the execution actions have changed. For permission conditions, compare the permission subject, permission level, and permission scope to determine if the permission conditions have changed. For time attributes, compare the release time, effective time, expiration time, and applicable time period to determine if the time attributes have changed. For applicable scope, compare the business object, organizational scope, product scope, personnel scope, geographical scope, and business scenario to determine if the applicable scope has changed.

[0033] Specifically, the difference comparison results are marked according to preset difference types, which include at least new differences, deleted differences, modified differences, conditional differences, parameter differences, time differences, and applicable scope differences.

[0034] For example, the knowledge content in the current valid knowledge version is "reimbursement amount exceeding 5,000 yuan requires department head approval", while the knowledge content to be updated is "reimbursement amount exceeding 3,000 yuan requires department head approval". After comparing the business fields and rule conditions, it was identified that the business object is "expense reimbursement" and the action is "department head approval". However, the approval amount threshold has changed from 5,000 yuan to 3,000 yuan. Therefore, this change is identified as a parameter threshold modification difference.

[0035] S104: Determine the target knowledge items based on the difference comparison results; Specifically, the difference comparison results are aggregated, and one or more differences belonging to the same business object, the same business matter, or having common business semantics are grouped into the same knowledge item, and the knowledge item containing the difference is determined as the target knowledge item.

[0036] In this embodiment, the target knowledge item is used to represent the internal knowledge unit of the enterprise actually involved in the update operation. The target knowledge item has at least a knowledge item identifier, a knowledge item type, a business object identifier, a business matter identifier, and a current valid version identifier.

[0037] Specifically, when the knowledge content to be updated involves only one knowledge item, that knowledge item is identified as the target knowledge item; when the knowledge content to be updated involves multiple knowledge items, the difference comparison results are grouped according to the business object, business matter, knowledge item identifier, and scope of application, and multiple target knowledge items are identified for each group.

[0038] For example, if a new corporate travel policy simultaneously modifies both the "accommodation standard" and the "transportation reimbursement standard," and both are managed as independent knowledge items in the knowledge base, then the "accommodation standard" and "transportation reimbursement standard" knowledge items will be determined as target knowledge items based on the difference comparison results, instead of updating the entire policy document as a single knowledge item.

[0039] By breaking down the knowledge content to be updated into knowledge items, we can avoid updating all unrelated knowledge content as a whole due to a change in one knowledge item, thus enabling subsequent evidence verification to address specific knowledge changes.

[0040] S105: Break down the difference comparison results into knowledge difference items; Specifically, for each target knowledge item, the difference comparison result is split into at least one independently verifiable knowledge difference item based on at least one of the following: business semantic boundary, field boundary, rule condition boundary, parameter boundary, time attribute boundary, and scope of application boundary.

[0041] The knowledge difference term is used to represent a specific knowledge change of the target knowledge item relative to the current effective knowledge version, and serves as the basic unit for subsequent verification of evidence association and evidence coverage judgment.

[0042] Specifically, when a difference involves only one business field or one parameter value, the difference is directly identified as a knowledge difference item; when a difference involves multiple fields, it is further determined whether the multiple fields need to be jointly verified as a business condition; when multiple fields jointly constitute an indivisible business rule condition, it is identified as a knowledge difference item; when multiple fields can be verified separately and the verification results are independent of each other, it is split into multiple knowledge difference items.

[0043] For example, regarding the modification of the "reimbursement approval rules", if the knowledge content to be updated is changed from "approval by the department head for amounts exceeding 5,000 yuan" to "approval by the department head for amounts exceeding 3,000 yuan", then "the threshold for approval amount changed from 5,000 yuan to 3,000 yuan" can be identified as a knowledge difference item.

[0044] For example, if the knowledge content to be updated also modifies "the approval amount threshold is adjusted from 5000 yuan to 3000 yuan" and "the approval entity is changed from the department head to joint approval by the department head and the financial officer," then because the two changes have different business fields and validation criteria, they can be generated separately as follows: Knowledge Difference 1: The threshold for approved amount has been changed from 5,000 yuan to 3,000 yuan; Knowledge Difference Item 2: The approval authority has changed from the department head to both the department head and the financial officer; In this way, during the subsequent evidence verification process, verification evidence from different sources can be matched for knowledge difference item 1 and knowledge difference item 2 respectively, without having to use the entire policy document as a single verification object.

[0045] S106: Establish difference item identifiers and difference attributes for knowledge difference items; Specifically, a unique difference item identifier is assigned to each knowledge difference item, and an association is established between the difference item identifier and the corresponding target knowledge item, the current valid knowledge version, and the knowledge content to be updated.

[0046] In this embodiment, the knowledge difference item is the smallest knowledge change unit that can be independently verified for evidence, and at least the following difference attributes are recorded: difference item identifier, target knowledge item identifier, parent version identifier, content to be updated identifier, difference type, original value, updated value, business object, business field, applicable scope, time attribute, and difference location.

[0047] The original value represents the value of the corresponding knowledge element in the current valid knowledge version, the updated value represents the value of the corresponding knowledge element in the knowledge content to be updated, and the difference type represents the change type such as addition, deletion, or modification.

[0048] For example, regarding changes in the approval amount threshold, the following knowledge difference data can be established: By establishing an independent identifier for each knowledge difference item, the verification evidence obtained subsequently can be directly associated with the specific knowledge difference item, rather than just the entire candidate update version, thus providing a data foundation for forming the correspondence between target knowledge items and knowledge difference items.

[0049] S107: Establish a correspondence between target knowledge items and knowledge difference items; Specifically, after determining the target knowledge item and its corresponding knowledge difference item, a one-to-many correspondence is established between the target knowledge item and the knowledge difference item, wherein one target knowledge item corresponds to at least one knowledge difference item, and one knowledge difference item corresponds to one target knowledge item.

[0050] Furthermore, a current valid knowledge version identifier is recorded for each target knowledge item, and a corresponding difference item identifier is recorded for each knowledge difference item under the target knowledge item. This forms a hierarchical correspondence between the target knowledge item, the current valid knowledge version, and the knowledge difference items. In the subsequent candidate update version generation and evidence verification process, the knowledge difference item is used as the minimum verification object. Verification evidence is obtained, associated, and verified according to each knowledge difference item. Based on the verification results of each knowledge difference item, it is determined whether the candidate update version of the corresponding target knowledge item meets the effective conditions.

[0051] S200: Generate a candidate update version based on the current valid knowledge version of the target knowledge item and the knowledge difference item, assign an independent version identifier to the candidate update version, set the candidate update version to a pending verification state, and maintain the current valid knowledge version as the current valid version of the target knowledge item while the candidate update version is in the pending verification state; In this embodiment, after determining the target knowledge item and its corresponding knowledge difference item, the current valid knowledge version of the target knowledge item is read, and a candidate update version is generated based on the version content of the current valid knowledge version and the knowledge difference item.

[0052] The currently valid knowledge version provides the effective knowledge content of the target knowledge item, while the knowledge difference item defines the specific content that changes in this update relative to the currently valid knowledge version. When generating a candidate update version, the currently valid knowledge version is used as the base version, and only the corresponding knowledge elements are updated according to the knowledge difference item. Unchanged knowledge elements remain consistent with the currently valid knowledge version.

[0053] By using the above methods, a clear version inheritance relationship is established between the candidate update version and the current effective knowledge version, and the specific changes of the candidate update version relative to the current effective knowledge version can be determined based on the knowledge difference items.

[0054] S201: Generate candidate update versions based on the current valid knowledge version and knowledge differences; Specifically, the current valid knowledge version of the target knowledge item is read, and the version content, version identifier, and version status of the current valid knowledge version are obtained; further, each knowledge difference item belonging to the target knowledge item is read, and the corresponding knowledge elements in the current valid knowledge version are updated according to the difference type, original value, and update value recorded in each knowledge difference item.

[0055] For knowledge difference items of the modification type, the original value of the corresponding knowledge element in the current valid knowledge version is replaced with the updated value corresponding to the knowledge difference item; for knowledge difference items of the addition type, the knowledge element corresponding to the knowledge difference item is added to the current valid knowledge version; for knowledge difference items of the deletion type, the corresponding knowledge element in the current valid knowledge version is marked as content to be deleted or removed from the candidate update version.

[0056] In this embodiment, when a target knowledge item corresponds to multiple knowledge difference items, the multiple knowledge difference items are applied to the current valid knowledge version according to their corresponding business fields, rule conditions, parameters, applicable scope and time attributes, thereby forming a candidate update version that includes all changes to be updated.

[0057] For example, the target knowledge item is "reimbursement approval rules." The current valid knowledge version V1 stipulates that "reimbursement amounts exceeding 5,000 yuan require approval from the department head." The content to be updated in this update creates two knowledge discrepancies: D1: The threshold for approved amount has been changed from 5,000 yuan to 3,000 yuan; D2: The approval body has been changed from the department head to both the department head and the financial officer.

[0058] The system reads V1 as the base version and applies the updated content corresponding to D1 and D2 to V1 respectively to form a candidate update version V2. At this time, V2 contains a new approval amount threshold and a new approval subject, while other unchanged knowledge content of V1 remains unchanged.

[0059] S202: Assign a unique version identifier to candidate update versions; Specifically, after generating candidate update versions, an independent version identifier, distinct from the current valid knowledge version, is assigned to each candidate update version, and a parent-child version association is established between the candidate update version and the current valid knowledge version. The independent version identifier uniquely identifies a candidate update version and is used for subsequent association with verification evidence, state transition records, and version update records.

[0060] The candidate update version records at least: candidate version identifier; target knowledge item identifier; parent version identifier; candidate version content; set of knowledge difference item identifiers; creation time; creation subject; and current status. The parent version identifier corresponds to the current valid knowledge version identifier of the target knowledge item. The set of knowledge difference item identifiers is used to record one or more knowledge difference items on which the candidate update version is based. When evidence verification is performed later, all the corresponding knowledge difference items can be located through the candidate version identifier, and the verification evidence corresponding to each knowledge difference item can be further located.

[0061] S203: Set the candidate update version to a pending verification status; Specifically, after completing the generation of candidate update versions and the allocation of version identifiers, the status attribute of the candidate update versions is set to a pending verification state, and the setting time of this state is recorded. The pending verification state indicates that the candidate update version has been generated based on the knowledge content to be updated, but the knowledge differences it contains have not yet fully met the preset evidence verification conditions, and therefore it is not qualified to replace the current valid knowledge version.

[0062] In this embodiment, the version status of a candidate update version includes at least a pending verification status, an effective status, a historical status, and a rejected status. Specifically: a pending verification status indicates that the candidate update version has been generated but has not yet met the preset effective determination conditions, awaiting supplementary verification evidence or undergoing evidence verification; an effective status indicates that the candidate update version has met the preset effective determination conditions and can be used as a valid knowledge version; a historical status indicates that the knowledge version was once in an effective state but has been replaced by a new effective version; and a rejected status indicates that the candidate update version is determined to be ineffective under preset termination conditions, and its verification process is terminated. It should be noted that "currently effective knowledge version" refers to the knowledge version pointed to by the currently effective version association of the target knowledge item. This concept belongs to version association relationships rather than version status enumeration. At any given time, the same target knowledge item establishes a currently effective version association with only one knowledge version in an effective state for external knowledge retrieval and business invocation.

[0063] While a candidate update version is in a pending verification state, only evidence supplementation, evidence verification, and state transition operations are allowed, and it is not provided as the current valid knowledge version of the target knowledge item to business systems that need to use the already effective knowledge.

[0064] S204: Keep the current valid knowledge version unchanged during the verification period; Specifically, while the candidate update version is in the pending verification state, the current valid version identifier of the target knowledge item is kept pointing to the original current valid knowledge version, and the candidate update version is prohibited from replacing the current valid knowledge version.

[0065] For example, while candidate update version V2 is in a pending verification state, the current valid version identifier of the target knowledge item still points to V1: Current version: V1 Candidate update version: V2 V1 continues to provide knowledge content to knowledge retrieval systems, business application systems, and other systems that need to access target knowledge items, while V2 is only saved as a version to be verified.

[0066] Therefore, even if V2 contains new business rules, parameters, or operating procedures, the business system will still execute according to the existing knowledge content corresponding to V1 until it has passed verification.

[0067] S300: For each knowledge difference item in the candidate update version, according to the difference type and business scope of the knowledge difference item, obtain corresponding verification evidence from at least two internal business data sources of the enterprise, and associate each verification evidence with at least the corresponding knowledge difference item, the source of the verification evidence, and the generation time of the verification evidence to form an evidence association set corresponding to the candidate update version. In this embodiment, each knowledge difference item is treated as an independent verification object, and corresponding evidence acquisition and association tasks are established for each knowledge difference item. For multiple knowledge differences items included in the same candidate update version, different internal business data sources can be determined for different knowledge differences items according to their difference type and business applicability, so that the acquired verification evidence can be correlated with specific knowledge changes.

[0068] Specifically, firstly, the difference item identifier, difference type, difference content, and business scope of each knowledge difference item are read; then, the corresponding internal business data source of the enterprise is determined according to the difference type and business scope; verification evidence is obtained from at least two determined business data sources; finally, each verification evidence is associated with the corresponding knowledge difference item, the source of the verification evidence, and the generation time of the verification evidence and stored, and the evidence association records corresponding to each knowledge difference item are summarized to form the evidence association set corresponding to the candidate update version.

[0069] Through the above processing, the knowledge changes in the candidate update version are further divided from overall verification into evidence association structures with knowledge difference items as the smallest verification object, providing a data foundation for subsequent judgment of the evidence coverage of each knowledge difference item and execution of multi-evidence consistency verification.

[0070] S301: Determine the internal business data source of the enterprise based on the difference type and business scope of the knowledge difference items; Specifically, the difference type and business scope of the knowledge difference item are read first. The difference type is used to characterize the specific changes that have occurred to the knowledge difference item, including at least one of the following: changes in business rules, changes in business parameters, changes in operation steps, changes in permission conditions, changes in applicable conditions, and changes in effective time. The business scope is used to characterize at least one of the following: business object, organizational scope, business scenario, product type, job role, or time range corresponding to the knowledge difference item.

[0071] Based on the difference type, the verification evidence requirements corresponding to the knowledge difference items are determined, and based on the business scope, internal business data sources that meet the business scope requirements are further screened.

[0072] For example, for knowledge differences arising from changes in business rules, the system document database and the approval process record database can be identified as business data sources; for knowledge differences arising from changes in business parameters, the business system configuration database and the actual business transaction record database can be identified as business data sources; for knowledge differences arising from changes in operating procedures, the operation manual database and the business operation log database can be identified as business data sources; and for knowledge differences arising from changes in permission conditions, the permission configuration database and the permission change approval record database can be identified as business data sources.

[0073] Furthermore, at least two different internal business data sources should be identified for the same knowledge discrepancy, so that the same knowledge discrepancy can obtain verification evidence from different business data sources.

[0074] In this embodiment, the enterprise's internal business data sources may include at least two of the following: a database of policy documents, a database of business systems, a database of approval processes, a database of operation logs, a database of work orders, a database of financial systems, a database of enterprise announcements, and a database of supply chain management.

[0075] By determining the business data source based on the type of difference and the scope of business application, the scope of verification evidence acquisition is matched with the knowledge changes to be verified, avoiding the use of the same evidence source for different types of knowledge differences and improving the correspondence between subsequent verification evidence and knowledge differences.

[0076] S302: Obtain corresponding verification evidence from at least two internal business data sources of the enterprise; Specifically, for each knowledge difference item, corresponding evidence acquisition conditions are generated based on the difference item identifier, difference type, difference content, and business scope of the knowledge difference item, and evidence acquisition requests are initiated to the determined internal business data sources of the enterprise according to the evidence acquisition conditions.

[0077] The evidence acquisition conditions include at least one of the following: the business object, business scenario, target parameters, rule conditions, effective time range, and applicable organizational scope corresponding to the knowledge difference item.

[0078] For the policy document database, the corresponding policy document can be retrieved based on the policy name, policy number, business rule, or effective date corresponding to the knowledge difference item; for the business system database, the corresponding business system record can be retrieved based on the business object, business field, and parameter value; for the approval process database, the corresponding approval record can be retrieved based on the approval item, approval node, and approval time; for the operation log database, the corresponding operation record can be retrieved based on the business operation type, operation object, and operation time.

[0079] For example, when the knowledge difference item is "the reimbursement approval amount threshold is adjusted from 5,000 yuan to 3,000 yuan", the new version of the reimbursement system document can be obtained from the system document database, the corresponding approval amount configuration record can be obtained from the approval process database, and the execution record after the rule adjustment can be obtained from the actual reimbursement business record, based on the difference type and business scope of the knowledge difference item.

[0080] After obtaining the verification evidence, the original business data returned by different business data sources are processed for evidence-based processing. The business data that can be used to prove or reflect the corresponding knowledge differences are identified as verification evidence, and the source relationship between the verification evidence and the original business data is retained.

[0081] The verification evidence can be rules and clauses in the system documents, parameter configuration records in the business system, approval records in the approval process, execution records in the operation log, or business records in the financial system.

[0082] By obtaining corresponding verification evidence from at least two different internal business data sources of enterprises, the same knowledge difference item has a multi-source verification basis, which provides a basis for subsequent judgment on whether the verification evidence is valid and whether there is a conflict between multiple verification evidences.

[0083] S303: Associate each piece of verification evidence with the corresponding knowledge discrepancy item, the source of the verification evidence, and the time when the verification evidence was generated; Specifically, for each piece of verification evidence obtained, the knowledge difference item identifier, the source of the verification evidence, and the generation time of the verification evidence are read or extracted.

[0084] The source of verification evidence is used to identify the internal business data source of the enterprise that generates the verification evidence, and may include at least one of the following: data source identifier, data source type, business system identifier, business department identifier, or data table identifier.

[0085] The verification evidence generation time is used to indicate the time when the corresponding verification evidence was formed, and can be determined according to the specific type of verification evidence. For example, for verification evidence of policy documents, the verification evidence generation time can be the policy document release time or the policy document revision time; for verification evidence of business system configuration, it can be the configuration modification time; for verification evidence of approval records, it can be the approval completion time; and for verification evidence of operation logs, it can be the time when the business operation occurred.

[0086] Subsequently, using the difference item identifier of the knowledge difference item as the verification association object, the verification evidence identifier, knowledge difference item identifier, verification evidence source, and verification evidence generation time are written into the corresponding evidence association record, thereby establishing the association relationship between the knowledge difference item, verification evidence, verification evidence source, and verification evidence generation time.

[0087] In one implementation, each evidence association record may further include information such as a summary of the verification evidence content, the scope of business application, the evidence acquisition time, and the evidence status. The summary of the verification evidence content characterizes the core content of the verification evidence used to support the corresponding knowledge difference item; the evidence acquisition time records the actual time the system acquired the verification evidence; and the evidence status records whether the verification evidence has passed validity verification.

[0088] Furthermore, after forming the evidence association record, the validity of each piece of verification evidence is verified, and the verification result is written into the evidence status field. The validity of the evidence is determined based at least on the reliability of the source of the verification evidence, the consistency between the time of the verification evidence generation and the target business time range, the data integrity, and the degree of business relevance with the knowledge difference item. The evidence status includes at least valid, invalid, and pending confirmation, which are used for subsequent overlay status records and effectiveness determination.

[0089] Through the above association processing, all verification evidence corresponding to any knowledge difference item can be retrieved based on the difference item identifier, and the source and generation time of each verification evidence can be obtained, providing an association basis for subsequent evidence validity verification and multi-evidence consistency verification based on knowledge difference items.

[0090] S304: Summarize the evidence association records corresponding to each knowledge difference item to form an evidence association set corresponding to the candidate update version; Specifically, firstly, the version identifier of the candidate update version is read, and the version identifier is associated with the difference item identifier of each knowledge difference item; then, the verification evidence association record corresponding to each knowledge difference item is added to the evidence association set corresponding to the candidate update version. The evidence association set includes at least: candidate update version identifier; knowledge difference item identifier; verification evidence identifier; verification evidence source; and verification evidence generation time.

[0091] Furthermore, the evidence association set may also include at least one of the following: a summary of the evidence content, the time of evidence acquisition, the scope of business application, and the status of the evidence.

[0092] In this embodiment, a one-to-one or one-to-many association relationship is established between the candidate update version and the evidence association set. One candidate update version can correspond to multiple knowledge difference items, and one knowledge difference item can correspond to multiple verification evidence from different sources. This forms a multi-level association structure of candidate update version, knowledge difference item, verification evidence, verification evidence source and generation time. In the subsequent evidence verification process, the corresponding evidence association set is first obtained according to the candidate update version identifier, and then the corresponding verification evidence is obtained according to the knowledge difference item identifier. Evidence validity verification and multi-evidence consistency verification are performed for each knowledge difference item. Based on the verification results of each knowledge difference item, it is further determined whether the candidate update version meets the preset effective conditions.

[0093] Through the above steps, an evidence association set is formed with "knowledge difference items" as the smallest verification object, "multi-business data source verification evidence" as the verification basis, and "evidence source and generation time" as evidence attributes. This enables subsequent version state transitions to be based on the verification results of specific knowledge difference items, rather than solely on the overall credibility of candidate update versions.

[0094] S400: Establish an evidence coverage matrix based on the evidence association set. The evidence coverage matrix uses the knowledge difference item as the difference item index and the verification evidence as the evidence index, and records the coverage status of each verification evidence for the corresponding knowledge difference item. In this embodiment, the evidence coverage matrix is ​​used to describe the correspondence between each knowledge difference item and each verification evidence in the candidate update version. The rows of the matrix correspond to different knowledge difference items, the columns of the matrix correspond to different verification evidence, and the intersection of the rows and columns is used to record whether the corresponding verification evidence establishes an evidence association with the corresponding knowledge difference item and whether it forms effective coverage.

[0095] Specifically, firstly, the evidence association set corresponding to the candidate update version is read, and knowledge difference item identifiers and verification evidence identifiers are extracted from the evidence association set; then, difference item indexes and evidence indexes are established respectively; based on the knowledge difference item identifiers and verification evidence identifiers recorded in each evidence association record, the correspondence between knowledge difference items and verification evidence is established at the corresponding matrix positions; further, combined with the evidence status of the verification evidence, the coverage status is recorded at the corresponding matrix positions.

[0096] Through the above processing, the relationship between the scattered knowledge difference items and the verification evidence is transformed into a matrix form, which enables subsequent judgment on the evidence coverage of individual knowledge difference items by row, and also enables analysis of the knowledge difference items covered by individual verification evidence by column, providing a structured basis for subsequent evidence validity verification and judgment of the conditions for the effectiveness of candidate update versions.

[0097] S401: Construct an evidence coverage matrix using knowledge difference items as difference item indices; Specifically, firstly, the evidence association set corresponding to the candidate update version is read, knowledge difference item identifiers are extracted from each evidence association record, and the extracted knowledge difference item identifiers are deduplicated to obtain the knowledge difference item set corresponding to the candidate update version. Subsequently, a difference item index for the evidence coverage matrix is ​​established based on the knowledge difference item set. Each knowledge difference item corresponds to a row in the evidence coverage matrix, and the difference item identifier of the knowledge difference item is used as the index value of the corresponding row.

[0098] Furthermore, knowledge difference items can be sorted according to their preset order in candidate update versions, difference type, or degree of business impact, and a corresponding row index can be built according to the sorting results.

[0099] For example, the candidate update version includes three knowledge differences, namely: D1: The threshold for reimbursement approval amount has been adjusted from 5,000 yuan to 3,000 yuan; D2: New approval criteria for department heads; D3: The deadline for submitting reimbursement applications has been changed from the end of the month to the beginning of the month.

[0100] Then D1, D2, and D3 can be used as the indices of the three differences in the evidence coverage matrix, corresponding to the three rows in the matrix.

[0101] In this embodiment, each knowledge difference item corresponds to an independent difference item identifier. The difference item identifier can be used to associate the matrix rows in the evidence coverage matrix with the specific knowledge difference items in the candidate update version.

[0102] By using knowledge difference items as difference item indexes, each knowledge difference item can have an independent verification position in the evidence coverage matrix. This allows for the determination of whether each knowledge difference item has obtained valid verification evidence, thus avoiding multiple knowledge difference items sharing the same overall verification result.

[0103] S402: Establish an evidence coverage matrix using verified evidence as an evidence index; Specifically, the verification evidence identifier is first extracted from the evidence association set, and the verification evidence identifier is deduplicated to obtain the verification evidence set corresponding to the candidate update version.

[0104] Subsequently, an evidence index is established for the evidence coverage matrix based on the set of verified evidence. Each piece of verified evidence corresponds to a column in the evidence coverage matrix, and the verification evidence identifier of the verified evidence is used as the index value of the corresponding column.

[0105] In this context, even if different verification evidence originates from the same internal business data source, they are indexed as independent verification evidence; verification evidence originating from different internal business data sources are indexed as independent evidence.

[0106] For example, for the knowledge difference item D1 in step S401, verification evidence E1 can be obtained from the institutional document database, verification evidence E2 can be obtained from the approval process database, and verification evidence E3 can be obtained from the actual reimbursement business records. Then, E1, E2, and E3 can be used as evidence indices in the evidence coverage matrix.

[0107] Furthermore, in addition to the verification evidence identifier corresponding to the evidence index, the source of the verification evidence and the time of the verification evidence generation can be associated with the record, so that each evidence index in the matrix can be further mapped to a specific business data source and time attribute.

[0108] Thus, the evidence coverage matrix forms a two-dimensional relational structure with knowledge difference items as rows and verification evidence as columns. For example: Among them, D1, D2, and D3 are knowledge difference indexes, and E1, E2, and E3 are verification evidence indexes.

[0109] By using verification evidence as an evidence index, multiple verification evidences from different sources corresponding to the same knowledge difference item can be represented separately in the evidence coverage matrix, providing a basis for subsequent judgment on whether a single knowledge difference item is supported by multiple sources of evidence and for performing consistency checks between different pieces of evidence.

[0110] S403: Record the coverage status based on the correspondence between verification evidence and knowledge discrepancies; Specifically, for each matrix position in the evidence coverage matrix, the corresponding knowledge difference item identifier and verification evidence identifier are read, and the evidence association set is searched to see if there is an evidence association record that contains both the knowledge difference item identifier and the verification evidence identifier.

[0111] When no corresponding evidence-related record exists, the coverage status of the corresponding matrix position is set to uncovered. When a corresponding evidence-related record exists, the evidence status of the verification evidence is further read and the coverage status is recorded: when the evidence status is valid, the coverage status of the corresponding matrix position is set to valid coverage; when the evidence status is invalid, the coverage status of the corresponding matrix position is set to invalid coverage; when the evidence status is pending confirmation, the coverage status of the corresponding matrix position is set to pending confirmation coverage.

[0112] In this embodiment, the number of valid coverages in the corresponding matrix row can be counted for each knowledge difference item, and the degree of evidence coverage for that knowledge difference item can be determined based on the number of valid coverages. For example, if there is at least one valid coverage position in the matrix row corresponding to a knowledge difference item, the knowledge difference item is marked as having evidence coverage; if there is no valid coverage position, the knowledge difference item is marked as having no valid evidence coverage.

[0113] Furthermore, for key knowledge discrepancies, higher evidence coverage requirements can be set. For example, it can be required that each key knowledge discrepancy has at least one valid verification piece of evidence, or that at least two verification pieces of evidence from different business data sources provide valid coverage. Knowledge discrepancies that do not meet the corresponding coverage requirements are marked as having insufficient evidence coverage.

[0114] Through the above processing, the evidence coverage matrix not only records whether there is a correlation between knowledge difference items and verification evidence, but also reflects the current actual coverage status of the verification evidence on the knowledge difference items. This provides a basis for subsequent evidence validity verification based on knowledge difference items and for judging whether the candidate update version meets the effective conditions based on the coverage results of all key knowledge difference items.

[0115] S404: Determine the evidence coverage of each knowledge difference item based on the evidence coverage matrix; Specifically, for each knowledge difference item, its corresponding matrix row is read, and the number of verification evidence items in the matrix row that are in a state of effective coverage and the number of corresponding verification evidence sources are counted.

[0116] When there is verification evidence in the matrix row corresponding to a knowledge difference item that meets the preset validity requirements, the knowledge difference item is determined to have valid evidence coverage; when there is no verification evidence in the matrix row that meets the preset validity requirements, the knowledge difference item is determined to have no valid evidence coverage.

[0117] Furthermore, for multiple verification evidences corresponding to the same knowledge difference item, it is also possible to determine whether the multiple verification evidences come from different internal business data sources of the enterprise based on their evidence source identifiers, and use the effective coverage of different sources as the multi-source coverage indicator of the knowledge difference item.

[0118] For example, if a knowledge difference item D1 corresponds to three verification evidences E1, E2, and E3, where E1 comes from the institutional document database, E2 comes from the approval process database, and E3 comes from the business system database, and all three verification evidences pass the validity check, then D1 has coverage by three valid evidences and is supported by evidence from three different business data sources.

[0119] Therefore, the evidence coverage matrix can refine the evidence verification process of candidate update versions from "version-level verification" to "knowledge difference item-level verification", enabling subsequent evidence validity and consistency judgments to be performed according to the coverage status of each knowledge difference item, and further controlling the state transition of candidate update versions based on the verification results of each knowledge difference item.

[0120] S500: For multiple verification evidences corresponding to the same knowledge difference item, generate corresponding evidence assertions based on the verification conclusions contained in each verification evidence, establish conflict assertion relationships based on the exclusion relationship between different evidence assertions, and determine the conflict state of the knowledge difference item based on the conflict assertion relationships. In this embodiment, for multiple verification evidences corresponding to the same knowledge difference item, the verification conclusions used to prove the knowledge difference item in each verification evidence are extracted and standardized to form multiple evidence assertions corresponding to the knowledge difference item; furthermore, conflict assertion relationships are established based on the exclusion relationship between the business conclusions expressed by different evidence assertions, and the conflict status of the corresponding knowledge difference item is determined based on the conflict assertion relationships to identify whether there are unresolved verification conclusion conflicts among multiple verification evidences.

[0121] S501: For the same knowledge difference item, extract the corresponding verification conclusions from multiple verification evidences associated with the knowledge difference item, and perform structured processing on the verification conclusions according to the preset assertion expression rules to generate the corresponding evidence assertions. Specifically, the corresponding verification conclusion extraction method is determined according to the type of difference in the knowledge difference item. For example: when the knowledge difference item is a change in parameter value, the parameter name, parameter value, and corresponding comparison relationship are extracted from the verification evidence; when the knowledge difference item is a change in business rule, the rule conditions, execution actions, and judgment results on whether the rule is valid are extracted from the verification evidence; when the knowledge difference item is a change in scope of application, the applicable object, business scenario, or organizational scope are extracted from the verification evidence; when the knowledge difference item is a change in time attribute, the effective time, expiration time, or time interval are extracted from the verification evidence.

[0122] The extracted verification conclusions are represented according to a unified assertion structure, enabling verification evidence from different sources to be converted into comparable evidentiary assertions. For example, a verification conclusion can be represented as: "The value of business field A within the scope of application B is C," or as: "When business condition A is met, action B should be executed." This provides a unified basis for comparison of verification conclusions from different business data sources.

[0123] S502: Determine the exclusion relationship between evidentiary assertions; Specifically, determining the exclusion relationship between evidential assertions is not simply a matter of comparing whether the texts of two evidential assertions are different. Instead, it involves using business semantic normalization, applicable condition matching, and mutual exclusion rule determination to determine whether two evidential assertions address the same verification condition and provide conclusions that they cannot both be true simultaneously. This includes the following steps: S502-1: For multiple verification evidences corresponding to the same knowledge difference item, extract attributes such as business object, business field, assertion value, comparison relationship, business scenario and effective time range from the evidence assertions corresponding to each verification evidence, and convert the extracted attributes into a unified structured assertion representation; Specifically, the evidence assertions in natural language form are converted into a unified structured representation. For example, "the threshold for the approved amount is 20,000 yuan" is converted into: Business target: Approval procedures; Business field: Approval amount threshold; Comparison relation: equal to; Assertion value: 20,000 yuan; Convert "Approval amount threshold is 10,000 yuan" to: Business target: Approval procedures; Business field: Approval amount threshold; Comparison relation: equal to; Assertion value: 10,000 yuan; By converting evidence assertions in natural language form into a unified structured representation, a comparable set of attributes is provided for subsequent exclusion relation determinations.

[0124] S502-2: For any two evidentiary assertions, match the business objects, business scenarios, and effective time ranges corresponding to the two evidentiary assertions to determine whether the two evidentiary assertions have the same or overlapping applicable conditions. Specifically, it can be determined whether two evidential assertions refer to the same business object by matching business object identifiers; whether two evidential assertions refer to the same business scenario by business scenario identifiers, business type identifiers, or business process node identifiers; and whether there is an overlap in the effective time range of two evidential assertions by comparing the effective start and end times.

[0125] When two evidentiary assertions have the same business object, the same business scenario, and an overlapping effective time range, the two evidentiary assertions are identified as a candidate conflicting assertion pair under the same applicable conditions.

[0126] S502-3: For two evidence assertions that are determined to be a candidate conflict assertion pair, further compare the business fields and attribute types corresponding to the two evidence assertions; Specifically, when two evidentiary assertions correspond to the same business field, and the attribute type of the business field is numeric, enumerated, stateful, boolean, or other field types with mutually exclusive value relationships, it is determined that the two evidentiary assertions describe the same business attribute.

[0127] For example, for the numerical business field "approval amount threshold", "approval amount threshold is 20,000 yuan" and "approval amount threshold is 10,000 yuan" are identified as two evidentiary assertions for the same business field.

[0128] S502-4: For candidate conflict assertion pairs with consistent business fields, determine whether the two evidence assertions can be true simultaneously based on the comparison relationship between the two evidence assertions and the logical relationship between the assertion values. Specifically, assertion relationship determination rules for different business field types can be established in advance, and logical relationship determination can be performed on two evidence assertions according to the assertion relationship determination rules.

[0129] For example: When two assertions both use the "equal to" relationship and their values ​​are different, the two assertions are determined to be mutually exclusive. When one assertion is "equal to 20,000 yuan" and the other assertion is "equal to 10,000 yuan", the two assertions cannot be true at the same time. When one assertion is "approval amount threshold is greater than 10,000 yuan" and the other assertion is "approval amount threshold is less than or equal to 5,000 yuan", the value ranges of the two assertions do not overlap, thus determining that the two assertions have an exclusive relationship. When the value ranges of the two assertions overlap, they are not directly determined to be exclusive, but are determined to be non-exclusive or pending further verification.

[0130] Therefore, it is possible to avoid judging conflicts solely based on textual differences, and instead determine whether evidence assertions are mutually exclusive based on the logical relationships between business fields, comparison relationships, and assertion values.

[0131] S503: Associate the evidence assertions that have an exclusion relationship to form a conflicting assertion relationship for the knowledge difference item; Specifically, using evidence assertions as assertion nodes and the exclusion relationship between two evidence assertions as association edges, evidence assertions with exclusion relationships are connected to form conflict assertion relationships for corresponding knowledge difference items.

[0132] For example, for three evidence assertions A, B, and C corresponding to the same knowledge difference item, if there is an exclusion relationship between evidence assertion A and evidence assertion B, and an exclusion relationship between evidence assertion B and evidence assertion C, then conflict assertion relationships are established between A and B, and between B and C, respectively.

[0133] By establishing the conflict assertion relationship, mutually exclusive verification conclusions that were originally scattered in business data sources within different enterprises can be linked, enabling the subsequent verification process to handle specific knowledge differences that conflict, without having to re-verify the entire candidate update version without discrimination.

[0134] S504: Determine the conflict state of the knowledge difference item based on the conflict assertion relation corresponding to the knowledge difference item; Specifically, when there is no exclusion relationship among multiple evidence assertions corresponding to the same knowledge difference item, the knowledge difference item is determined to be in a conflict-free state; when there is an exclusion relationship between at least two evidence assertions, and the exclusion relationship has not been eliminated by subsequent evidence verification, source priority judgment or scope of application judgment, the knowledge difference item is determined to be in a state of unresolved conflict.

[0135] Furthermore, for knowledge differences involving conflicting assertions, the conflicting relationships can be further labeled based on information such as the source, generation time, scope of application, and validity of the verification evidence corresponding to the conflicting assertions. For example, conflict states can be categorized as no conflict, pending conflict, and resolved conflict.

[0136] The conflict-free state indicates that there is no mutual exclusion relationship between the verification conclusions formed by multiple verification evidences; the conflict-pending state indicates that different verification evidences form mutually exclusive verification conclusions, and it is not yet possible to determine whether any of the verification conclusions has priority; the conflict-resolved state indicates that although there were mutually exclusive evidence assertions, after further verification, it has been determined that one of the evidence assertions is valid, or it has been determined that different evidence assertions are applicable to different business scopes or effective time periods.

[0137] Through the above processing, the conflict of verification conclusions between multiple pieces of verification evidence is transformed into a conflict state for specific knowledge difference items. This allows the existence of unresolved knowledge difference items to be used as one of the judgment conditions for restricting the effectiveness of candidate update versions during the subsequent determination of the effectiveness of candidate update versions.

[0138] S600: Determine whether each key knowledge difference item has valid evidence coverage based on the evidence coverage matrix, and determine whether there are unresolved conflicts for each key knowledge difference item based on the conflict assertion relationship. In this embodiment, for each key knowledge difference item in the candidate update version, the effective evidence coverage status of each key knowledge difference item is determined according to the evidence coverage matrix, and the conflict status of each key knowledge difference item is determined according to the conflict assertion relationship, thereby forming the verification judgment result corresponding to each key knowledge difference item, providing a basis for subsequently determining whether the candidate update version meets the effective conditions.

[0139] S601: Determine the coverage status of key knowledge discrepancies based on the evidence coverage matrix; Specifically, for each key knowledge difference item in the candidate update version, each verification evidence associated with the corresponding key knowledge difference item is obtained according to the evidence coverage matrix, and the coverage status of each verification evidence for the key knowledge difference item is read.

[0140] The coverage status is used to characterize whether the corresponding verification evidence can provide effective verification of the key knowledge difference item, including effective coverage status, invalid coverage status, coverage status to be confirmed, and uncovered status.

[0141] By summarizing multiple pieces of verification evidence corresponding to the same key knowledge difference item, the number of valid verification evidence items and the sources of valid verification evidence items corresponding to the key knowledge difference item are determined.

[0142] S602: Determine effective evidence coverage for key knowledge discrepancies; Specifically, for each key knowledge difference item, the preset evidence coverage conditions are determined based on the number of valid verification evidence items and the source of the valid verification evidence. When the number of valid verification evidence items corresponding to a key knowledge difference item reaches the preset coverage quantity, and the valid verification evidence meets the preset business data source conditions, the key knowledge difference item is determined to have valid evidence coverage; otherwise, the key knowledge difference item is determined to not have valid evidence coverage.

[0143] The preset business data source conditions can be used to limit whether valid verification evidence comes from different internal business data sources of the enterprise, so as to avoid duplicate business records generated by the same business data source being repeatedly included in the evidence coverage result.

[0144] Furthermore, the effective evidence coverage results for all key knowledge discrepancies are summarized to determine whether there is at least one key knowledge discrepancy that does not have effective evidence coverage.

[0145] S603: Determine the conflict status of key knowledge discrepancies based on conflict assertion relationships; Specifically, for each key knowledge difference item, the exclusion relationship between the evidence assertions associated with the key knowledge difference item is obtained based on the conflict assertion relationship, and it is determined whether the conflict corresponding to the exclusion relationship has been resolved. When there is no conflict assertion relationship between multiple evidence assertions corresponding to the same key knowledge difference item, the key knowledge difference item is determined to be without conflict. When there is a conflict assertion relationship, the conflict resolution result corresponding to the conflict assertion relationship is further obtained.

[0146] When at least one of the mutually exclusive evidentiary assertions no longer holds due to the invalidation of the corresponding verification evidence, its exceeding the scope of business application, or its negation by subsequent verification evidence, the corresponding conflict assertion relationship is determined to be resolved, and the key knowledge difference item is determined to have no unresolved conflict.

[0147] When mutually exclusive evidentiary assertions remain valid and there is no verification result that can eliminate the exclusion relationship, the key knowledge difference item is identified as having an unresolved conflict.

[0148] Furthermore, the conflict status of all key knowledge discrepancies is summarized to determine whether there is at least one key knowledge discrepancy with unresolved conflict.

[0149] S604: Verification and judgment results for key knowledge discrepancies; Specifically, for each key knowledge difference item, the effective evidence coverage status is associated with the conflict status to form a corresponding verification judgment result. When a key knowledge difference item has effective evidence coverage and no unresolved conflicts, the verification judgment result of the key knowledge difference item is determined to be verified as passed; when the key knowledge difference item does not have effective evidence coverage, it is determined to be insufficient evidence coverage; when the key knowledge difference item has unresolved conflicts, it is determined to be unresolved conflicts.

[0150] Furthermore, the verification results corresponding to all key knowledge discrepancies are summarized: When all key knowledge differences are covered by valid evidence and there are no unresolved conflicts for any of the key knowledge differences, the candidate update version is determined to meet the preset validity criteria; when at least one key knowledge difference is not covered by valid evidence, or at least one key knowledge difference has unresolved conflicts, the candidate update version is determined not to meet the validity criteria.

[0151] Therefore, the "sufficiency of evidence" reflected by the evidence coverage matrix and the "existence of unresolved contradictions among the evidence" reflected by the conflict assertion relationship are jointly determined to avoid making the candidate update version effective based on a single piece of verification evidence, and to provide a clear basis for the state transition of the candidate update version in the subsequent S700.

[0152] S700: When all key knowledge discrepancies have valid evidence coverage and there are no unresolved conflicts, the candidate update version is changed from the pending verification state to the effective state, and the candidate update version is determined as the new current effective knowledge version of the target knowledge item. At the same time, the original current effective knowledge version is converted into a historical knowledge version, and a version update record containing version identifier, state transition information and evidence association information is established. In this embodiment, when it is determined that all key knowledge differences have valid evidence coverage and there are no unresolved conflicts, the candidate update version is determined to meet the preset effective conditions, and the state transition of the candidate update version and the switching of the current effective version of the target knowledge item are performed based on the effective conditions. At the same time, the original current effective knowledge version is processed as history and a corresponding version update record is established.

[0153] S701: Determine the conditions for the candidate update version to take effect; Specifically, the verification results corresponding to each key knowledge difference item formed in S604 are obtained, and it is determined whether all key knowledge difference items meet the condition of "having valid evidence coverage and having no unresolved conflicts". When all key knowledge difference items have valid evidence coverage and there are no unresolved conflicts for each key knowledge difference item, the candidate update version is determined to meet the preset activation conditions; otherwise, the activation process of the candidate update version is not executed.

[0154] Using the above method, the key knowledge difference items are taken as the smallest verification object, and the verification results of each knowledge difference item are summarized into the effective determination result of the candidate update version.

[0155] S702: Change the effective status of candidate update versions; Specifically, when it is determined that the candidate update version meets the preset effective conditions, the version identifier and current version status corresponding to the candidate update version are obtained, the version status of the candidate update version is changed from the pending verification status to the effective status, and the effective time of the candidate update version is recorded.

[0156] Before performing the state transition, the candidate update version can be verified again to ensure it is still in the pending verification state, so as to avoid the version that has already completed the state transition being executed repeatedly.

[0157] After the state transition is complete, the candidate update version is marked as the current valid version that can be accessed by the enterprise's internal knowledge base.

[0158] S703: Switch the current valid knowledge version of the target knowledge item; Specifically, after converting the candidate update version to an effective state, the candidate update version is associated with the corresponding target knowledge item as a current effective version, and the candidate update version is determined as the new current effective knowledge version of the target knowledge item.

[0159] At the same time, the association between the target knowledge item and the original current valid knowledge version is removed, so that subsequent knowledge queries, business rule calls or knowledge matching operations for the target knowledge item will prioritize reading the candidate updated version.

[0160] This enables the target knowledge item to switch from its current valid knowledge version to a candidate updated version.

[0161] S704: Transform the historical state of the current valid knowledge version; Specifically, the original currently valid knowledge version that was associated with the target knowledge item before the candidate update version took effect is obtained, and the version status of the original currently valid knowledge version is changed from the effective state to the historical state.

[0162] At the same time, the version identifier, knowledge content, historical effective time, and associated verification evidence corresponding to the original currently valid knowledge version are retained to form a traceable historical version.

[0163] The historical knowledge version is no longer considered the current valid knowledge version of the target knowledge item, but it is still retained in the version record of the enterprise's internal knowledge base for subsequent version queries, version backtracking, and update process tracking.

[0164] S705: Establish version update records; Specifically, a version update record is established based on the effective process of the candidate update version, and the version update record is associated with the target knowledge item. The version update record includes at least: The version identifier of the currently valid knowledge version; Version identifier for candidate update versions; The status transition information of the candidate update version from the pending verification state to the effective state; The effective time corresponding to the state transition; Evidence association information corresponding to the knowledge difference items; Verification and judgment results for each key knowledge difference item.

[0165] The evidence association information is used to record the verification evidence and its source that makes the corresponding knowledge difference item meet the valid evidence coverage conditions, so that the effective result of the candidate update version can be traced back to the specific knowledge difference item and the corresponding internal business data evidence of the enterprise.

[0166] S706: Establishes a valid version association after a version switch; Specifically, after the candidate update version takes effect, the current valid version of the target knowledge item is switched, and the original current valid knowledge version is historically processed, the association between the target knowledge item, the candidate update version, and the version update record is established.

[0167] Specifically, the target knowledge item is associated with the candidate updated version as a current valid relationship, the target knowledge item is associated with the original current valid knowledge version as a historical version relationship, and the two versions are associated with the corresponding version update records.

[0168] This leads to: The target knowledge item, the new current valid knowledge version, the version update record, the verification evidence, and the version tracing relationship of the original historical knowledge version enable the knowledge base not only to determine the current valid knowledge content, but also to trace the formation process of the current version and its evidence based on the version update record.

[0169] S800: When at least one key knowledge difference item does not have valid evidence coverage, or at least one key knowledge difference item has unresolved conflicts, the candidate update version is kept in a pending verification state, and the current valid knowledge version remains unchanged. In this embodiment, when it is determined from the verification result of S600 that at least one key knowledge difference item does not have valid evidence coverage, or at least one key knowledge difference item has unresolved conflicts, it is determined that the candidate update version does not meet the preset effective conditions, and the candidate update version is kept in the state of pending verification, while the current valid knowledge version of the target knowledge item is not switched.

[0170] S801: The candidate update version does not meet the conditions for taking effect; Specifically, the evidence coverage status and conflict status corresponding to each key knowledge difference item are obtained, and it is determined whether there is at least one key knowledge difference item that does not have effective evidence coverage, or whether there is at least one key knowledge difference item that has unresolved conflict.

[0171] When any of the following conditions are met: At least one key knowledge discrepancy item lacks effective evidence coverage; There is at least one key knowledge discrepancy item with an unresolved conflict; The candidate update version will then be determined as not meeting the preset activation conditions.

[0172] In this process, the knowledge discrepancies that cause the candidate update version to fail to take effect can be further identified, and the corresponding knowledge discrepancies can be marked as pending further verification or pending conflict resolution.

[0173] S802: Maintain the pending verification status of candidate update versions; Specifically, after determining that the candidate update version does not meet the conditions for effectiveness, the current version status of the candidate update version is obtained, and it is determined whether the current version status is a pending verification status. When the candidate update version is in a pending verification status, the version identifier and version content of the candidate update version remain unchanged, and the state transition from the pending verification status to the effective status is not performed.

[0174] Simultaneously, the verification results that cause the candidate update version to fail to meet the effectiveness conditions are associated with the candidate update version to form a corresponding reason to be verified. In one embodiment, when the candidate update version meets the preset termination conditions, the candidate update version is changed from the pending verification state to the rejection state, and the rejection reason and the corresponding knowledge difference item identifier are recorded. The candidate update version in the rejection state will no longer participate in the subsequent new evidence-driven effectiveness determination process.

[0175] For example, when knowledge difference item A has valid evidence coverage but there are two mutually exclusive and valid evidence assertions, "knowledge difference item A has unresolved conflicts" can be used as the reason for verification of the candidate update version; when knowledge difference item B does not have valid verification evidence that meets the coverage conditions, "knowledge difference item B has insufficient valid evidence coverage" can be used as the reason for verification.

[0176] S803: Maintain the current effective knowledge version unchanged; Specifically, while the candidate update version remains in the pending verification state, the current valid version association between the target knowledge item and the original current valid knowledge version is maintained. The current valid version association between the candidate update version and the target knowledge item is not established, and the current valid version association between the original current valid knowledge version and the target knowledge item is not terminated. The original current valid knowledge version continues to be used as the valid version for the target knowledge item when providing knowledge query, business rule call or knowledge matching.

[0177] Therefore, in the absence of a candidate update version that has not yet been verified, we should avoid replacing already effective knowledge content with unverified knowledge content.

[0178] S804: Record the reason why the candidate update version is pending verification; Specifically, based on the key knowledge differences that do not meet the conditions for effectiveness and their verification results, a pending verification status record corresponding to the candidate update version is established or updated. The pending verification status record includes at least the version identifier of the candidate update version, the reason for pending verification, the corresponding knowledge difference identifier, and the corresponding evidence coverage status or conflict status.

[0179] When the same candidate update version has both insufficient evidence coverage and unresolved conflicts, the corresponding knowledge difference items and their reasons for failure can be recorded separately so that the knowledge difference items that need to be re-verified can be determined after new verification evidence arrives.

[0180] S805: Maintain the parallel state of candidate update versions and the current effective knowledge version; Specifically, while the candidate update version is in the pending verification state, the candidate update version is kept in the enterprise's internal knowledge base as a pending verification version, while the original currently valid knowledge version continues to be the current valid version of the target knowledge item.

[0181] The candidate updated version is used to receive new verification evidence and perform re-verification of the corresponding knowledge difference items; the currently valid knowledge version continues to handle knowledge invocation in the current business scenario.

[0182] Therefore, the same target knowledge item can exist in two different states at the same time: the currently valid knowledge version and the candidate update version to be verified, without changing the knowledge version currently used by the business because the candidate update version has not yet been verified.

[0183] S900: Monitor the enterprise's internal business data source while the candidate update version is in the pending verification state. When a new business record is obtained, associate and match the new business record with the knowledge difference item in the candidate update version in the pending verification state to determine whether the new business record constitutes new verification evidence corresponding to at least one knowledge difference item. In this embodiment, while the candidate update version is in the pending verification state, the internal business data source corresponding to the candidate update version is continuously monitored to obtain the new business records generated after the candidate update version is generated. The new business records are then associated and matched with the knowledge difference items in the candidate update version to determine whether the new business records can serve as new verification evidence for the corresponding knowledge difference items.

[0184] Specifically, the following steps are included: S901: Monitor internal business data sources of the enterprise; Specifically, while the candidate update version is in the pending verification state, a continuous monitoring mechanism is established for the enterprise's internal business data source corresponding to the candidate update version, and new business records generated in the enterprise's internal business data source are obtained according to a preset monitoring cycle or a new record notification mechanism of the business data source.

[0185] The internal business data sources may include business approval systems, order management systems, financial management systems, production management systems, customer management systems, access control systems, and internal business databases. For different business data sources, new business records can be obtained using methods such as periodic queries, incremental reads, event notifications, or log monitoring, depending on their data update methods.

[0186] To avoid using existing business records from before the candidate update version was generated as verification evidence, the generation time of the candidate update version can be recorded, and the new business record can be determined as a new business record after the candidate update version was generated based on the generation time, entry time, or business processing time of the new business record.

[0187] S902: Retrieve new business records; Specifically, business data that meets the conditions for adding new records is obtained from the monitored internal business data sources of the enterprise, and the business object, business field, field value, business scenario, business processing time, business status and data source identifier corresponding to the new business record are extracted.

[0188] For example, when the knowledge difference item in the candidate update version involves "the approval amount threshold is adjusted from 10,000 yuan to 20,000 yuan", the approval business record newly added after the candidate update version is generated is obtained from the approval system, and the fields such as approval amount, approval type, approval object, approval time, approval result and approval system identifier in the approval business record are extracted.

[0189] By extracting business fields from newly added business records, these records can be converted into data representations that are identical or comparable to the knowledge difference items, providing input for subsequent association matching.

[0190] S903: Establish a matching association between newly added business records and knowledge difference items; Specifically, for each newly acquired business record, the business fields, business objects, business scenarios, time attributes, and business types in the newly acquired business record are associated and matched with each knowledge difference item in the candidate update version that is in the pending verification state.

[0191] Specifically, matching conditions can be determined in advance based on each knowledge difference item. The matching conditions include at least the business fields involved in the knowledge difference item and the business scope of application, and further include one or more of the following: business object, business scenario, time range, or business type.

[0192] The corresponding fields in the newly added business record are compared with the matching conditions. When the newly added business record meets both the corresponding business field matching conditions and the business scope matching conditions, an association is established between the newly added business record and the corresponding knowledge difference item.

[0193] For example, for the knowledge difference item "the approval amount threshold is adjusted from 10,000 yuan to 20,000 yuan", "approval amount" can be identified as the target business field, and the corresponding approval business can be identified as the business scope. When a new business record belongs to the approval business and contains the approval amount field, the new business record can be associated with the above knowledge difference item.

[0194] S904: Identify new verification evidence; Specifically, after a new business record is associated with at least one knowledge difference item, the system determines whether the new business record can form a verification conclusion for the corresponding knowledge difference item based on the business field values, business status, and business processing results in the new business record.

[0195] When the newly added business record contains business data that can be directly used to verify the knowledge difference item, and the business data meets the preset evidence formation conditions, the newly added business record is determined as new verification evidence associated with the corresponding knowledge difference item.

[0196] For example, regarding the knowledge difference item "the approval amount threshold is adjusted from 10,000 yuan to 20,000 yuan", when there is a business record in the newly added approval business record with an approval amount of 15,000 yuan and the approval record has been completed according to the new threshold, a verification conclusion can be formed based on the newly added business record that "the business record with an approval amount of 15,000 yuan has been approved according to the 20,000 yuan threshold rule", and the newly added business record can be identified as the new verification evidence for the corresponding knowledge difference item.

[0197] If a newly added business record only contains business fields related to the knowledge difference item, but lacks the business status, processing result, or other necessary fields required to form a verification conclusion, then the newly added business record will not be determined as valid new verification evidence. Instead, its association with the corresponding knowledge difference item will be retained, and subsequent business records will be used to supplement and form complete verification evidence.

[0198] S905: Record the associated information of newly added verification evidence; Specifically, when it is determined that a newly added business record constitutes new verification evidence, the knowledge difference item identifier, business data source identifier, new business record identifier, verification evidence generation time, and verification conclusion formed by the newly added business record are recorded, and the new verification evidence is added to the evidence association set of the corresponding candidate update version.

[0199] Therefore, while the candidate update version is in the pending verification state, the existing evidence can be dynamically supplemented by the internal business data continuously generated after the candidate update version is generated, without the need to regenerate the candidate update version. This allows subsequent steps to update the evidence coverage status only for the corresponding knowledge difference items based on the newly added verification evidence and re-execute the verification.

[0200] S1000: When the newly added business record constitutes new verification evidence, only the coverage status of the knowledge difference item corresponding to the new verification evidence in the evidence coverage matrix is ​​updated, and only the evidence validity check and conflict assertion relationship check are re-executed for the knowledge difference item, thereby determining whether the candidate update version containing the knowledge difference item meets the effective conditions, and performing the state transition of the candidate update version according to the effective conditions. In this embodiment, when the newly added business record is determined to be new verification evidence corresponding to at least one knowledge difference item, the new verification evidence is associated with the corresponding knowledge difference item, and based on the verification result of the new verification evidence, only the corresponding coverage status of the knowledge difference item in the evidence coverage matrix is ​​updated; on this basis, only for the knowledge difference item affected by the new verification evidence, the evidence validity verification and conflict assertion relationship verification are re-executed, and based on the re-verification result, it is determined whether the candidate update version containing the knowledge difference item meets the effective conditions, thereby performing the state transition of the candidate update version.

[0201] Specifically, the following steps are included: S1001: Locate the knowledge difference item corresponding to the newly added verification evidence; Specifically, when it is determined that the newly added business record constitutes new verification evidence, the knowledge difference item corresponding to the new verification evidence is determined according to the knowledge difference item identifier recorded in the new verification evidence, and the association between the new verification evidence and the knowledge difference item is established.

[0202] The knowledge difference item identifier can be generated during the association matching process in S903 and transmitted along with the newly added verification evidence, so that the newly added verification evidence can directly locate the corresponding knowledge difference item in the candidate update version without having to re-traverse all the knowledge difference items in the candidate update version.

[0203] For example, for the knowledge difference item "the approval amount threshold is adjusted from 10,000 yuan to 20,000 yuan", when the newly added approval business record is determined by S904 to be the newly added verification evidence corresponding to the knowledge difference item, the knowledge difference item that needs to be re-verified is directly determined to be the knowledge difference item corresponding to "approval amount threshold adjustment" based on the knowledge difference item identifier associated with the newly added verification evidence.

[0204] S1002: Update the coverage status of the corresponding knowledge difference item; Specifically, based on the verification result corresponding to the newly added verification evidence, it is determined whether the newly added verification evidence can form effective evidence coverage of the corresponding knowledge difference item, and only the coverage status of the knowledge difference item in the evidence coverage matrix is ​​updated. The newly added verification evidence is added to the evidence coverage matrix as a new evidence index, and the coverage status of the newly added verification evidence is recorded at the matrix position corresponding to the knowledge difference item.

[0205] The coverage status can be used to indicate whether the newly added verification evidence can effectively support the corresponding knowledge difference item, such as being set to "effective coverage", "invalid coverage" or "pending confirmation" status.

[0206] During the update process, the coverage status of other knowledge difference items that are not related to the newly added verification evidence is not modified, thereby avoiding the need to recalculate the entire evidence coverage matrix due to the arrival of a single new business record.

[0207] S1003: Re-execute the evidence validity verification for the corresponding knowledge difference item; Specifically, for the knowledge difference item corresponding to the newly added verification evidence, the existing verification evidence and the newly added verification evidence associated with the knowledge difference item are obtained, and the evidence validity verification is re-executed. The evidence validity verification can be judged based on the reliability of the source of the verification evidence, the generation time, the business status, the data integrity, and the degree of business association between the verification evidence and the knowledge difference item.

[0208] For example, regarding the knowledge difference item "the approval amount threshold is adjusted from 10,000 yuan to 20,000 yuan", when the newly added approval record comes from an approval system with a valid business data source identifier, and the approval record has complete fields for approval amount, approval type, approval time, and approval result, it can be determined that the newly added verification evidence meets the corresponding evidence validity conditions.

[0209] By using the above method, only the knowledge difference items corresponding to the newly added verification evidence are re-verified, while other knowledge difference items that are not affected by the newly added verification evidence are not re-verified.

[0210] S1004: Re-execute the conflict assertion relation check for the corresponding knowledge difference item; Specifically, for knowledge discrepancies after evidence validity verification, corresponding evidence assertions are generated or updated based on newly added verification evidence and existing verification evidence corresponding to the knowledge discrepancy. The newly added evidence assertions are compared with the existing evidence assertions of the knowledge discrepancy. When the newly added evidence assertions and existing evidence assertions meet a preset exclusion relationship determination condition, the conflict assertion relationship corresponding to the knowledge discrepancy is updated, and the knowledge discrepancy is marked as conflicting. When there is no exclusion relationship between the newly added evidence assertions and existing evidence assertions, or when the original conflict can be resolved by adding verification evidence, the corresponding conflict state is updated.

[0211] For example, if existing verification evidence forms an evidentiary assertion that "the threshold for the approval amount is 20,000 yuan," and newly added verification evidence forms an evidentiary assertion that "the threshold for the approval amount is 20,000 yuan," then there is no exclusion relationship between the two evidentiary assertions. However, if the newly added verification evidence forms an evidentiary assertion that "the threshold for the approval amount is 10,000 yuan," and its business object, business scenario, and effective time range are the same as the previous evidentiary assertion, then the corresponding conflict assertion relationship is re-established, and this knowledge difference item is identified as an unresolved conflict.

[0212] S1005: Determine whether the candidate update version meets the conditions for taking effect; Specifically, based on the evidence coverage status and conflict status of the knowledge difference item, the verification result of the candidate update version containing the knowledge difference item is updated. First, it is determined whether the knowledge difference item has valid evidence coverage and whether there are unresolved conflicts. On this basis, combined with the existing verification results of other key knowledge difference items in the candidate update version, it is determined whether the candidate update version meets the preset effective conditions.

[0213] When all key knowledge differences are covered by valid evidence and there are no unresolved conflicts for any of the key knowledge differences, the candidate update version is determined to meet the conditions for effectiveness; when at least one key knowledge difference is not covered by valid evidence, or at least one key knowledge difference has unresolved conflicts, the candidate update version is determined not to meet the conditions for effectiveness.

[0214] S1006: Perform state transition for candidate update versions based on the effective conditions; Specifically, when it is determined that the candidate update version meets the activation conditions, the candidate update version is changed from the pending verification state to the activated state, and the candidate update version is determined as the new current valid knowledge version of the corresponding target knowledge item. At the same time, the original current valid knowledge version is converted to a historical knowledge version, and the corresponding version update record is updated. When it is determined that the candidate update version still does not meet the activation conditions, the candidate update version remains in the pending verification state, and the original current valid knowledge version remains unchanged, so that subsequent newly added business records can still continue to participate in the dynamic verification of the candidate update version as verification evidence.

[0215] Therefore, by performing evidence coverage status updates, evidence validity checks, and conflict assertion relationship checks only on the knowledge difference items corresponding to newly added verification evidence, the verification process of candidate update versions is transformed from repetitive verification of the entire version to local incremental verification of the affected knowledge difference items. While maintaining the stability of the current effective knowledge version, the verification evidence is gradually supplemented by subsequent new business records, and the candidate update version is driven to meet the effective conditions.

[0216] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0217] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for updating enterprise knowledge based on knowledge version management and dynamic evidence verification, characterized in that: include: The process involves acquiring knowledge content generated within the enterprise that needs to be updated, comparing the knowledge content to be updated with the current valid knowledge version in the enterprise's internal knowledge base, and determining at least one target knowledge item and at least one knowledge difference item corresponding to each target knowledge item based on the comparison results. Based on the current valid knowledge version of the target knowledge item and the knowledge difference item, a candidate update version is generated, and an independent version identifier is assigned to the candidate update version. The candidate update version is set to a pending verification state. During the period when the candidate update version is in the pending verification state, the current valid knowledge version is maintained as the current valid version of the target knowledge item. For each knowledge difference item in the candidate update version, according to the difference type and business scope of the knowledge difference item, corresponding verification evidence is obtained from at least two internal business data sources of the enterprise, and each verification evidence is associated with at least the corresponding knowledge difference item, the source of the verification evidence, and the generation time of the verification evidence to form an evidence association set corresponding to the candidate update version. An evidence coverage matrix is ​​established based on the evidence association set. The evidence coverage matrix uses the knowledge difference item as the difference item index and the verification evidence as the evidence index, and records the coverage status of each verification evidence for the corresponding knowledge difference item. For multiple verification evidences corresponding to the same knowledge difference item, corresponding evidence assertions are generated based on the verification conclusions contained in each verification evidence, and conflict assertion relationships are established based on the exclusion relationship between different evidence assertions. The conflict state of the knowledge difference item is determined based on the conflict assertion relationships. Based on the evidence coverage matrix, determine whether each key knowledge difference item has valid evidence coverage, and based on the conflict assertion relationship, determine whether there are unresolved conflicts for each key knowledge difference item. When all key knowledge discrepancies are covered by valid evidence and there are no unresolved conflicts, the candidate update version is changed from the pending verification state to the effective state, and the candidate update version is determined as the new current effective knowledge version of the target knowledge item. At the same time, the original current effective knowledge version is converted into a historical knowledge version, and a version update record containing version identifier, state transition information and evidence association information is established. When at least one key knowledge difference item does not have valid evidence coverage, or at least one key knowledge difference item has unresolved conflicts, the candidate update version is kept in a pending verification state, and the current valid knowledge version remains unchanged. During the period when the candidate update version is in the pending verification state, the internal business data source of the enterprise is monitored. When a new business record is obtained, the new business record is associated and matched with the knowledge difference items in the candidate update version in the pending verification state to determine whether the new business record constitutes new verification evidence corresponding to at least one knowledge difference item. When the newly added business record constitutes new verification evidence, only the coverage status of the knowledge difference item corresponding to the new verification evidence in the evidence coverage matrix is ​​updated, and only the evidence validity check and conflict assertion relationship check are re-executed for the knowledge difference item, thereby determining whether the candidate update version containing the knowledge difference item meets the effective conditions, and performing the state transition of the candidate update version according to the effective conditions.

2. The method as described in claim 1, characterized in that, The knowledge difference item is the smallest knowledge change unit that can be independently verified for evidence, and at least the difference item identifier, target knowledge item identifier, parent version identifier, content to be updated identifier, difference type, original value, updated value, business object, business field, scope of application, time attribute, and difference location are recorded.

3. The method as described in claim 1, characterized in that, The candidate update version records at least: candidate version identifier; target knowledge item identifier; parent version identifier; candidate version content; set of knowledge difference item identifiers; creation time; creation subject; and current status, wherein the parent version identifier corresponds to the current valid knowledge version identifier of the target knowledge item.

4. The method as described in claim 1, characterized in that, The enterprise's internal business data sources may include at least two of the following: policy document database, business system database, approval process database, operation log database, work order database, financial system database, enterprise announcement database, and supply chain management database.

5. The method as described in claim 1, characterized in that, The version status of a candidate update version includes at least the following: pending verification, effective, historical, and rejected. Specifically: pending verification indicates that the candidate update version has been generated but has not yet met the preset effective conditions and is awaiting supplementary verification evidence or is undergoing evidence verification; effective indicates that the candidate update version has met the preset effective conditions and can be used as a valid knowledge version; historical indicates that the knowledge version was once in an effective state but has been replaced by a new effective version; and rejected indicates that the candidate update version has been determined not to be effective under preset termination conditions and its verification process has been terminated.

6. The method as described in claim 1, characterized in that, The evidence association set should include at least: candidate update version identifier; knowledge difference item identifier; verification evidence identifier; verification evidence source; and verification evidence generation time.

7. The method as described in claim 1, characterized in that, The evidence coverage matrix is ​​used to describe the correspondence between each knowledge difference item and each verification evidence in the candidate update version. The rows of the matrix correspond to different knowledge differences items, the columns of the matrix correspond to different verification evidence, and the intersection of the rows and columns is used to record whether the corresponding verification evidence establishes an evidence association with the corresponding knowledge difference item and whether it forms effective coverage.

8. The method as described in claim 1, characterized in that, For multiple verification evidences corresponding to the same knowledge difference item, the process involves generating corresponding evidence assertions based on the verification conclusions contained in each verification evidence, establishing conflict assertion relationships based on the exclusion relationships between different evidence assertions, and determining the conflict state of the knowledge difference item based on the conflict assertion relationships, including: For the same knowledge difference item, the corresponding verification conclusions are extracted from multiple verification evidences associated with the knowledge difference item, and the verification conclusions are structured according to the preset assertion expression rules to generate the corresponding evidence assertions. Establish the exclusionary relationship between evidentiary assertions; Associating evidence assertions that are mutually exclusive to form conflicting assertion relationships for the knowledge differences mentioned above; The conflict state of the knowledge difference item is determined based on the conflict assertion relation corresponding to the knowledge difference item.

9. The method as described in claim 8, characterized in that, The exclusionary relationship between the assertions of evidence includes: For multiple verification evidences corresponding to the same knowledge difference item, attributes such as business object, business field, assertion value, comparison relationship, business scenario and effective time range are extracted from the evidence assertions corresponding to each verification evidence, and the extracted attributes are converted into a unified structured assertion representation. For any two evidentiary assertions, the business objects, business scenarios, and effective time ranges corresponding to the two evidentiary assertions are matched to determine whether the two evidentiary assertions have the same or overlapping applicable conditions. For the two evidence assertions identified as a candidate conflict assertion pair, further compare the business fields and attribute types corresponding to the two evidence assertions; For candidate conflict assertion pairs with consistent business fields, determine whether the two evidence assertions can be true simultaneously based on the comparison relationship between the two evidence assertions and the logical relationship between the assertion values.

Images