An ontology model enhancement-based data mining method and system

Abstract

The application discloses a data mining method and system based on ontology model enhancement, relates to the field of data mining, and comprises the following steps: acquiring multi-source heterogeneous data, analyzing data structure, generating data assets and corresponding resource identifiers; mapping the data assets with an ontology model to generate an ontology runtime version; generating quality discovery and quality reports according to semantic constraints in the ontology runtime version, and deciding whether to block a cleaning version according to the severity of the quality discovery; when the cleaning version is not blocked, generating a feature portrait according to the ontology runtime version, performing data mining based on the feature portrait, and generating a mining result; binding the mining result to each version identifier, generating an evidence graph, and giving a review state to support business action management; and the application realizes the technical effects of semantic unification of multi-source heterogeneous data, controllable quality, interpretable features, traceable results and reviewable business by taking the ontology model as an executable semantic control layer throughout the whole data mining process.

Description

Technical Field

[0001] This application relates to the field of data mining technology, and in particular to a data mining method and system based on ontology model enhancement. Background Technology

[0002] Existing data mining systems typically revolve around data tables, fields, feature engineering, and algorithm models. When performing association rule analysis, sequence pattern analysis, clustering, trend analysis, anomaly detection, or predictive modeling, these systems primarily rely on field names, statistical distributions, data types, and manual configuration to determine input features and algorithm parameters. This approach can achieve certain results with a single data source, clearly defined field meanings, and relatively good data quality. However, when enterprise data comes from multiple business systems, API interfaces, databases, files, and documents, and there are inconsistencies in field naming, business meanings, frequent quality issues, and stringent access and auditing requirements, traditional data mining processes are prone to problems such as unclear input semantics, non-reusable cleaning rules, uninterpretable feature generation, difficulty in tracking model results, and the inability to apply insight reports to business objects and action governance.

[0003] On the other hand, existing ontology modeling or knowledge graph systems typically focus on the representation of entities, attributes, relationships, and rules, and use ontology models for querying, displaying, reasoning, or knowledge management, but do not deeply embed ontology models into the entire data mining process. In other words, ontology models are often just interpretive material after data mining, rather than the actual control boundaries for data access, quality inspection, cleaned version generation, feature profiling, algorithm selection, mining result verification, and action execution. Summary of the Invention

[0004] The purpose of this application is to provide a data mining method and system based on ontology model enhancement, so as to solve the above-mentioned problems existing in the prior art.

[0005] To achieve the above objectives, this application provides a data mining method based on ontology model enhancement, comprising the following steps: S1: Acquire multi-source heterogeneous data, parse the data structure of multi-source heterogeneous data, and generate data assets and corresponding resource identifiers; S2: Obtain the ontology model, which includes entity types, attribute types, relationship types, and business rules. Map the data assets to the entity types, attribute types, relationship types, and business rules in the ontology model to generate the runtime version of the ontology. S3: Based on the semantic constraints in the runtime version of the ontology, perform a quality scan on the data assets and generate quality findings and quality reports; decide whether to block the generation of the cleaned version based on the severity of the quality findings and the preset blocking strategy. S4: If the cleaned version is not blocked, generate a feature profile based on the cleaned data assets and the runtime version of the ontology. The feature profile is generated based on the object type, attribute type, relationship path and time semantics in the ontology model. S5: Based on the semantic task objectives and semantic constraints in the runtime version of the ontology, perform data mining based on feature profiles to generate mining results; S6: Bind the mining results to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile, generate an evidence graph, and assign a review status to support business action governance.

[0006] Preferably, S1 specifically includes: Multi-source heterogeneous data includes structured data, semi-structured data, document data, API interface data, and streaming event data; The data structure of multi-source heterogeneous data is parsed, and structured metadata is extracted. The structured metadata includes data tables, field definitions, primary key constraints, data types, and field statistical information. Generate stable resource identifiers for data sources, data tables, and fields, and record access permission contexts and tenant information.

[0007] Preferably, generating the runtime version of the ontology in S2 specifically includes: Obtain the ontology model, which includes entity types, attribute types, relationship types, and a set of business rules; Map data assets to entity types, attribute types, relationship types, and business rules in the ontology model to establish a correspondence between the structured metadata of data assets and the ontology model; Compile business rules into an executable set of semantic constraints; Package the mapping relationships and semantic constraint sets to generate the runtime version of the ontology.

[0008] Preferably, in S3, a quality scan of the data assets is performed based on the semantic constraints in the ontology runtime version, generating quality findings and quality reports, specifically including: Generate quality inspection rules based on the entity primary keys, attribute types, enumeration values, relationship integrity, and business rules in the ontology model; Data assets are scanned using quality inspection rules to detect issues such as missing primary keys, mismatched attribute types, abnormal enumeration values, excessive field missing rates, and incomplete relationships. The detected issues are recorded as quality findings, and quality reports are generated based on these findings.

[0009] Preferably, in S3, the decision on whether to block the generation of the cleaned version is based on the severity of the quality discovery and a preset blocking strategy, specifically including: Obtain the review results of the quality findings, including acceptance, ignoring, or blocking; If at least one quality discovery verification result is a block, then the generation of the current cleaned version is terminated; If no blocking verification result is found, the data asset is cleaned according to the preset cleaning rules, a cleaned version is generated, and the lineage information of the data asset to the cleaned version is recorded.

[0010] Preferably, in S4, feature profiles are generated based on the cleaned data assets and the ontology runtime version, specifically including: By combining ontology objects, attribute types, time fields, category fields, relationship paths, and quality results from the ontology runtime version, feature profiles are generated from the cleaned data assets.

[0011] Preferably, S5 specifically includes: Based on the semantic task objectives and the semantic constraints in the runtime version of the ontology, and based on the feature profile, a data mining task is selected from the preset mining task types and executed to generate mining results and record algorithm metadata. Data mining tasks include association rule mining, sequence pattern mining, cluster analysis, trend analysis, anomaly detection, and predictive modeling.

[0012] Preferably, in S6, the mining results are bound to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile to generate an evidence graph, specifically including: The resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile are used as evidence nodes; Use the lineage information of data assets to the cleaned version as evidence edges, and use the ontology mapping information and review status information as evidence edges; By combining evidence nodes and evidence edges, an evidence graph is generated, making each mining result traceable to the original data asset, the cleaned version, the feature profile, and the runtime version of the ontology.

[0013] Preferably, in S6, a review status is assigned to support business action governance, specifically including: Each mining result is associated with its corresponding ontology object, evidence graph, and review status to generate an insight report; The application center displays insight reports and receives review decisions for each mining result, including accepting, ignoring, or blocking. Write the review decision into the audit event and generate an exportable evidence package containing evidence graphs, lineage information, review status, and runtime object evidence.

[0014] An ontology-based data mining system employs an ontology-based data mining method, comprising: The data access module is used to acquire multi-source heterogeneous data, parse the data structure of multi-source heterogeneous data, and generate data assets and corresponding resource identifiers. The ontology model module is used to obtain the ontology model, which includes entity types, attribute types, relationship types, and business rules. It maps data assets to the entity types, attribute types, relationship types, and business rules in the ontology model and generates the runtime version of the ontology. The data refining module is used to perform quality scans on data assets based on semantic constraints in the runtime version of the ontology, generate quality findings and quality reports, and decide whether to block the generation of the cleaned version based on the severity of the quality findings and preset blocking strategies. The feature generation module is used to generate feature profiles based on the cleaned data assets and the runtime version of the ontology, provided that the cleaned version is not blocked. The feature profiles are generated based on the object type, attribute type, relationship path and temporal semantics in the ontology model. The data mining execution module is used to perform data mining based on feature profiles and generate mining results, according to the semantic task objectives and semantic constraints in the runtime version of the ontology. The results governance module is used to bind the mining results to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile, generate evidence graphs, and assign a review status to support business action governance.

[0015] Therefore, the above-mentioned data mining method and system based on ontology model enhancement has the following beneficial effects: (1) This application uses ontology model to constrain data access, quality detection and cleaning version generation, which can improve the semantic consistency and quality controllability of multi-source data before entering the mining algorithm.

[0016] (2) By incorporating ontology objects, attributes, relationships and rules into the feature generation process, this application can reduce the misuse of features caused by simply relying on field names and statistical distributions.

[0017] (3) By binding the mining results to input resources, cleaned versions, feature profiles, ontology runtime versions and audit events, this application can significantly improve the interpretability and traceability of the results.

[0018] (4) By introducing review status, blocking mechanism, business action governance and exportable evidence package, this application can transform data mining results into business objects that are reviewable, executable and auditable for enterprise applications.

[0019] (5) This application can reduce the risk of crosstalk between different projects, different data versions and different ontology runtime versions by using a combination of project identifier, cleaning version identifier, feature profile identifier and ontology runtime version identifier for isolation.

[0020] The technical solution of this application will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating a data mining method based on ontology model enhancement according to this application. Detailed Implementation

[0022] The following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0023] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning as understood by a person of ordinary skill in the art to which this application pertains.

[0024] The terms "comprising" or "including," as used in this application, mean that the element preceding the term encompasses the element listed after the term, and do not exclude the possibility of encompassing other elements as well. The terms "inner," "outer," "upper," and "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. When the absolute position of the described object changes, the relative positional relationship may also change accordingly. In this application, unless otherwise expressly specified and limited, the term "attached," etc., should be interpreted broadly. For example, it can refer to a fixed connection, a detachable connection, or an integral part; it can refer to a direct connection or an indirect connection through an intermediate medium; it can refer to the internal communication of two elements or the interaction relationship between two elements. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0025] Example 1: A data mining method based on ontology model enhancement, such as Figure 1 As shown, it includes the following steps: S1: Acquire multi-source heterogeneous data, parse the data structure of multi-source heterogeneous data, and generate data assets and corresponding resource identifiers; Multi-source heterogeneous data includes structured data, semi-structured data, document data, API interface data, and streaming event data; The data structure of multi-source heterogeneous data is parsed, and structured metadata is extracted. The structured metadata includes data tables, field definitions, primary key constraints, data types, and field statistical information. Generate stable resource identifiers for the data source, data table, and fields, and record access permission context and tenant information. This step provides stable resource identifiers and structured input for subsequent ontology mapping.

[0026] S2: Obtain the ontology model, which includes entity types, attribute types, relation types, and business rules. Map the data assets to the entity types, attribute types, relation types, and business rules in the ontology model to generate the runtime version of the ontology, which serves as the core source of semantic constraints and mining task context. Obtain the ontology model, which includes entity types, attribute types, relationship types, and a set of business rules; Map data assets to entity types, attribute types, relationship types, and business rules in the ontology model to establish a correspondence between the structured metadata of data assets and the ontology model; Compile business rules into an executable set of semantic constraints; Package the mapping relationships and semantic constraint sets to generate the runtime version of the ontology.

[0027] S3: Based on the semantic constraints in the runtime version of the ontology, perform a quality scan on the data assets and generate quality findings and quality reports; decide whether to block the generation of the cleaned version based on the severity of the quality findings and the preset blocking strategy. Based on the semantic constraints in the ontology runtime version, a quality scan is performed on the data assets to generate quality findings and quality reports, specifically including: Generate quality inspection rules based on the entity primary keys, attribute types, enumeration values, relationship integrity, and business rules in the ontology model; Data assets are scanned using quality inspection rules to detect issues such as missing primary keys, mismatched attribute types, abnormal enumeration values, excessive field missing rates, and incomplete relationships. The detected issues are recorded as quality findings, and quality reports are generated based on these findings.

[0028] The decision to block the generation of the cleaned version is based on the severity of the quality findings and the preset blocking strategy, specifically including: Obtain the review results of the quality findings, including acceptance, ignoring, or blocking; If at least one quality discovery verification result is a block, then the generation of the current cleaned version is terminated; If no blocking verification result is found, the data asset is cleaned according to the preset cleaning rules, a cleaned version is generated, and the lineage information of the data asset to the cleaned version is recorded.

[0029] This step determines the quality inspection rules, cleaning rules, and blocking strategies based on ontological constraints.

[0030] S4: If the cleaned version is not blocked, generate a feature profile based on the cleaned data assets and the runtime version of the ontology. The feature profile is generated based on the object type, attribute type, relationship path and time semantics in the ontology model. Based on the cleaned data assets and the ontology runtime version, a feature profile is generated, specifically including: By combining ontology objects, attribute types, time fields, category fields, relationship paths, and quality results from the ontology runtime version, feature profiles are generated from the cleaned data assets.

[0031] S5: Based on the semantic task objectives and semantic constraints in the runtime version of the ontology, perform data mining based on feature profiles to generate mining results; Based on the semantic task objectives and the semantic constraints in the runtime version of the ontology, and based on the feature profile, a data mining task is selected from the preset mining task types and executed to generate mining results and record algorithm metadata. Data mining tasks include association rule mining, sequence pattern mining, cluster analysis, trend analysis, anomaly detection, and predictive modeling.

[0032] S6: Bind the mining results to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile, generate an evidence graph, and assign a review status to support business action governance. Bind the mining output to the ontology object, runtime action, and audit chain, and realize result review, action governance, and traceable consumption.

[0033] The mining results are bound to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile to generate an evidence graph, specifically including: The resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile are used as evidence nodes; Use the lineage information of data assets to the cleaned version as evidence edges, and use the ontology mapping information and review status information as evidence edges; By combining evidence nodes and evidence edges, an evidence graph is generated, making each mining result traceable to the original data asset, the cleaned version, the feature profile, and the runtime version of the ontology.

[0034] Assigning a review status to support business action governance includes: Each mining result is associated with its corresponding ontology object, evidence graph, and review status to generate an insight report; The application center displays insight reports and receives review decisions for each mining result, including accepting, ignoring, or blocking. Write the review decision into the audit event and generate an exportable evidence package containing evidence graphs, lineage information, review status, and runtime object evidence.

[0035] Example 2: An ontology-based data mining system employs an ontology-based data mining method, comprising: The data access module is used to acquire multi-source heterogeneous data, parse the data structure of multi-source heterogeneous data, and generate data assets and corresponding resource identifiers. The ontology model module is used to obtain the ontology model, which contains entity types, attribute types, relationship types, and business rules. It maps data assets to entity types, attribute types, relationship types, and business rules in the ontology model and generates the runtime version of the ontology. The data refining module is used to perform quality scans on data assets based on semantic constraints in the runtime version of the ontology, generate quality findings and quality reports, and decide whether to block the generation of the cleaned version based on the severity of the quality findings and preset blocking strategies. The feature generation module is used to generate feature profiles based on the cleaned data assets and the runtime version of the ontology, provided that the cleaned version is not blocked. The feature profiles are generated based on the object type, attribute type, relationship path and temporal semantics in the ontology model. The data mining execution module is used to perform data mining based on feature profiles and generate mining results, according to the semantic task objectives and semantic constraints in the runtime version of the ontology. The results governance module is used to bind the mining results to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile, generate evidence graphs, and assign a review status to support business action governance.

[0036] Example 3: To further illustrate the effectiveness of the method in this application, a specific retail business scenario will be used as an example for detailed description below.

[0037] The system integrates a sales details table with fields including transaction number, customer number, product category, product name, unit price, quantity, payment method, store location, transaction date, and whether a discount is applied. The system first registers the table in the data asset management center and generates stable resource identifiers for the data source, table, and fields. Subsequently, the ontology model layer establishes entity types such as customer, product, transaction, store, and payment method, mapping table fields to corresponding attributes or relationships. The system performs a quality scan based on the primary key constraints of the transaction entity, the time attribute of the transaction date, the category attribute of the product category, and the enumeration attribute of the payment method. When issues such as missing primary keys, excessively high field missing rates, or enumeration anomalies are detected, a quality discovery is generated, allowing users to perform accept, ignore, or block operations.

[0038] When quality issues are not prevented, the system generates a cleaned version according to the cleaning rules, recording the pattern checksum, number of rows, byte size, content hash, and lineage information from the original data asset to the cleaned version. Subsequently, the system combines the ontology model to generate feature profiles, such as identifying transaction dates as time indices, sales amounts as numerical indicators, product categories and payment methods as classification dimensions, and customer IDs and transaction IDs as identifier features. Based on these features, the system can perform trend analysis to identify sales change windows, execute association rules to identify combinations of product categories and payment methods, execute sequence patterns to identify customer transaction paths, and perform clustering to discover abnormal consumer groups. Each mining result includes the corresponding data asset, cleaned version, feature profile, ontology object, algorithm configuration, and evidence graph nodes.

[0039] During the results output phase, the system writes the mined insights into an insight report and displays the findings, recommended actions, and audit notes related to the ontology objects in the application center. For example, when the system detects an abnormal growth trend of a certain type of product in a specific store, this insight will be associated with the product entity, store entity, sales transaction entity, trend window evidence, cleaned versions, and quality reports. Business users can review the insight in the application center and choose to accept, ignore, or block subsequent actions; the system simultaneously writes audit events and allows the export of evidence packages containing evidence of lineage, audits, graph nodes, edges, and runtime objects.

[0040] Therefore, this application adopts the above-mentioned data mining method and system based on ontology model enhancement. By using the ontology model as an executable semantic control layer throughout the entire data mining process, it achieves the technical effects of semantic unification, quality control, feature interpretability, result traceability, and business verification of multi-source heterogeneous data. It is applicable to scenarios where there are structured data, semi-structured data, document data, API data, and runtime business object data within an enterprise. It can establish an executable semantic context before the data enters the mining algorithm and form a traceable, verifiable, and auditable evidence chain after the mining results are output.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and not to limit them. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of this application, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of this application.

Claims

1. A data mining method based on ontology model enhancement, characterized in that, Includes the following steps: S1: Acquire multi-source heterogeneous data, parse the data structure of multi-source heterogeneous data, and generate data assets and corresponding resource identifiers; S2: Obtain the ontology model, which includes entity types, attribute types, relationship types, and business rules. Map the data assets to the entity types, attribute types, relationship types, and business rules in the ontology model to generate the runtime version of the ontology. S3: Based on the semantic constraints in the runtime version of the ontology, perform a quality scan on the data assets and generate quality findings and quality reports; decide whether to block the generation of the cleaned version based on the severity of the quality findings and the preset blocking strategy. S4: If the cleaned version is not blocked, generate a feature profile based on the cleaned data assets and the runtime version of the ontology. The feature profile is generated based on the object type, attribute type, relationship path and time semantics in the ontology model. S5: Based on the semantic task objectives and semantic constraints in the runtime version of the ontology, perform data mining based on feature profiles to generate mining results; S6: Bind the mining results to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile, generate an evidence graph, and assign a review status to support business action governance.

2. The data mining method based on ontology model enhancement as described in claim 1, characterized in that, S1 specifically includes: Multi-source heterogeneous data includes structured data, semi-structured data, document data, API interface data, and streaming event data; The data structure of multi-source heterogeneous data is analyzed, and structured metadata is extracted. The structured metadata includes data tables, field definitions, primary key constraints, data types, and field statistical information. Generate stable resource identifiers for data sources, data tables, and fields, and record access permission contexts and tenant information.

3. The data mining method based on ontology model enhancement as described in claim 2, characterized in that, The generation of the runtime version of the ontology in S2 specifically includes: Obtain the ontology model, which includes entity types, attribute types, relationship types, and a set of business rules; Map data assets to entity types, attribute types, relationship types, and business rules in the ontology model to establish a correspondence between the structured metadata of data assets and the ontology model; Compile business rules into an executable set of semantic constraints; Package the mapping relationships and semantic constraint sets to generate the runtime version of the ontology.

4. The data mining method based on ontology model enhancement as described in claim 1, characterized in that, S3 performs quality scans on data assets based on semantic constraints in the ontology runtime version, generating quality findings and quality reports, specifically including: Generate quality inspection rules based on the entity primary keys, attribute types, enumeration values, relationship integrity, and business rules in the ontology model; Data assets are scanned using quality inspection rules to detect issues such as missing primary keys, mismatched attribute types, abnormal enumeration values, excessive field missing rates, and incomplete relationships. The detected issues are recorded as quality findings, and quality reports are generated based on these findings.

5. The data mining method based on ontology model enhancement according to claim 2, characterized in that, S3 determines whether to block the generation of the cleaned version based on the severity of the quality discovery and the preset blocking strategy, specifically including: Obtain the review results of the quality findings, including acceptance, ignoring, or blocking; If at least one quality discovery verification result is a block, then the generation of the current cleaned version is terminated; If no blocking verification result is found, the data asset is cleaned according to the preset cleaning rules, a cleaned version is generated, and the lineage information of the data asset to the cleaned version is recorded.

6. The data mining method based on ontology model enhancement as described in claim 1, characterized in that, S4 generates feature profiles based on the cleaned data assets and the ontology runtime version, specifically including: By combining ontology objects, attribute types, time fields, category fields, relationship paths, and quality results from the ontology runtime version, feature profiles are generated from the cleaned data assets.

7. The data mining method based on ontology model enhancement as described in claim 1, characterized in that, S5 specifically includes: Based on the semantic task objectives and the semantic constraints in the runtime version of the ontology, and based on the feature profile, a data mining task is selected from the preset mining task types and executed to generate mining results and record algorithm metadata. Data mining tasks include association rule mining, sequence pattern mining, cluster analysis, trend analysis, anomaly detection, and predictive modeling.

8. The data mining method based on ontology model enhancement as described in claim 1, characterized in that, In S6, the mining results are bound to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile to generate an evidence graph, specifically including: The resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile are used as evidence nodes; Use the lineage information of data assets to the cleaned version as evidence edges, and use the ontology mapping information and review status information as evidence edges; By combining evidence nodes and evidence edges, an evidence graph is generated, making each mining result traceable to the original data asset, the cleaned version, the feature profile, and the runtime version of the ontology.

9. The data mining method based on ontology model enhancement as described in claim 1, characterized in that, In S6, a review status is assigned to support business action governance, specifically including: Each mining result is associated with its corresponding ontology object, evidence graph, and review status to generate an insight report; The application center displays insight reports and receives review decisions for each mining result, including accepting, ignoring, or blocking. Write the review decision into the audit event and generate an exportable evidence package containing evidence graphs, lineage information, review status, and runtime object evidence.

10. A data mining system based on ontology model enhancement, employing a data mining method based on ontology model enhancement as described in any one of claims 1-9, characterized in that, include: The data access module is used to acquire multi-source heterogeneous data, parse the data structure of multi-source heterogeneous data, and generate data assets and corresponding resource identifiers. The ontology model module is used to obtain the ontology model, which includes entity types, attribute types, relationship types, and business rules. It maps data assets to the entity types, attribute types, relationship types, and business rules in the ontology model and generates the runtime version of the ontology. The data refining module is used to perform quality scans on data assets based on semantic constraints in the runtime version of the ontology, generate quality findings and quality reports, and decide whether to block the generation of the cleaned version based on the severity of the quality findings and preset blocking strategies. The feature generation module is used to generate feature profiles based on the cleaned data assets and the runtime version of the ontology, provided that the cleaned version is not blocked. The feature profiles are generated based on the object type, attribute type, relationship path and temporal semantics in the ontology model. The data mining execution module is used to perform data mining based on feature profiles and generate mining results, according to the semantic task objectives and semantic constraints in the runtime version of the ontology. The results governance module is used to bind the mining results to the resource identifier of the data asset, the version identifier of the ontology runtime version, the version identifier of the cleaned version, and the version identifier of the feature profile, generate evidence graphs, and assign a review status to support business action governance.

Images