ETL semantic ambiguity resolution method, device, equipment and medium

Abstract

The ETL semantic ambiguity resolution method, apparatus, device, and medium provided in this disclosure include: determining a target subgraph corresponding to the target instruction information based on target instruction information submitted by the target object and a target domain knowledge graph; generating multiple candidate semantic paths and the information entropy of each candidate semantic path based on the target subgraph and the target domain knowledge graph; and determining the target semantic path based on each candidate semantic path and the corresponding information entropy. This reduces the semantic ambiguity of the generated target semantic path.

Description

Technical Field

[0001] This invention relates to the field of data processing technology and related technical fields, specifically to an ETL semantic ambiguity resolution method, apparatus, device, and medium. Background Technology

[0002] In modern enterprise data architecture, the Extract-Transform-Load (ETL) process is the underlying foundation for building data platforms, data warehouses, and data lakes. Its core task is to process complex data from diverse and heterogeneous environments in a structured, repeatable, and highly reliable manner. Traditional ETL development heavily relies on data engineers writing SQL scripts or performing tedious manual configurations using graphical integrated development environments. This not only leads to long delays in responding to business needs but also easily introduces human errors during complex logical transformations. To lower the development threshold and improve data democratization, the introduction of Natural Language Interfaces (NLEs) for semantic parsing—automatically converting unstructured natural language descriptions into executable process definitions—has become a cutting-edge direction of common interest in both industry and academia.

[0003] Despite significant advancements in semantic parsing technology in recent years, existing techniques still reveal a series of deep-seated shortcomings in natural language applications for ETL. These problems are primarily attributed to the inherent gap between large language models, as probabilistic prediction engines, and the high demands of ETL tasks on determinism, compliance, and formal logic.

[0004] Given the problems with existing technologies, there is an urgent need for an ETL semantic ambiguity resolution method. Summary of the Invention

[0005] The embodiments described herein provide an ETL semantic ambiguity resolution method, apparatus, device, and medium that reduces the semantic ambiguity of the generated target semantic path.

[0006] Firstly, based on the content of this disclosure, an ETL semantic ambiguity resolution method is provided, including: Based on the target instruction information submitted by the target object and the target domain knowledge graph, determine the target subgraph corresponding to the target instruction information; Based on the target subgraph and the target domain knowledge graph, generate multiple candidate semantic paths and the information entropy of each semantic candidate path; The target semantic path is determined based on each of the candidate semantic paths and the information entropy corresponding to each candidate semantic path.

[0007] In some embodiments of this disclosure, determining the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph includes: Based on the target instruction information submitted by the target object, determine the seed nodes included in the target instruction information, wherein the seed nodes include Abox entity nodes and Tbox concept nodes; Based on the seed node and the target domain knowledge graph, the target subgraph corresponding to the target instruction information is determined.

[0008] In some embodiments of this disclosure, determining the target subgraph corresponding to the target instruction information based on the seed node and the target domain knowledge graph includes: Starting from the seed node, multiple initial subgraphs are searched from the target domain knowledge graph based on a bidirectional breadth-first search strategy; Based on the correlation score between each initial subgraph, the initial subgraphs are spliced ​​together to obtain the target subgraph corresponding to the target instruction information.

[0009] In some embodiments of this disclosure, generating multiple candidate semantic paths and the information entropy of each semantic candidate path based on the target subgraph and the target domain knowledge graph includes: Obtain the node information of each Abox entity node in the target subgraph, wherein the node information includes entity label, entity type and entity attributes; Based on the node information of each Abox entity node in the target subgraph and the connection relationship of each Abox entity node in the target subgraph, determine the natural language relationship description corresponding to the target subgraph; Based on the Abox entity nodes included in the target subgraph and the target domain knowledge graph, determine the Tbox concept nodes corresponding to each Abox entity node and the descriptive logic axioms between Abox entity nodes and Tbox concept nodes; Based on the descriptive logic axioms between the Abox entity nodes and the Tbox concept nodes, the natural language constraints are determined. Based on the natural language relation description and the natural language constraints, multiple candidate semantic paths and the information entropy of each candidate semantic path are generated.

[0010] In some embodiments of this disclosure, determining the Tbox concept node corresponding to each Abox entity node and the descriptive logic axiom between Abox entity nodes and Tbox concept nodes based on the Abox entity nodes included in the target subgraph and the target domain knowledge graph includes: Based on the Abox entity nodes included in the target subgraph, determine the Tbox concept nodes corresponding to the Abox entity nodes; Based on the Tbox concept node and the target domain knowledge graph, the descriptive logic axioms between the Tbox concept node and the Abox entity node are determined.

[0011] In some embodiments of this disclosure, determining the target semantic path based on each of the candidate semantic paths and the information entropy corresponding to each candidate semantic path includes: The semantic ambiguity score is determined based on the information entropy corresponding to each candidate semantic path. If the semantic ambiguity score is greater than or equal to the preset semantic ambiguity score, the target semantic path is determined in response to receiving the instruction information submitted by the target object.

[0012] In some embodiments of this disclosure, the method further includes: If the semantic ambiguity score is less than the preset semantic ambiguity score, the candidate semantic path with the largest information entropy value is selected as the target semantic path based on the information entropy corresponding to each candidate semantic path.

[0013] Secondly, according to the present disclosure, an ETL semantic ambiguity resolution apparatus is provided, comprising: The target subgraph determination module is used to determine the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph; The information determination module is used to generate multiple candidate semantic paths and the information entropy of each semantic candidate path based on the target subgraph and the target domain knowledge graph. The target semantic path determination module is used to determine the target semantic path based on each of the candidate semantic paths and the information entropy corresponding to each of the candidate semantic paths.

[0014] Thirdly, according to this disclosure, a computer device is provided, comprising: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any of the first aspects.

[0015] Fourthly, according to this disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the methods described in any of the first aspects.

[0016] The ETL semantic ambiguity resolution method, apparatus, device, and medium provided in this disclosure first determine the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph. Then, based on the target subgraph and the target domain knowledge graph, multiple candidate semantic paths and the information entropy of each candidate semantic path are generated. Finally, the target semantic path is determined based on each candidate semantic path and the semantic ambiguity score corresponding to each candidate semantic path. First, by generating corresponding natural language relation descriptions from the structured target subgraph, and then finding the corresponding TBox concept node along the instanceOf edge for each ABox entity node included in the target subgraph, all descriptive logical axioms associated with the TBox concept node in the target domain knowledge graph are extracted as natural language constraints for the ABox entity node, significantly improving the interpretability of the disambiguation results. Finally, based on the information entropy corresponding to each candidate semantic path, the target semantic path is determined from multiple candidate semantic paths. The process of determining the target semantic path based on information entropy avoids "blind guessing" in ambiguous scenarios and effectively reduces semantic ambiguity.

[0017] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments will be briefly described below. It should be understood that the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure, wherein: Figure 1 This is a flowchart illustrating an ETL semantic ambiguity resolution method provided in an embodiment of this disclosure; Figure 2 This is a schematic diagram of the structure of an ETL semantic ambiguity resolution device provided in an embodiment of this disclosure; Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure.

[0019] In the accompanying diagram, markers with the same last two digits correspond to the same elements. It should be noted that the elements in the diagram are schematic and not drawn to scale. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are also within the scope of protection of this disclosure.

[0021] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this subject matter pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having the meaning consistent with their meaning in the context of the specification and in the relevant art, and shall not be interpreted in an idealized or overly formal form unless otherwise explicitly defined herein. As used herein, the statement of “connecting” or “coupling” two or more parts together shall mean that these parts are directly joined together or joined through one or more intermediate components.

[0022] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of the phrase "embodiment" in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0023] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists, A and B exist simultaneously, or B exists. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0024] Furthermore, in all embodiments of this disclosure, terms such as “first” and “second” are used only to distinguish one component (or part of a component) from another component (or another part of a component).

[0025] In the description of this application, unless otherwise stated, "multiple" means two or more (including two), and similarly, "multiple groups" means two or more (including two groups).

[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0027] In view of the problems existing in the prior art, the present disclosure provides an ETL semantic ambiguity resolution method. Figure 1 This is a flowchart illustrating an ETL semantic ambiguity resolution method provided in an embodiment of this disclosure, as shown below. Figure 1 As shown, ETL semantic ambiguity resolution methods include: S110. Based on the target instruction information submitted by the target object and the target domain knowledge graph, determine the target subgraph corresponding to the target instruction information.

[0028] This disclosure addresses the "multi-semantic traps" and "logical illusions" that exist in Large Language Models (LLMs) in specific vertical domains (such as ETL data integration). It proposes an ETL semantic ambiguity resolution method, which uses a constructed target domain knowledge graph (ETL domain knowledge graph) as a "truth source" to limit the parsing process of LLM for target instructions submitted by the target object to a path that conforms to the domain syntax and logical axioms, thereby achieving high-precision and interpretable semantic parsing.

[0029] Among them, the target domain knowledge graph, also known as the ETL domain knowledge graph, deeply integrates description logic and realizes the leap from "data storage" to "knowledge reasoning".

[0030] The target domain knowledge graph consists of a terminology layer and an assertion layer. The terminology layer includes multiple predefined Tbox concept nodes, and the assertion layer includes multiple predefined Abox entity nodes. The terminology layer defines rules, defining the "metadata" and "physical laws" of the ETL domain. It doesn't concern itself with the specific data, but only with the relationships and constraints between the data, such as SumAggregator. hasParameter.NumericColumn (The summation aggregator must be applied to a numeric column), Join A Transformer (a connection is a converter). The term "store instance" stores specific business metadata, or "facts," which serve as the basis for linking entities and helping the system understand the specific business objects mentioned in user commands. Examples include specific table names, field names (e.g., customer_table.id), business aliases (e.g., "outlier" linked to Negative_Value_Constraint), and data relationships (the output of table A is the input of table B).

[0031] In a specific implementation, the target subgraph corresponding to the target instruction information is determined based on the target instruction information submitted by the target object and the target domain knowledge graph. This includes: determining the seed nodes included in the target instruction information based on the target instruction information submitted by the target object, wherein the seed nodes include Abox entity nodes and Tbox concept nodes; and determining the target subgraph corresponding to the target instruction information based on the seed nodes and the target domain knowledge graph.

[0032] Specifically, the process of determining the target subgraph corresponding to the target instruction information based on the seed node and the target domain knowledge graph includes: starting from the seed node, searching for multiple initial subgraphs from the target domain knowledge graph using a bidirectional breadth-first search strategy; and concatenating the initial subgraphs based on the relevance scores between them to obtain the target subgraph corresponding to the target instruction information.

[0033] For example, the target instruction submitted by the target object is: "Please help me merge the customer table and the order table, and filter out the outliers from last year." The identified seed nodes include: merge, customer table, order table, and outliers. Among them, the merge seed node corresponds to Join / Union in the target domain knowledge graph, the customer table seed node corresponds to customer_table in the target domain knowledge graph, the order table seed node corresponds to order_table in the target domain knowledge graph, and the outlier seed node corresponds to Negative_Value_Constraint in the target domain knowledge graph. Among the above seed nodes, the merge seed node is a Tbox concept node, and the customer table seed node, order table seed node, and outlier seed node are all Abox entity nodes.

[0034] The specific process for determining the seed nodes included in the target instruction information is to perform word segmentation on the target instruction information, and determine the seed nodes included in the target instruction information based on the similarity between each word vector after word segmentation and the preset word vectors included in the target domain knowledge graph (that is, the word vectors corresponding to the preset Abox entity nodes and preset Tbox concept nodes included in the target domain knowledge graph).

[0035] After identifying the seed nodes included in the target instruction information submitted by the target object, a bidirectional breadth-first search strategy is adopted, expanding outward from the seed nodes. During the outward expansion based on the bidirectional breadth-first search strategy, associated nodes with various child nodes are determined according to the preset hop count parameter. Various child nodes and their corresponding associated nodes form an initial subgraph. After determining the initial subgraphs corresponding to various child nodes, the initial subgraphs are spliced ​​together according to the correlation score between them to obtain the target subgraph corresponding to the target instruction information.

[0036] The preset number of jumps is usually set to 1 step or 2 steps.

[0037] A specific example, starting from the seed node `customer_table` in the customer table (with a default hop count parameter of 1 step), the determined associated nodes include: → hasField → id (field) → hasField → name (field) → hasField → age (field) → instanceOf → Table (TBox) → hasPrimaryKey → id (constraint) Based on the associated nodes found after a preset hop count of 1 step, starting from the id (a field in customer_table) node (preset hop count parameter is 2 steps), the identified associated nodes include: → isReferencedBy → (order_table.customer_id) → instanceOf → PrimaryKey Through the above implementation process, the initial subgraphs corresponding to various child nodes are determined in sequence. Then, by using the correlation scores of each node in the initial subgraphs corresponding to various child nodes, the initial subgraphs are spliced ​​together to obtain the target subgraph corresponding to the target instruction information.

[0038] S120. Based on the target subgraph and the target domain knowledge graph, generate multiple candidate semantic paths and the information entropy of each candidate semantic path.

[0039] The information entropy of each semantic candidate path measures the degree of dispersion of the probability distribution of the semantic candidate paths.

[0040] In a specific implementation, multiple candidate semantic paths and the information entropy of each semantic candidate path are generated based on the target subgraph and the target domain knowledge graph. This includes: obtaining node information of each Abox entity node in the target subgraph, wherein the node information includes entity label, entity type, and entity attributes; determining the natural language relation description corresponding to the target subgraph based on the node information of each Abox entity node in the target subgraph and the connection relationship between each Abox entity node in the target subgraph; determining the Tbox concept node corresponding to each Abox entity node and the descriptive logic axiom between Abox entity nodes and Tbox concept nodes based on the Abox entity nodes included in the target subgraph and the target domain knowledge graph; determining natural language constraints based on the descriptive logic axiom between Abox entity nodes and Tbox concept nodes; and generating multiple candidate semantic paths and the information entropy of each semantic candidate path based on the natural language relation description and the natural language constraints.

[0041] Specifically, based on the Abox entity nodes included in the target subgraph and the target domain knowledge graph, the Tbox concept nodes corresponding to each Abox entity node and the descriptive logic axioms between Abox entity nodes and Tbox concept nodes are determined, including: determining the Tbox concept nodes corresponding to Abox entity nodes based on the Abox entity nodes included in the target subgraph; and determining the descriptive logic axioms between Tbox concept nodes and Abox entity nodes based on Tbox concept nodes and the target domain knowledge graph.

[0042] In this implementation step, firstly, all ABox entity nodes in the target subgraph are traversed to extract the entity label, entity type, and entity attributes of each entity node. Then, the edges in the target subgraph, that is, the connection relationships of each ABox entity node in the target subgraph, are traversed. Based on the node information of each ABox entity node and the connection relationships of each ABox entity node, the target subgraph is transformed into a natural language relation description.

[0043] In the process of converting the target subgraph into a natural language relation description, the node information of each Abox entity node and the connection relationship of each Abox entity node can be input into the LLM model, and a natural language relation description corresponding to the target subgraph can be generated based on a preset template.

[0044] Secondly, for each ABox entity node, its corresponding TBox concept node is found along the instanceOf edge, and all the descriptive logic axioms associated with the TBox concept node in the target domain knowledge graph are extracted. Then, the descriptive logic axioms between the TBox concept node and the ABox entity node are determined. By inputting the descriptive logic axioms between the TBox concept node and the ABox entity node into the LLM model, the natural language constraints corresponding to the descriptive logic axioms are output by the LLM model. The natural language constraints corresponding to each descriptive logic axiom are used as constraints for the natural language relation description.

[0045] Finally, based on the natural language relation description and natural language constraints, multiple candidate semantic paths corresponding to the target subgraph and the information entropy of each semantic candidate path are generated using the LLM model.

[0046] In this implementation process, the structured target subgraph is used to generate a corresponding natural language relation description. For each ABox entity node, its corresponding TBox concept node is found along the instanceOf edge, and all the descriptive logic axioms associated with the TBox concept node in the target domain knowledge graph are extracted. Thus, the descriptive logic axioms between the Tbox concept node and the Abox entity node can be used as natural language constraints for the Abox entity node, which significantly improves the interpretability of the disambiguation results.

[0047] S130. Determine the target semantic path based on each candidate semantic path and the information entropy corresponding to each candidate semantic path.

[0048] In a specific implementation, the target semantic path is determined based on each candidate semantic path and the information entropy corresponding to each candidate semantic path, including: determining a semantic ambiguity score based on the information entropy corresponding to each candidate semantic path; and determining the target semantic path in response to receiving instruction information submitted by the target object when the semantic ambiguity score is greater than or equal to a preset semantic ambiguity score.

[0049] If the semantic ambiguity score is less than the preset semantic ambiguity score, the candidate semantic path with the largest information entropy value is selected as the target semantic path based on the information entropy corresponding to each candidate semantic path.

[0050] Specifically, the information entropy corresponding to each candidate semantic path reflects the degree of dispersion of the probability distribution of the candidate semantic paths. If the information entropy corresponding to a certain candidate semantic path is close to 1 and the information entropy corresponding to other candidate semantic paths is close to 0, then the semantic ambiguity score determined according to the information entropy corresponding to each candidate semantic path is lower than the preset semantic ambiguity score. In this case, the candidate semantic path with an information entropy close to 1 is selected as the target semantic path. If the information entropy of multiple candidate semantic paths is close, then the semantic ambiguity score determined according to the information entropy corresponding to each candidate semantic path is greater than or equal to the preset semantic ambiguity score. This can be understood as ambiguity between multiple candidate semantic paths. In this case, by receiving the instruction information submitted by the target object, the target semantic path is selected from multiple candidate semantic paths according to the instruction information submitted by the target object.

[0051] The process of determining the target semantic path based on the instruction information submitted by the target object is the ambiguity resolution process. Based on the instruction information submitted by the target object, the instruction information is used to select the connection relationship between the Tbox concept node and the Abox entity node, thus avoiding "blind guessing" in ambiguous scenarios.

[0052] Among them, the semantic ambiguity score satisfies:

[0053] in, The information entropy corresponding to each candidate semantic path, Score the semantic ambiguity for each of the n candidate semantic paths.

[0054] The ETL semantic ambiguity resolution method provided in this disclosure first determines the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph. Then, based on the target subgraph and the target domain knowledge graph, multiple candidate semantic paths and the information entropy of each candidate semantic path are generated. Finally, the target semantic path is determined based on each candidate semantic path and the semantic ambiguity score corresponding to each candidate semantic path. First, by generating a corresponding natural language relation description from the structured target subgraph, and then finding the corresponding TBox concept node along the instanceOf edge for each ABox entity node included in the target subgraph, all descriptive logical axioms associated with the TBox concept node in the target domain knowledge graph are extracted as natural language constraints for the ABox entity node, significantly improving the interpretability of the disambiguation results. Finally, based on the information entropy corresponding to each candidate semantic path, the target semantic path is determined from multiple candidate semantic paths. The process of determining the target semantic path based on information entropy avoids "blind guessing" in ambiguous scenarios and effectively reduces semantic ambiguity.

[0055] Based on the above embodiments, this disclosure also provides an ETL semantic ambiguity resolution apparatus. Figure 2 This is a schematic diagram of the structure of an ETL semantic ambiguity resolution device provided in an embodiment of this disclosure, as shown below. Figure 2 As shown, the ETL semantic ambiguity resolution apparatus includes: The target subgraph determination module 210 is used to determine the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph; The information determination module 220 is used to generate multiple candidate semantic paths and the information entropy of each semantic candidate path based on the target subgraph and the target domain knowledge graph. The target semantic path determination module 230 is used to determine the target semantic path based on each of the candidate semantic paths and the information entropy corresponding to each of the candidate semantic paths.

[0056] The ETL semantic ambiguity resolution apparatus provided in this embodiment first determines the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph. Then, based on the target subgraph and the target domain knowledge graph, multiple candidate semantic paths and the information entropy of each candidate semantic path are generated. Finally, the target semantic path is determined based on each candidate semantic path and the semantic ambiguity score corresponding to each candidate semantic path. First, by generating a corresponding natural language relation description from the structured target subgraph, and then finding the corresponding TBox concept node along the instanceOf edge for each ABox entity node included in the target subgraph, all descriptive logical axioms associated with the TBox concept node in the target domain knowledge graph are extracted as natural language constraints for the ABox entity node, significantly improving the interpretability of the disambiguation results. Finally, based on the information entropy corresponding to each candidate semantic path, the target semantic path is determined from multiple candidate semantic paths. The process of determining the target semantic path based on information entropy avoids "blind guessing" in ambiguous scenarios and effectively reduces semantic ambiguity.

[0057] In some embodiments of this disclosure, determining the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph includes: Based on the target instruction information submitted by the target object, determine the seed nodes included in the target instruction information, wherein the seed nodes include Abox entity nodes and Tbox concept nodes; Based on the seed node and the target domain knowledge graph, the target subgraph corresponding to the target instruction information is determined.

[0058] In some embodiments of this disclosure, determining the target subgraph corresponding to the target instruction information based on the seed node and the target domain knowledge graph includes: Starting from the seed node, multiple initial subgraphs are searched from the target domain knowledge graph based on a bidirectional breadth-first search strategy; Based on the correlation score between each initial subgraph, the initial subgraphs are spliced ​​together to obtain the target subgraph corresponding to the target instruction information.

[0059] In some embodiments of this disclosure, generating multiple candidate semantic paths and the information entropy of each semantic candidate path based on the target subgraph and the target domain knowledge graph includes: Obtain the node information of each Abox entity node in the target subgraph, wherein the node information includes entity label, entity type and entity attributes; Based on the node information of each Abox entity node in the target subgraph and the connection relationship of each Abox entity node in the target subgraph, determine the natural language relationship description corresponding to the target subgraph; Based on the Abox entity nodes included in the target subgraph and the target domain knowledge graph, determine the Tbox concept nodes corresponding to each Abox entity node and the descriptive logic axioms between Abox entity nodes and Tbox concept nodes; Based on the descriptive logic axioms between the Abox entity nodes and the Tbox concept nodes, the natural language constraints are determined. Based on the natural language relation description and the natural language constraints, multiple candidate semantic paths and the information entropy of each candidate semantic path are generated.

[0060] In some embodiments of this disclosure, determining the Tbox concept node corresponding to each Abox entity node and the descriptive logic axiom between Abox entity nodes and Tbox concept nodes based on the Abox entity nodes included in the target subgraph and the target domain knowledge graph includes: Based on the Abox entity nodes included in the target subgraph, determine the Tbox concept nodes corresponding to the Abox entity nodes; Based on the Tbox concept node and the target domain knowledge graph, the descriptive logic axioms between the Tbox concept node and the Abox entity node are determined.

[0061] In some embodiments of this disclosure, determining the target semantic path based on each of the candidate semantic paths and the information entropy corresponding to each candidate semantic path includes: The semantic ambiguity score is determined based on the information entropy corresponding to each candidate semantic path. If the semantic ambiguity score is greater than or equal to the preset semantic ambiguity score, the target semantic path is determined in response to receiving the instruction information submitted by the target object.

[0062] In some embodiments of this disclosure, the method further includes: If the semantic ambiguity score is less than the preset semantic ambiguity score, the candidate semantic path with the largest information entropy value is selected as the target semantic path based on the information entropy corresponding to each candidate semantic path.

[0063] This application also provides a computer device, please refer to 3 for details. Figure 3 This is a basic structural block diagram of the computer device in this embodiment.

[0064] The computer device includes a memory 510 and a processor 520 that are interconnected via a system bus. It should be noted that only a computer device with components 510-520 is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components may be implemented alternatively. Those skilled in the art will understand that the computer device described herein is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0065] Computer devices can include desktop computers, laptops, handheld computers, and cloud servers. These devices allow for human-computer interaction with users through keyboards, mice, remote controls, touchpads, or voice-activated devices.

[0066] The memory 510 includes at least one type of readable storage medium, including non-volatile memory or volatile memory, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. RAM may include static RAM or dynamic RAM. In some embodiments, the memory 510 may be an internal storage unit of a computer device, such as the hard disk or memory of the computer device. In other embodiments, the memory 510 may also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, or flash card equipped on the computer device. Of course, the memory 510 may include both internal storage units and external storage devices of the computer device. In this embodiment, the memory 510 is typically used to store the operating system and various application software installed on the computer device, such as the program code of the method described above. In addition, the memory 510 may also be used to temporarily store various types of data that have been output or will be output.

[0067] The processor 520 is typically used to perform the overall operation of a computer device. In this embodiment, the memory 510 is used to store program code or instructions, including computer operation instructions. The processor 520 is used to execute the program code or instructions stored in the memory 510 or to process data, such as program code that runs the methods described above.

[0068] In this article, the bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus system can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0069] Another embodiment of this application also provides a computer-readable medium, which may be a computer-readable signal medium or a computer-readable medium. A processor in a computer reads computer-readable program code stored in the computer-readable medium, enabling the processor to execute the functional actions specified in each step or combination of steps in the above method; and to generate means for implementing the functional actions specified in each block or combination of blocks in the block diagram.

[0070] Computer-readable media include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared memory or semiconductor systems, devices or apparatuses, or any suitable combination thereof, wherein the memory is used to store program code or instructions, the program code including computer operation instructions, and the processor is used to execute the program code or instructions of the above-described methods stored in the memory.

[0071] The definitions of memory and processor can be found in the description of the foregoing computer device embodiments, and will not be repeated here.

[0072] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0073] In the various embodiments of this application, the functional units or modules can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0074] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0075] Unless otherwise expressly indicated by the context, the singular form of words used herein and in the appended claims includes the plural form, and vice versa. Thus, when referring to the singular, the plural form of the corresponding term is generally included. Similarly, the terms “comprising” and “including” shall be interpreted as including rather than exclusively. Likewise, the terms “including” and “or” shall be interpreted as including unless such interpretation is expressly prohibited herein. Where the term “example” is used herein, particularly when it follows a set of terms, the “example” is merely exemplary and illustrative and should not be considered exclusive or extensive.

[0076] Further aspects and scope of adaptation become apparent from the description provided herein. It should be understood that various aspects of this application may be implemented individually or in combination with one or more other aspects. It should also be understood that the descriptions and specific embodiments herein are for illustrative purposes only and are not intended to limit the scope of this application.

[0077] Several embodiments of this disclosure have been described in detail above. However, it is obvious that those skilled in the art can make various modifications and variations to the embodiments of this disclosure without departing from the spirit and scope of this disclosure. The scope of protection of this disclosure is defined by the appended claims.

Claims

1. A method for resolving semantic ambiguity in ETL, characterized in that, include: Based on the target instruction information submitted by the target object and the target domain knowledge graph, determine the target subgraph corresponding to the target instruction information; Based on the target subgraph and the target domain knowledge graph, generate multiple candidate semantic paths and the information entropy of each semantic candidate path; The target semantic path is determined based on each of the candidate semantic paths and the information entropy corresponding to each candidate semantic path.

2. The method according to claim 1, characterized in that, The step of determining the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph includes: Based on the target instruction information submitted by the target object, determine the seed nodes included in the target instruction information, wherein the seed nodes include Abox entity nodes and Tbox concept nodes; Based on the seed node and the target domain knowledge graph, the target subgraph corresponding to the target instruction information is determined.

3. The method according to claim 2, characterized in that, The step of determining the target subgraph corresponding to the target instruction information based on the seed node and the target domain knowledge graph includes: Starting from the seed node, multiple initial subgraphs are searched from the target domain knowledge graph based on a bidirectional breadth-first search strategy; Based on the correlation score between each initial subgraph, the initial subgraphs are spliced ​​together to obtain the target subgraph corresponding to the target instruction information.

4. The method according to claim 1, characterized in that, The step of generating multiple candidate semantic paths and the information entropy of each semantic candidate path based on the target subgraph and the target domain knowledge graph includes: Obtain the node information of each Abox entity node in the target subgraph, wherein the node information includes entity label, entity type and entity attributes; Based on the node information of each Abox entity node in the target subgraph and the connection relationship of each Abox entity node in the target subgraph, determine the natural language relationship description corresponding to the target subgraph; Based on the Abox entity nodes included in the target subgraph and the target domain knowledge graph, determine the Tbox concept nodes corresponding to each Abox entity node and the descriptive logic axioms between Abox entity nodes and Tbox concept nodes; Based on the descriptive logic axioms between the Abox entity nodes and the Tbox concept nodes, the natural language constraints are determined. Based on the natural language relation description and the natural language constraints, multiple candidate semantic paths and the information entropy of each candidate semantic path are generated.

5. The method according to claim 4, characterized in that, The step of determining the Tbox concept node corresponding to each Abox entity node and the descriptive logical axiom between Abox entity nodes and Tbox concept nodes based on the Abox entity nodes included in the target subgraph and the target domain knowledge graph includes: Based on the Abox entity nodes included in the target subgraph, determine the Tbox concept nodes corresponding to the Abox entity nodes; Based on the Tbox concept node and the target domain knowledge graph, the descriptive logic axioms between the Tbox concept node and the Abox entity node are determined.

6. The method according to claim 1, characterized in that, The step of determining the target semantic path based on each candidate semantic path and the information entropy corresponding to each candidate semantic path includes: The semantic ambiguity score is determined based on the information entropy corresponding to each candidate semantic path. If the semantic ambiguity score is greater than or equal to the preset semantic ambiguity score, the target semantic path is determined in response to receiving the instruction information submitted by the target object.

7. The method according to claim 6, characterized in that, The method further includes: If the semantic ambiguity score is less than the preset semantic ambiguity score, the candidate semantic path with the largest information entropy value is selected as the target semantic path based on the information entropy corresponding to each candidate semantic path.

8. An ETL semantic ambiguity resolution device, characterized in that, include: The target subgraph determination module is used to determine the target subgraph corresponding to the target instruction information based on the target instruction information submitted by the target object and the target domain knowledge graph; The information determination module is used to generate multiple candidate semantic paths and the information entropy of each semantic candidate path based on the target subgraph and the target domain knowledge graph. The target semantic path determination module is used to determine the target semantic path based on each of the candidate semantic paths and the information entropy corresponding to each of the candidate semantic paths.

9. A computer device, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 7.

Images