Data processing method, apparatus, device, medium, and program product

Abstract

This application provides a data processing method applicable to the fields of artificial intelligence, digital human, and fintech, specifically involving the application of large-scale models in fintech. The method includes: receiving a natural language request input by a user through a digital human interface; parsing the natural language request using a first large-scale model in a large-scale model set to obtain the user intent and business entity corresponding to the natural language request; generating an executable task chain based on the user intent and the business entity using a second large-scale model in the large-scale model set; and executing the executable task chain by calling a pre-packaged data processing service corresponding to the executable task chain to obtain an execution result. This application also provides a data processing apparatus, device, storage medium, and program product.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence, digital human and financial technology, and in particular to the application of large models in the field of financial technology, specifically to a data processing method, apparatus, device, medium and program product. Background Technology

[0002] With the development of the banking service industry, the information retrieval methods of bank employees for the large amount of data stored in the bank have become more diversified and complex in order to meet different marketing methods and product promotion methods. However, the traditional data retrieval process relies heavily on the support of the technology department and the functions of the basic data management system of each bank.

[0003] With the progress of digital transformation, the data platform system of digital banks has been able to basically meet the data-driven intelligence needs of headquarters and branches. However, as the digital transformation work continues to deepen, the contradiction between the inclusive data needs within banks and the limited data productivity is becoming increasingly prominent. In other words, the contradiction between the data needs of ordinary users in their business operations and the insufficient data tools produced by the bank's digital departments to meet these needs is becoming increasingly apparent, and there is still a certain gap from the goal of empowering branches to use data efficiently and with low barriers to entry. At present, grassroots users within banks still face problems such as "difficulty" in data retrieval, "cumbersome" data extraction, and "high barriers to entry." For example, with the exponential growth of data scale, traditional retrieval models are unable to quickly and accurately locate the required data information, affecting the efficient flow and value presentation of data. Another example is that the development of personalized data retrieval models is highly dependent on the support of the technology department, but the traditional data extraction process requires multi-stage collaboration, with a response cycle of several weeks and relatively low response efficiency. Furthermore, the data analysis capabilities of different branches vary, grassroots business personnel lack technical reserves, and there is a lack of low-barrier data analysis tools. Traditional methods can no longer meet the urgent needs of business personnel to quickly conduct data analysis and make decisions. Summary of the Invention

[0004] In view of the above problems, this application provides data processing methods, apparatus, devices, media and program products.

[0005] According to a first aspect of this application, a data processing method is provided, the method comprising: receiving a natural language request input by a user through a digital human interface; parsing the natural language request through a first large model in a large model set to obtain a user intent and a business entity corresponding to the natural language request; generating an executable task chain based on the user intent and the business entity using a second large model in the large model set; and calling a pre-encapsulated data processing service corresponding to the executable task chain to execute the executable task chain and obtain an execution result.

[0006] According to an embodiment of this application, the step of parsing the natural language request using a first large model in a large model set to obtain the user intent and business entity corresponding to the natural language request includes: parsing the natural language request using the first large model; retrieving metadata and business rule data associated with the parsed natural language request from a data middleware; and, based on the retrieved metadata and business rule data, identifying the natural language request using a third large model in the large model set to obtain the user intent and the business entity corresponding to the natural language request, wherein the business entity includes data tags and definition rules corresponding to the user intent.

[0007] According to an embodiment of this application, the step of generating an executable task chain based on the user intent and the business entity using the second largest model in the large model set includes: generating multiple structured query statements using the second largest model in the large model set based on the business entity and the user intent; and combining the multiple structured query statements to form the executable task chain based on the combination logic corresponding to the data tags and definition rules of the business entity.

[0008] According to an embodiment of this application, the step of generating an executable task chain based on the user intent and the business entity using the second largest model in the large model set further includes: performing syntax rule verification on the multiple structured query statements respectively; and performing user permission verification on the multiple structured query statements respectively, wherein the user permission verification is performed by calling a pre-configured permission approval service through the third largest model in the large model set.

[0009] According to an embodiment of this application, the user permission verification of the structured query statement includes: determining a target data table based on the data tag of the business entity corresponding to the structured query statement; and verifying the access rights of the user who initiated the natural language request on the target data table.

[0010] According to an embodiment of this application, the executable task chain includes multiple tasks. The step of calling a pre-packaged data processing service corresponding to the executable task chain to execute the executable task chain and obtain an execution result includes: parsing the executable task chain using the third major model to obtain the task type corresponding to each task and the dependencies between the task types; determining the execution sequence of each task in the executable task chain based on the task type and the dependencies between the task types using the third major model; and, based on the execution sequence, sequentially calling the data processing service corresponding to each task using the third major model to execute each task and obtain the execution result.

[0011] According to an embodiment of this application, the method further includes: using the first large model to convert the execution result data into result data in the form of a chart; performing desensitization processing on the personal information in the result data to obtain desensitized result data; integrating the desensitized result data into a preset business report template, and sending the business report template to the user terminal in the form of natural language through the digital human interface.

[0012] According to an embodiment of this application, the method for determining the combination logic corresponding to the data tag and definition rule of the business entity includes: matching the data tag with a pre-configured tag rule library in the data platform to determine the combination rule associated with the data tag in the tag rule library, wherein the combination rule is used to define the processing method of the data tag; and determining the combination logic for combining the data tag and the definition rule according to the definition rule and the combination rule.

[0013] According to an embodiment of this application, the step of identifying the natural language request using the intent understanding service in the large model set based on the retrieved metadata and business rule data, and obtaining the user intent and the business entity corresponding to the natural language request, specifically includes: dynamically selecting a local large model or a cloud large model in the large model set to identify the natural language request based on the retrieved metadata and business rule data and based on the complexity of the natural language request, and obtaining the user intent and the business entity corresponding to the natural language request.

[0014] According to an embodiment of this application, the method further includes: inputting the corresponding natural language request, the business entity, the user intent, and the structured query statement into the case library of the data platform to update the case library.

[0015] A second aspect of this application provides a data processing apparatus, comprising: a first acquisition module configured to receive a natural language request input by a user through a digital human interface; a first processing module configured to parse the natural language request using a first large model in a large model set to obtain a user intent and a business entity corresponding to the natural language request; a second processing module configured to generate an executable task chain based on the user intent and the business entity using a second large model in the large model set; and a first execution module configured to call a pre-encapsulated data processing service corresponding to the executable task chain to execute the executable task chain and obtain an execution result.

[0016] A third aspect of this application provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0017] A fourth aspect of this application also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0018] The fifth aspect of this application also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method. Attached Figure Description

[0019] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0020] Figure 1 The illustrations depict application scenarios of data processing methods, apparatuses, devices, media, and program products according to embodiments of this application.

[0021] Figure 2 A flowchart illustrating a data processing method according to an embodiment of this application is shown schematically.

[0022] Figure 3 A flowchart illustrating step S200 of the data processing method according to an embodiment of this application is shown schematically.

[0023] Figure 4 A flowchart illustrating step S300 of a data processing method according to an embodiment of this application is shown schematically.

[0024] Figure 5 A supplementary flowchart illustrating step S300 of the data processing method according to an embodiment of this application is shown schematically;

[0025] Figure 6 A flowchart illustrating step S302 of the data processing method according to an embodiment of this application is shown schematically.

[0026] Figure 7 A flowchart illustrating step S400 of a data processing method according to an embodiment of this application is shown schematically.

[0027] Figure 8 A supplementary flowchart illustrating a data processing method according to an embodiment of this application is shown schematically;

[0028] Figure 9The flowchart illustrates a method for determining the combination logic corresponding to the data tags and definition rules of the business entities in the data processing method according to an embodiment of this application.

[0029] Figure 10 This schematic diagram illustrates a structural block diagram of a data processing apparatus according to an embodiment of the present application;

[0030] Figure 11 A first supplementary structural block diagram of a data processing apparatus according to an embodiment of this application is schematically shown;

[0031] Figure 12 A block diagram schematically illustrates an electronic device suitable for implementing a data processing method according to an embodiment of this application; and

[0032] Figure 13 The diagram illustrates a framework suitable for implementing a data processing method according to an embodiment of this application. Detailed Implementation

[0033] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0034] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0035] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0036] When using expressions such as "at least one of A, B, and C," the meaning should generally be interpreted according to the understanding of someone skilled in the art. For example, "a system having at least one of A, B, and C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C. Similarly, when using expressions such as "at least one of A, B, or C," the meaning should generally be interpreted according to the understanding of someone skilled in the art. For example, "a system having at least one of A, B, or C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C.

[0037] The data processing methods, apparatus, devices, media, and program products of this application can be used in the fields of artificial intelligence, digital human, and fintech, particularly in the application of large-scale models in fintech. They can also be used in any field other than finance, and the application areas of the data processing methods, apparatus, devices, media, and program products of this application are not limited.

[0038] In the technical solution of this application, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, application, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0039] In scenarios involving automated decision-making using personal information, the methods, devices, and systems provided in this application all offer users corresponding entry points for choosing to agree to or reject the automated decision-making results. If the user chooses to reject, the process proceeds to the expert decision-making stage. Here, "automated decision-making" refers to the activity of automatically analyzing and evaluating an individual's behavioral habits, interests, or economic, health, and credit status through computer programs, and then making a decision. Here, "expert decision-making" refers to the activity of making decisions by personnel who specialize in a particular field, possess specialized experience, knowledge, and skills, and have reached a certain level of professional expertise.

[0040] It's understandable that a data platform can be divided into three layers from bottom to top: the source layer, the aggregation layer, and the extraction layer. The source layer primarily consists of basic data that has been compliantly and systematically incorporated into the data lake through source integration, providing all users with the ability to integrate and share internal and external enterprise data. The aggregation layer refers to a data area that, based on the basic data, has been organized, filtered, processed, and refined to form data information with certain business significance and universality. The extraction layer can be divided into general-purpose shared data and domain-specific shared data. General-purpose data refers to general information built based on cross-professional and cross-domain common dimensions, including customer themes, funding themes, and channel themes. Domain-specific data refers to data processed from basic data, oriented towards specific domain applications, but possessing bank-wide reusability. The data asset types in the extraction layer can include: customer tags, which are indicators or variables produced and processed at the customer granularity level for the purpose of understanding customers and supporting business operations, reflecting customer attributes and characteristics. Extraction layer data asset types can also include: statistical indicators: statistical indicators are concepts that describe the overall comprehensive quantitative characteristics. A complete indicator generally consists of two parts: the indicator name and the indicator value. It reflects the characteristics of both the substance and the quantity of the matter.

[0041] The embodiments of this application provide a data processing method. Figure 1 The illustrations depict application scenarios of data processing methods, apparatuses, devices, media, and program products according to embodiments of this application.

[0042] like Figure 1 As shown, network 104 is a medium used to provide a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0043] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as financial service applications, web browser applications, email clients, social media platform software, etc.

[0044] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0045] Server 105 can be a server that provides various services, such as a backend management server that supports information browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0046] It should be noted that, in the first aspect, the data processing method provided in the embodiments of this application can be executed by server 105; in the second aspect, the data processing method provided in the embodiments of this application can be executed by first terminal device 101, second terminal device 102, or third terminal device 103. Correspondingly, in the third aspect, the data processing apparatus provided in the embodiments of this application can be disposed in server 105; in the fourth aspect, the data processing apparatus provided in the embodiments of this application can be disposed in first terminal device 101, second terminal device 102, or third terminal device 103. The data processing method provided in the embodiments of this application can also be executed by a server or server cluster that is different from server 105 and capable of communicating with first terminal device 101, second terminal device 102, third terminal device 103, and / or server 105. Correspondingly, the data processing apparatus provided in the embodiments of this application can also be disposed in a server or server cluster that is different from server 105 and capable of communicating with first terminal device 101, second terminal device 102, third terminal device 103, and / or server 105.

[0047] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0048] The following will be based on Figure 1 The described scene, through Figures 2-9 The data processing method of the application embodiments will be described in detail.

[0049] Figure 2 A flowchart illustrating a data processing method according to an embodiment of this application is shown.

[0050] like Figure 2 As shown, the data processing method of this embodiment includes operations S100 to S400.

[0051] In operation S100, natural language requests input by the user are received through the digital human interface. In some embodiments, the digital human interface located at layer 1 provides natural language interaction capabilities. Users input their data processing needs in a conversational manner through this interface. The digital human interface receives natural language instructions input by business personnel, including needs for data queries, customer group analysis, or report generation. Efficient input is achieved through conversational interaction, facilitating business personnel to express their needs. For example, the user inputs a first business objective description statement, such as a request to activate customer group screening, or a second data query request statement. For instance, the user inputs a natural language request through the digital human interface to "screen customer groups with financial potential but low mobile banking activity." The digital human interface converts the voice or text request into structured input data.

[0052] In operation S200, the natural language request is parsed using the first major model in the large model set to obtain the user intent and business entity corresponding to the natural language request. In some embodiments, the first major model located at layer N can be a natural language processing model, which initially understands the core objective of the user request and the key business objects involved, that is, it maps the user request to an identifiable operation scenario type and key business parameters through semantic parsing. In some embodiments, the first major model serves as the core of natural language processing to parse the request content to identify the scenario type and achieve semantic analysis. For example, after parsing, it obtains the first user intent, such as a data query scenario, and the second business entity, such as customer tag definition rules. For example, the first major model deconstructs the user request into the intent of "customer segmentation" and business entities such as "financial management preferences" and "mobile banking activity," wherein the business entities are associated with the customer topic tag definition rules of the data platform.

[0053] like Figure 3 As shown, in some other exemplary embodiments, the step of parsing the natural language request through the first large model in the large model set to obtain the user intent and business entity corresponding to the natural language request includes: operations S210 to S230.

[0054] In operation S210, the first large model is used to parse the natural language request. In some embodiments, the first large model located at layer N initially decomposes the request content to extract key elements. For example, when a user inputs "filter customers with inactive mobile banking transactions but financial needs", the model extracts the core verb "filter" and the entities "mobile banking transactions" and "financial needs". This parsing process simultaneously identifies the business scenario as belonging to the customer group selection type.

[0055] In operation S220, metadata and business rule data associated with the parsed natural language request are retrieved from the data platform. In some embodiments, based on the preliminary parsing results of the first large model, relevant data table structure information and field definitions are retrieved from the data asset catalog of the data platform, and relevant indicator calculation rules, tag definition descriptions, and business constraints are retrieved from the business knowledge base. For example, the data asset catalog of the data platform is queried in real time to obtain the basic table structure located in the source layer, the customer tag definitions located in the extraction layer (such as the threshold rule for the "inactive mobile banking customer" tag), the statistical indicator caliber located in the aggregation layer, and the compliance constraint library composed of the bank's data security management methods.

[0056] In operation S230, based on the retrieved metadata and business rule data, the third major model in the large model set is used to identify the natural language request, obtaining the user intent and the business entity corresponding to the natural language request. The business entity includes data tags and definition rules corresponding to the user intent. In some embodiments, the third major model located at layer N integrates the preliminary parsing results, retrieved metadata, and business rule data for deep semantic understanding to accurately identify the user's specific operational intent (such as data querying, customer group selection, or report generation) and the specific business entity information upon which this intent depends. The business entity information explicitly includes the required data tags and their calculation or definition rules. For example, the third major model integrates the retrieved metadata and business rule data to enhance semantic understanding. For instance, it maps "inactive customer" to a customer tag entity in the extraction layer that has "≤x transactions in the last x months," while simultaneously associating the tag with its storage location and permission constraints in the data platform. By dynamically retrieving metadata and business rule data, the accuracy of semantic understanding is enhanced, resolving natural language ambiguity issues.

[0057] In some embodiments, the step of identifying the natural language request using the intent understanding service in the large model set based on the retrieved metadata and business rule data, and obtaining the user intent and the business entity corresponding to the natural language request, specifically includes: dynamically selecting a local large model or a cloud-based large model in the large model set to identify the natural language request based on the retrieved metadata and business rule data and the complexity of the natural language request, and obtaining the user intent and the business entity corresponding to the natural language request. That is, for requests with relatively simple semantics and high response speed requirements, the lightweight large model deployed locally is preferentially called for identification; for requests involving complex semantic understanding and requiring stronger cognitive capabilities, the more powerful large model deployed in the cloud is called for identification, thereby achieving a balance between response speed and understanding accuracy. For example, simple query scenarios (such as single-tag retrieval) call the local model to ensure response timeliness, while complex analysis scenarios (such as multi-tag combination customer acquisition) schedule the cloud-based large model to perform deep semantic parsing. For example, when a user requests a simple query (such as "query customer A's account balance"), the local lightweight model is selected. When a complex request is encountered (such as "query product penetration rate of key customer groups"), the system automatically switches to the cloud-based large model, making dynamic routing decisions based on the number of nested clauses or the complexity of entity relationships in the request.

[0058] In operation S300, based on the user intent and the business entity, an executable task chain is generated using the second largest model in the large model set. In some embodiments, the second largest model located at layer N focuses on transforming the understood user intent and the business entity into a specific, executable sequence of data processing instructions. This executable task chain typically manifests as a structured set of database query statements. For example, when the user intent is customer segmentation, the second largest model generates an executable task chain that includes data querying, tag calculation, and segmentation operations.

[0059] like Figure 4 As shown, in some other exemplary embodiments, the step of generating an executable task chain based on the user intent and the business entity using the second largest model in the large model set includes: operations S310 to S320.

[0060] Specifically, in operation S310, based on the business entity and the user intent, multiple structured query statements are generated using the second major model in the large model set. In some embodiments, this means generating basic data operation instructions by identifying precise user intent (such as data query or customer segmentation) and specific business entity information (including required data tags and their definition rules). For requests involving complex business needs, multiple related query fragments may need to be generated. In some embodiments, the second major model is a dedicated model fine-tuned with domain data, capable of accurately understanding the meaning and definition rules of data tags in business entities and converting them into valid query conditions, that is, converting user intent and business entities into standard query syntax. For example, after understanding the intent to "filter inactive mobile banking customers holding wealth management products," tag calculation statements are generated as follows: [Query inactive mobile banking users (with ≤ x transactions in the last x months)], asset holding query statement (filter users holding wealth management / fund products), and behavioral feature filtering statement (filter users with login behavior in the last x months). Each structured query statement corresponds to data asset operations at different levels of the data platform, such as accessing the source layer basic table or calculating the extraction layer tag.

[0061] Specifically, in operation S320, based on the data tags of the business entities and the combination logic corresponding to the defined rules, multiple structured query statements are combined to form the executable task chain. In some embodiments, based on the predefined relationship rules between data tags in the business entities and the definition constraints of the tags themselves, it is determined how to logically combine multiple generated independent structured query statements to form a complete operation sequence that can achieve the user's final intention. For example, the generated tag calculation statement [Query inactive mobile banking users (no more than x transactions in the last x months)], asset holding query statement (filter users holding wealth management / fund products), and behavior feature filtering statement (filter users with login behavior in the last x months) are combined by an AND relationship (must simultaneously satisfy inactive, holding wealth management, and recent login), and a grouping operation is added: GROUP BY customer_id HAVING COUNT(behavior) > x, thereby generating an executable task chain: [tag calculation] → [asset query] → [behavior filtering] → [customer grouping], that is, generating a structured query chain based on the logical combination of business entity tags, realizing the automated construction of the task chain.

[0062] like Figure 4 and Figure 13As shown, in the process example from operation S310 to operation S320, when a user submits a natural language request to "analyze high-potential customer characteristics and output a segmentation report" through the digital human interface at layer 1, the first major model at layer N first parses the user's intent as customer segmentation analysis and extracts key business entities such as "potential characteristic tags" (corresponding to the tag definition rules of customer topics in the data platform); subsequently, the second major model at layer N generates multiple structured query statements based on this intent and entities: the first structured query statement is used to extract basic customer attributes (such as age range, product ownership status), the second structured query statement is used to calculate behavioral characteristic indicators (such as login frequency, historical active period), and the third structured query statement is used to... The system incorporates business analysis dimensions (such as recent marketing response records). During this process, based on the data tag combination logic of business entities (e.g., "potential customer = basic attributes ∩ behavioral characteristics ∩ marketing sensitivity"), it constructs logical relationships between query statements, forming an executable task chain of "basic attribute filtering → behavioral characteristic calculation → cluster result output." Simultaneously, traditional functions at layer X (permission approval service) are dynamically invoked to perform real-time authentication on query fields involving customer privacy data. Ultimately, the task chain is constructed within a three-layer framework of 1 (natural language interaction) + N (large model intelligent scheduling) + X (traditional authentication module). The execution of the data query solution is achieved through conversational interaction using natural language within this three-layer framework, completely breaking down data barriers.

[0063] like Figure 5 As shown, in some other exemplary embodiments, the step of generating an executable task chain based on the user intent and the business entity using the second largest model in the large model set further includes: operations S301 to S302.

[0064] Specifically, in operation S301, the syntax rules of each of the multiple structured query statements are validated. In some embodiments, each structured query statement generated by the second large model undergoes syntax and semantic compliance checks to ensure it conforms to the syntax specifications of the target database system, thereby preventing execution failure due to syntax errors. In some embodiments, a large model automatic regeneration mechanism is triggered for statements that fail validation. For example, the structured query statement is first validated to ensure that the table join syntax conforms to the database specifications, and then the use of the structured query statement is checked to ensure it matches the target data source structure.

[0065] Specifically, in operation S302, user permission verification is performed on each of the multiple structured query statements. This user permission verification is executed by invoking a pre-configured permission approval service. In some embodiments, for each data resource that the structured query statement intends to access, the system verifies whether the current user has the corresponding operation permissions. This is achieved through a dual protection mechanism of syntax rule validation and permission verification, ensuring the legality and security of the task chain.

[0066] like Figure 6 As shown, in some other exemplary embodiments, performing user permission verification on each of the multiple structured query statements includes: operations S3021 to S3022.

[0067] Specifically, in operation S3021, the target data table is determined based on the data tags of the business entity corresponding to the structured query statement. In some embodiments, by analyzing the data tags referenced in the structured query statement and combining the metadata mapping relationship of the data platform (the association between tags and underlying data tables), the specific database table that the query statement actually needs to access is accurately located.

[0068] Specifically, in operation S3022, the access permissions of the user initiating the natural language request on the target data table are verified. In some embodiments, based on the determined target database table, the user's permission configuration is queried to confirm whether the user is authorized to read or operate data in the target data table, thereby ensuring the security of data access. In some embodiments, the permission approval service configured in the enterprise's traditional functions is invoked to verify the matching between the user and the data security level according to the industry's data security management methods, and unauthorized operations are blocked in real time. For example, the read permission of the user ID on the target data table (such as the product holding table) is verified. If not authorized, an automatic application process is triggered, that is, permission verification is associated with a specific data table to achieve dynamic authentication.

[0069] In operation S400, a pre-encapsulated data processing service corresponding to the executable task chain is invoked to execute the executable task chain and obtain the execution result. For example, the executable task chain is ultimately transformed into actual execution behavior on the underlying data source (such as source layer data, aggregation layer data, or extraction layer data). For instance, the executable task chain for customer segmentation sequentially executes SQL queries, customer segmentation, and profile generation operations. This avoids the inefficiency caused by the need for customized development by enterprise technical personnel for personalized data query functions. Through natural language conversational interaction, it completely breaks down data usage barriers, reduces the time consumption of traditional data queries, and enables frontline business personnel to improve their ability to use data independently, achieving data democratization.

[0070] like Figure 7As shown, in some other exemplary embodiments, the executable task chain includes multiple tasks, and the step of calling a pre-encapsulated data processing service corresponding to the executable task chain to execute the executable task chain and obtain the execution result includes: operations S410 to S430.

[0071] In operation S410, the third major model is used to parse the executable task chain to obtain the task type corresponding to each task and the dependencies between the task types. In some embodiments, the third major model located at layer N is used to analyze the structure of the executable task chain, identifying the operational nature of each independent task (e.g., data query, data filtering, data aggregation, or tag calculation) and the logical sequence or data dependencies between tasks. For example, parsing the executable task chain may reveal that it includes a first data query task, a second analysis task, and a third visualization task, identifying that the second analysis task depends on the output result of the first data query task, and the third visualization task depends on the completion status of the first two tasks.

[0072] In operation S420, based on the task types and the dependencies between them, the execution sequence of each task in the executable task chain is determined using the third major model. In some embodiments, based on the identified task dependencies, a reasonable task execution order is planned using the third major model located at layer N, ensuring that the output generated by the preceding task can correctly meet the input requirements of the subsequent task. In some embodiments, tasks without dependencies are scheduled in parallel, and sequential execution queues are generated for tasks with strong dependencies.

[0073] In operation S430, based on the execution sequence, the third major model sequentially calls the data processing services corresponding to each task to execute each task and obtain the execution result. In some embodiments, the third major model located at layer N uses the planned execution sequence to trigger pre-encapsulated data processing services (such as dedicated services for different data layers or different operation types) one by one to execute specific tasks, and transmits or combines the execution results of each task as needed to finally form complete execution result data. That is, it realizes the parsing of task dependencies and dynamic orchestration of execution sequences, solving the problems of correct logic and efficiency in task scheduling. In some embodiments, the event-driven architecture asynchronously triggers the execution of data processing services, while the digital human interface event communicates asynchronously with the driving architecture. That is, it realizes non-blocking execution of request submission and result return (the user does not need to wait for all data processing services to complete), which is used to solve the problems of high concurrency and long-process task blocking.

[0074] like Figure 8 As shown, the data processing method in this embodiment further includes operations S500 to S700.

[0075] In operation S500, the first large model is used to convert the execution result data into result data in the form of charts.

[0076] In operation S600, the personal information in the result data is de-identified to obtain de-identified result data.

[0077] During operation S700, the de-identified result data is integrated into a preset business report template, and the business report template is converted into natural language form through the digital human interface and sent to the user terminal.

[0078] In some embodiments, specifically in operation S500, for example, the first large model located at layer N analyzes the structure and content of the execution result data, automatically selects and generates appropriate visualization charts (such as bar charts, line charts, or pie charts) to intuitively display key business indicators and insights. For example, the first large model includes a visualization generation model that maps structured query results into graphical elements such as pie charts and trend lines, such as converting customer age distribution into a demographic pie chart.

[0079] In some embodiments, specifically in operation S600, for example, sensitive personal information (such as customer names, contact information, or identity identifiers) identified in the execution result data is masked, replaced, or generalized according to preset privacy protection rules to ensure data security during display and use. For example, sensitive fields such as ID card numbers and mobile phone numbers are matched using regular expressions and masked and replaced, while highly sensitive information is dynamically obfuscated according to data security level policies.

[0080] In some embodiments, specifically in operation S700, for example, the anonymized result data is automatically populated into the corresponding positions of a predefined business report template. Then, the first main model at layer N transmits the report template content, including charts and key data, to the digital human interface at layer 1, which converts it into easily understandable natural language descriptions and provides feedback to the user in the form of dialogues or messages for easy customer comprehension. For example, the first main model at layer N automatically populates the anonymized charts and analysis conclusions into the branch marketing activity report template, and generates a natural language conclusion containing "A1-A2 year old customers with a high demand for financial products" through the digital human interface at layer 1 and returns it to the user terminal. For example, the anonymized customer group percentage data is populated into the "Market Expansion" section of the business report template, and a natural language description is generated through the digital human interface: "The target customer group size has reached XXX level, of which the second type of characteristic users account for a large proportion," and a dynamic chart is attached and sent to the user terminal.

[0081] like Figure 9As shown, in some exemplary embodiments, the method for determining the combination logic corresponding to the data tags and definition rules of the business entity includes operations S010 to S020.

[0082] In operation S010, based on the data tag, a pre-configured tag rule base in the data platform is matched to determine the combination rules associated with the data tag in the tag rule base. These combination rules define the processing method for the data tag. In some embodiments, the tag rule base stores the association relationships, calculation constraints, or filtering condition constraints between various data tags predefined by business experts. For example, when a user request involves multiple data tags, the allowed combination methods between these tags and their corresponding processing logic (such as how to calculate or filter) are obtained by matching the rule base.

[0083] In operation S020, based on the definition rules and the combination rules, combination logic for combining the data tags with the definition rules is determined. In some embodiments, the definition rules carried by the data tags themselves (such as the tag calculation formula or filtering conditions) are integrated with the acquired combination rules (such as the association constraints between tags). This integration determines how to connect the definition rules of each tag according to the requirements of the combination rules, through logical operators (such as AND, OR) or set execution priorities, to form the complete combination logic finally applied to data processing.

[0084] In some embodiments, the tag rule base, as part of the data platform's metadata management, stores predefined logical relationships, operation rules, or constraints between different data tags. For example, when a user request involves multiple data tags, the rule base retrieves the allowed combinations of these tags (such as "AND," "OR," and "NOT" logic) and the calculation methods for specific tags.

[0085] like Figure 13 As shown, in some exemplary embodiments, the corresponding natural language requests, business entities, user intents, and structured query statements are input into the case library of the data platform to update the case library. In some embodiments, a successfully processed user request, its corresponding intermediate processing results (i.e., the identified business entity and user intent), and the finally generated executable code (structured query statement) are stored as a complete case. In this embodiment, the case library is used for subsequent model optimization training or intelligent recommendation reference, and can also optimize the tag recommendation efficiency of subsequent similar requests. For example, when a user initiates a natural language request to query a specific customer group in a certain region, the finally generated SQL statement, the identified relevant tags, and intents will be associated and stored in the case library.

[0086] In this embodiment, the natural language request sent by the client can also be executed on the system side through the operations S100~S700, S210, S220, S230, S310, S320, S301, S302, S3021, S3022, S410, S420, S430, S010, and S020 described above, which will not be repeated here.

[0087] Based on the above data processing method, this application also provides a data processing apparatus, which will be described below in conjunction with... Figure 10 The device is described in detail.

[0088] Figure 10 A schematic block diagram of a data processing apparatus according to an embodiment of this application is shown.

[0089] like Figure 10 As shown, the data processing device 200 in this embodiment includes a first acquisition module 210, a first processing module 220, a second processing module 230, and a first execution module 240.

[0090] The first acquisition module 210 is configured to execute a natural language request to receive user input via a digital human interface. In one embodiment, the first acquisition module 210 can be used to execute the operation S100 described above, which will not be repeated here.

[0091] The first processing module 220 is configured to parse the natural language request using the first large model in the large model set to obtain the user intent and business entity corresponding to the natural language request. In one embodiment, the first processing module 220 can be used to execute the operation S200 described above, which will not be repeated here. Simultaneously, the first processing module 220 can also be used to execute the operations S210, S220, and S230 described above, which will not be repeated here.

[0092] The second processing module 230 is configured to generate an executable task chain based on the user intent and the business entity using the second largest model in the large model set. In one embodiment, the second processing module 230 can be used to execute the operation S300 described above. Simultaneously, the second processing module 230 can also be used to execute the operations S310, S320, S301, S302, S3021, S3022, S010, and S020 described above, which will not be elaborated further here.

[0093] The first execution module 240 is configured to execute the executable task chain by calling a pre-encapsulated data processing service corresponding to the executable task chain, and obtain the execution result. In one embodiment, the first execution module 240 can be used to execute the operation S400 described above. At the same time, the first execution module 240 can also be used to execute the operations S410, S420 and S430 described above, which will not be described again here.

[0094] like Figure 11 As shown, in one specific embodiment, the data processing device further includes: a third processing module 250, a fourth processing module 260, and a first transmission module 270.

[0095] The third processing module 250 is configured to use the first large model to convert the execution result data into result data in the form of a chart; in one embodiment, the third processing module 250 can be used to perform the operation S500 described above, which will not be repeated here.

[0096] The fourth processing module 260 is configured to perform desensitization processing on the personal information in the result data to obtain desensitized result data; in one embodiment, the fourth processing module 260 can be used to perform the operation S600 described above, which will not be repeated here.

[0097] The first transmission module 270 is configured to integrate the de-identified result data into a preset business report template, and convert the business report template into natural language form through the digital human interface and send it to the user terminal. In one embodiment, the first transmission module 270 can be used to perform the operation S700 described above, which will not be repeated here.

[0098] According to embodiments of this application, any multiple modules among the first acquisition module 210, first processing module 220, second processing module 230, first execution module 240, third processing module 250, fourth processing module 260, and first transmission module 270 can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the first acquisition module 210, first processing module 220, second processing module 230, first execution module 240, third processing module 250, fourth processing module 260, and first transmission module 270 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the first acquisition module 210, the first processing module 220, the second processing module 230, the first execution module 240, the third processing module 250, the fourth processing module 260, and the first transmission module 270 may be implemented at least partially as a computer program module, which can perform corresponding functions when the computer program module is run.

[0099] Figure 12 A block diagram schematically illustrates an electronic device suitable for implementing a data processing method according to an embodiment of this application.

[0100] like Figure 12 As shown, an electronic device 900 according to an embodiment of this application includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage portion 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0101] RAM 903 stores various programs and data required for the operation of electronic device 900. Processor 901, ROM 902, and RAM 903 are interconnected via bus 904. Processor 901 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 902 and / or RAM 903. It should be noted that programs may also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in one or more memories.

[0102] According to embodiments of this application, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to a bus 904. The electronic device 900 may also include one or more of the following components connected to the input / output (I / O) interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 910 as needed so that computer programs read from it can be installed into the storage section 908 as needed.

[0103] This application also provides a computer-readable storage medium, which may be included in the apparatus described in the above embodiments; or it may exist independently and not assembled into the apparatus. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0104] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 902 and / or RAM 903 and / or one or more memories other than ROM 902 and RAM 903 described above.

[0105] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to enable the computer system to implement the data processing methods provided in the embodiments of this application.

[0106] When the computer program is executed by the processor 901, it performs the functions defined in the apparatus of the embodiments of this application. According to the embodiments of this application, the apparatus and the like described above can be implemented by computer program modules.

[0107] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 909, and / or installed from a removable medium 911. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0108] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, it performs the functions defined in the system of the embodiments of this application. According to embodiments of this application, the apparatuses described above can be implemented by computer program modules.

[0109] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0110] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0111] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

[0112] The embodiments of this application have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of this application. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Without departing from the scope of this application, those skilled in the art can make various substitutions and modifications, all of which should fall within the scope of this application.

Claims

1. A data processing method, characterized in that, The method includes: Receive natural language requests from users via the digital human interface; The natural language request is parsed using the first major model in the large model set to obtain the user intent and business entity corresponding to the natural language request; Based on the user intent and the business entity, an executable task chain is generated using the second largest model in the large model set; The pre-encapsulated data processing service corresponding to the executable task chain is invoked to execute the executable task chain and obtain the execution result.

2. The method according to claim 1, characterized in that, The step of parsing the natural language request using the first large model in the large model set to obtain the user intent and business entity corresponding to the natural language request includes: The first large model is used to parse the natural language request; The data retrieved includes metadata and business rule data associated with the parsed natural language request in the data center. Based on the retrieved metadata and business rule data, the third major model in the large model set is used to identify the natural language request, and the user intent and the business entity corresponding to the natural language request are obtained. The business entity includes data tags and definition rules corresponding to the user intent.

3. The method according to claim 2, characterized in that, The step of generating an executable task chain based on the user intent and the business entity using the second largest model in the large model set includes: Based on the business entity and the user intent, multiple structured query statements are generated using the second largest model in the large model set; Based on the data tags of the business entities and the combination logic corresponding to the defined rules, multiple structured query statements are combined to form the executable task chain.

4. The method according to claim 3, characterized in that, The step of generating an executable task chain based on the user intent and the business entity using the second largest model in the large model set further includes: Syntax rule validation is performed on each of the structured query statements; and, User permission verification is performed on each of the multiple structured query statements, wherein the user permission verification is performed by calling a pre-configured permission approval service through the third major model in the large model set.

5. The method according to claim 4, characterized in that, The user permission verification of the structured query statement includes: Based on the data tags of the business entities corresponding to the structured query statement, the target data table is determined; Verify the user's access permissions on the target data table for the user who initiated the natural language request.

6. The method according to claim 3, characterized in that, The executable task chain includes multiple tasks. The step of calling a pre-encapsulated data processing service corresponding to the executable task chain to execute the executable task chain and obtain the execution result includes: The third major model is used to parse the executable task chain to obtain the task type corresponding to each task and the dependency relationship between each task type. Based on the task types and the dependencies between them, the execution sequence of each task in the executable task chain is determined by the third major model; Based on the execution sequence, the third major model sequentially calls the data processing services corresponding to each task to execute each task and obtain the execution results.

7. The method according to claim 3, characterized in that, The method further includes: The execution result data is converted into chart-style result data using the first major model; The personal information in the result data is anonymized to obtain anonymized result data; The anonymized data is integrated into a preset business report template, and the business report template is converted into natural language and sent to the user terminal through the digital human interface.

8. The method according to claim 3, characterized in that, The method for determining the combination logic corresponding to the data tags and definition rules of the business entities includes: Based on the data tag matching the pre-configured tag rule library in the data platform, determine the combined rules associated with the data tag in the tag rule library, wherein the combined rules are used to define the processing method of the data tag; Based on the defined rules and the combined rules, determine the combination logic for combining the data tags with the defined rules.

9. The method according to claim 2, characterized in that, The step of using the intent understanding service in the large model set to identify the natural language request based on the retrieved metadata and business rule data, and obtaining the user intent and business entity corresponding to the natural language request, specifically includes: Based on the retrieved metadata and business rule data, and based on the complexity of the natural language request, the local large model or cloud large model in the large model set is dynamically selected to identify the natural language request, thereby obtaining the user intent and the business entity corresponding to the natural language request.

10. The method according to claim 3, characterized in that, The method further includes: The corresponding natural language requests, business entities, user intents, and structured query statements are input into the case library of the data platform to update the case library.

11. A data processing apparatus, characterized in that, include: The first acquisition module is configured to execute natural language requests that receive user input through the digital human interface; The first processing module is configured to parse the natural language request through the first large model in the large model set to obtain the user intent and business entity corresponding to the natural language request; The second processing module is configured to generate an executable task chain based on the user intent and the business entity, using the second largest model in the large model set. The first execution module is configured to execute the executable task chain by calling a pre-encapsulated data processing service corresponding to the executable task chain, and obtain the execution result.

12. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 10.

14. A computer program product comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 10.

Images