Method and device for executing data analysis task, storage medium and electronic equipment
By performing instruction analysis and reconstruction on structured data and optimizing the data input of large language models, the problems of high resource overhead and low analytical capabilities are solved, achieving efficient data analysis.
Patent Information
- Application Number
- CN202510570707.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-30
- Publication Date
- 2025-09-05
AI Technical Summary
In the existing technology, when large language models perform structured data analysis tasks, resource consumption is high and analysis capabilities are reduced, resulting in low manual analysis efficiency.
By inputting structured data into multiple large language models, instruction analysis, data extraction and reconstruction are performed to remove unnecessary data, optimize data volume and retain key structured information.
It reduces resource overhead, improves information density and the accuracy of analysis results, and enhances the analytical capabilities of large language models.
Smart Images
Figure CN120596523A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to computer technology, and in particular to a method, device, storage medium and electronic device for performing data analysis tasks. Background Art
[0002] In some application scenarios, such as risk control scenarios, analysis tasks for structured data are often encountered. In the existing technology, the common method for such tasks is manual analysis. However, due to the limited attention of humans, manual analysis tasks are extremely inefficient when faced with massive amounts of data to be analyzed.
[0003] With the development of the times, large language models are increasingly being used for task analysis. Through autoregressive generative methods, large language models can respond to input textual data and commands. Existing techniques typically involve directly serializing the structured data to be analyzed into textual behavior, which is then combined with the corresponding analysis commands and fed into the large language model. Summary of the Invention
[0004] The purpose of the embodiments of this specification is to provide a method, device, storage medium, and electronic device for performing data analysis tasks.
[0005] An embodiment of this specification provides a method for performing a data analysis task, which can extract non-essential data from structured data by automatically extracting simplified data from the structured data. The method includes:
[0006] Inputting structured data to be analyzed and analysis instructions corresponding to the structured data into a first large language model, obtaining an instruction analysis result output by the first large language model, wherein the instruction analysis result is used to indicate a portion of the structured data required for the analysis instruction;
[0007] Data extraction is performed on the structured data according to the instruction analysis result to obtain simplified data, so as to execute a data analysis task on the serialized simplified data through the analysis instruction.
[0008] Furthermore, the method further comprises:
[0009] Inputting the structured data to be analyzed into the second language model to obtain a summary description text corresponding to the structured data output by the second language model;
[0010] The step of inputting the structured data to be analyzed and the analysis instructions corresponding to the structured data into the first large language model to obtain the instruction analysis results output by the first large language model includes:
[0011] The structured data, the summary description text, and the analysis instructions corresponding to the structured data are input into a first large language model to obtain an instruction analysis result output by the first large language model.
[0012] Furthermore, the method further comprises:
[0013] Inputting the structured data and the analysis instruction into a third language model to obtain a reconstruction analysis result corresponding to the structured data output by the third language model;
[0014] According to the reconstruction analysis result, a target reconstruction function is determined from a plurality of preset reconstruction functions, and the target reconstruction function is used to reconstruct the simplified data to obtain reconstructed data containing structured information, so as to perform a data analysis task on the serialized reconstructed data through the analysis instruction.
[0015] Furthermore, determining a target reconstruction function from a plurality of preset reconstruction functions according to the reconstruction analysis result includes:
[0016] Based on a plurality of preset reconstruction functions and an application scenario corresponding to each reconstruction function, a target application scenario is determined from the plurality of application scenarios according to the reconstruction analysis result, and the reconstruction function corresponding to the target application scenario is used as a target reconstruction function.
[0017] Furthermore, determining a target reconstruction function from a plurality of preset reconstruction functions according to the reconstruction analysis result includes:
[0018] Determining whether there is data missing in the simplified data according to the reconstruction analysis result;
[0019] If not, a target reconstruction function is determined from a plurality of preset reconstruction functions according to the reconstruction analysis result.
[0020] Furthermore, the method further comprises:
[0021] If it is determined that the simplified data has data missing, data extraction is performed again on the structured data according to the data missing analysis result corresponding to the simplified data and the instruction analysis result, wherein the data missing analysis result is used to indicate the part of the simplified data that has data missing.
[0022] The embodiments of this specification also provide an apparatus for performing a data analysis task, including:
[0023] a first obtaining module, configured to input structured data to be analyzed and an analysis instruction corresponding to the structured data into a first large language model, and obtain an instruction analysis result output by the first large language model, wherein the instruction analysis result indicates a portion of the structured data required for the analysis instruction;
[0024] The second obtaining module is used to extract data from the structured data according to the instruction analysis result to obtain simplified data, so as to perform a data analysis task on the serialized simplified data through the analysis instruction.
[0025] An embodiment of this specification further provides a storage medium, wherein the storage medium stores a computer program, and the computer program is suitable for being loaded by a processor and executing the steps of the above method.
[0026] An embodiment of this specification further provides an electronic device, comprising: a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the steps of the above method.
[0027] The embodiments of this specification also provide a computer program product having at least one instruction stored thereon, wherein the at least one instruction implements the steps of the above method when executed by a processor.
[0028] According to the solution of the embodiment of this specification, the structured data to be analyzed and the analysis instructions corresponding to the structured data are first input into the first large language model to obtain the instruction analysis result output by the first large language model, and the instruction analysis result is used to indicate the part of the structured data required by the analysis instruction. Then, data extraction is performed on the structured data according to the instruction analysis result to obtain simplified data, so as to perform data analysis tasks on the serialized simplified data through the analysis instruction. In this way, by extracting the simplified data required for the current analysis from the structured data, unnecessary data in the structured data can be removed, resource overhead can be reduced, and at the same time, information density can be improved, thereby improving the accuracy of the analysis results. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] Figure 1 A flowchart of a method for performing a data analysis task provided in an embodiment of this specification;
[0030] Figure 2 A schematic diagram of a process for extracting simplified data according to an example provided in the embodiments of this specification;
[0031] Figure 3 A schematic diagram of a process for data reconstruction provided as an example in an embodiment of this specification;
[0032] Figure 4 A schematic diagram of the structure of a device for performing data analysis tasks provided in an embodiment of this specification;
[0033] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this specification. DETAILED DESCRIPTION
[0034] To make the objectives, technical solutions, and advantages of this specification more clear, the following will clearly and completely describe the technical solutions of this specification in conjunction with the specific embodiments of this specification and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this specification.
[0035] See Figure 1 , is a flow chart of a method for performing a data analysis task provided in an embodiment of this specification. In an embodiment of this specification, the method for performing a data analysis task is applied to a device for performing a data analysis task (hereinafter referred to as a "task execution device") or an electronic device equipped with a task execution device. Figure 1 The process shown in FIG. 1 is described in detail. The method for performing a data analysis task may specifically include the following steps:
[0036] S102: Input the structured data to be analyzed and the analysis instructions corresponding to the structured data into the first large language model to obtain the instruction analysis result output by the first large language model, wherein the instruction analysis result is used to indicate the part of the structured data required by the analysis instruction.
[0037] In some embodiments, structured data refers to data stored in a predefined format and model. The representation of structured data includes, but is not limited to, tables and knowledge graphs. In some embodiments, structured data is tabular data (table-formatted data), which has a fixed length and each location has a specific meaning. For example, the tabular data modality corresponding to a user typically includes user identity information, user behavior information, etc.
[0038] In some embodiments, the analysis instruction represents an instruction for instructing the execution of an analysis task, for example, the analysis instruction is used to instruct the execution of a user risk analysis task, and for another example, the analysis instruction is used to indicate the characteristics of user secondary cohesion in an analysis graph; the analysis instruction can be a simple instruction or a complex analysis instruction.
[0039] In some embodiments, the instruction analysis result is used to indicate the analysis result obtained after analyzing the analysis instruction, and the instruction analysis result is used to indicate the part of the structured data that is actually required to execute the analysis instruction. For example, the structured data includes multiple nodes and edges for connecting the nodes, and the analysis instruction is used to instruct to perform risk analysis on node A. The first large language model outputs the corresponding instruction analysis result by analyzing the analysis instruction. The instruction analysis result indicates that the part of the structured data required to execute the analysis instruction is the data of node A and the data of nodes B and C directly connected to node A, that is, the data of other nodes other than nodes A, B, and C are non-essential data when executing the analysis instruction, that is, redundant data. It should be noted that the redundancy of structured data mainly lies in the part that is not required by the current analysis instruction. For example, a simple instruction usually only needs to pay attention to the data of one or two columns in the tabular structured data, and the other table parts are redundant. Based on the solution of the embodiment of this specification, it is possible to determine which part of the structured data is required to execute the analysis instruction and which part is not required to execute the analysis instruction by analyzing the analysis instruction.
[0040] S104 , extracting data from the structured data according to the instruction analysis result to obtain simplified data, so as to execute a data analysis task on the serialized simplified data through the analysis instruction.
[0041] In some embodiments, based on the instruction analysis result, part of the data required to execute the analysis instruction is extracted from the structured data, which is equivalent to removing the part of the structured data that is not required to execute the analysis instruction, thereby obtaining simplified data for executing the analysis instruction, and the amount of the simplified data is smaller than the amount of the structured data. As an example, assuming that the original structured data is D, the extracted Tabular data D ′ It can be expressed as:
[0042] D ′ =SelectData(LM(I),D)
[0043] Here, I is the analysis instruction input to the large language model, LM represents the large language model analysis, and SelectData is the automated extraction function.
[0044] In some embodiments, after obtaining the reduced data, the reduced data is serialized, and then the analysis instructions are used to perform data analysis tasks on the serialized reduced data to obtain an execution result. Experimental results show that reduced data extraction can significantly reduce the data size of large language model inputs, while reducing overhead while increasing information density and improving analysis accuracy.
[0045] The present application has found that in the prior art, when a large language model performs an analysis task, it usually directly serializes the structured data to be analyzed into text behavior, and then merges it with the corresponding analysis instructions and inputs it into the large language model. However, since the structured data is usually very long and contains a variety of repeated and redundant information, the solutions of the prior art will cause the token length of the large language model input to be too long, resulting in excessive resource overhead and a decrease in the analysis capability of the large language model.
[0046] According to the solution of the embodiment of this specification, the structured data to be analyzed and the analysis instructions corresponding to the structured data are first input into the first large language model to obtain the instruction analysis result output by the first large language model, and the instruction analysis result is used to indicate the part of the structured data required by the analysis instruction. Then, data extraction is performed on the structured data according to the instruction analysis result to obtain simplified data, so as to perform data analysis tasks on the serialized simplified data through the analysis instruction. In this way, by extracting the simplified data required for the current analysis from the structured data, unnecessary data in the structured data can be removed, resource overhead can be reduced, and at the same time, information density can be improved, thereby improving the accuracy of the analysis results.
[0047] In some embodiments, the method further includes: inputting the structured data to be analyzed into a second large language model, and obtaining a summary description text corresponding to the structured data output by the second large language model; wherein, inputting the structured data to be analyzed and the analysis instructions corresponding to the structured data into the first large language model, and obtaining the instruction analysis results output by the first large language model, includes: inputting the structured data, the summary description text, and the analysis instructions corresponding to the structured data into the first large language model, and obtaining the instruction analysis results output by the first large language model.
[0048] In some embodiments, the second large language model and the first large language model are the same large language model; in other embodiments, the second large language model and the first large language model are different large language models. In some embodiments, by inputting structured data into the second large language model, a data description for the structured data can be obtained, that is, a summary description text corresponding to the structured data, and then the summary description text, the structured data, and the analysis instruction to be executed can be input together into the first large language model for instruction analysis, and the instruction analysis result for the analysis instruction output by the first large language model can be obtained. It should be noted that by using a large language model to summarize and describe the structured data, and using it as auxiliary information input for instruction analysis of the analysis instruction, the instruction analysis effect can be effectively improved.
[0049] In some embodiments, the method further includes: inputting the structured data and the analysis instructions into a third language model to obtain a reconstruction analysis result corresponding to the structured data output by the third language model; determining a target reconstruction function from a plurality of preset reconstruction functions based on the reconstruction analysis result, and using the target reconstruction function to reconstruct the streamlined data to obtain reconstructed data containing structured information, so as to perform a data analysis task on the serialized reconstructed data through the analysis instructions.
[0050] In some embodiments, the third largest language model and the first largest language model / the second largest language model may be the same model or different models. In some embodiments, the third largest language model is used to perform reconstruction analysis on structured data for data reconstruction of simplified data. In some embodiments, the reconstruction analysis result is used to indicate how to reconstruct the structured information of the simplified data corresponding to the analysis instruction, for example, the reconstruction analysis result indicates adding edge information between nodes on the basis of the simplified data; the purpose of data reconstruction is to maintain certain structured information in the reconstructed data.
[0051] As an example, assume that the data to be reconstructed (ie, the simplified data) is D ′ , the reconstructed data D″ can be expressed as:
[0052] D″=SelectDataReconstruct(LM(I),D′)
[0053] Here, I represents the analysis instruction input to the large language model, LM represents large language model analysis, and SelectDataReconstruct represents the reconstruction function. Experiments have shown that reconstructing streamlined data can improve the large language model's ability to understand and analyze structured information.
[0054] The present application has discovered that after serialization, structured data will change from a structured representation to a serialized form, which will lead to the loss of structural information. The relationship that was originally obvious in the original structured data will become blurred after serialization, which will have a significant negative impact on the execution effect of the analysis instruction. According to the solution of the embodiment of this specification, the structured data can be reorganized (reconstructed) and serialized in combination with the current analysis instruction, so that the structured information required by the analysis instruction is added to the serialized data as much as possible, thereby improving the final task execution effect.
[0055] In some embodiments, the determining of the target reconstruction function from a plurality of preset reconstruction functions according to the reconstruction analysis result includes: based on a plurality of preset reconstruction functions and the application scenarios corresponding to each reconstruction function, determining a target application scenario from a plurality of application scenarios according to the reconstruction analysis result, and using the reconstruction function corresponding to the target application scenario as the target reconstruction function. In some embodiments, the application scenario corresponding to the reconstruction function can characterize the applicable task of the reconstruction function to a certain extent. For example, if the application scenario corresponding to a certain reconstruction function is a risk analysis scenario, then if the analysis instruction is an instruction for performing risk analysis, the reconstruction function can be selected as the target reconstruction function. In some embodiments, the reconstruction analysis result includes one or more application scenarios associated with the analysis instruction. In some embodiments, the target application scenario can be selected from a plurality of application scenarios associated with the analysis instruction based on the reconstruction analysis result and a preset selection rule. For example, the application scenario with the highest matching degree or correlation degree can be selected as the target application scenario.
[0056] In some embodiments, determining a target reconstruction function from a plurality of preset reconstruction functions based on the reconstruction analysis result includes: determining whether the reduced data has data missing based on the reconstruction analysis result; if not, determining a target reconstruction function from a plurality of preset reconstruction functions based on the reconstruction analysis result. In some embodiments, the reconstruction analysis result also includes indication information for indicating whether the reduced data has data missing. If data missing exists, it indicates that the reduced data lacks data for executing the analysis instruction. In some embodiments, the reconstruction analysis result also includes descriptive information related to the missing data, such as data object, data location, data type, etc. In some embodiments, if it is determined that the reduced data does not have data missing, a target reconstruction function that meets the requirements is determined from a plurality of preset reconstruction functions based on the reconstruction analysis result. If data missing exists, the reduced data may be re-extracted or supplementary information corresponding to the reduced data may be extracted from the digitized structure based on the descriptive information of the missing data, and the supplementary information may be serialized together with the reduced data for executing the data analysis task.
[0057] In some embodiments, the method further comprises: if it is determined that the reduced data contains missing data, re-extracting data from the structured data based on a data missing analysis result corresponding to the reduced data and the instruction analysis result, wherein the data missing analysis result indicates the portion of the reduced data containing missing data. In some embodiments, if it is determined that the reduced data contains missing data and the missing data is key data associated with the analysis instruction, re-extracting data from the structured data based on the data missing analysis result corresponding to the reduced data and the instruction analysis result. In some embodiments, the data missing analysis result is determined by analyzing the reduced data and indicates whether the current reduced data includes all data used to execute the analysis instruction. In some embodiments, if the reduced data contains missing data based on the reconstruction analysis result, the reduced data and the analysis instruction are input into a preset model to obtain a data missing analysis result corresponding to the reduced data. In some embodiments, if the reduced data contains missing data based on the reconstruction analysis result, the reduced data, the analysis instruction, and the reconstruction analysis result are input into a preset model to obtain a data missing analysis result corresponding to the reduced data.
[0058] Figure 2 This is a flow chart of an example of extracting simplified data provided in the embodiments of this specification. Figure 2 As shown, structured data, data descriptions corresponding to the structured data, and analysis instructions can be input into a large language model for instruction analysis to obtain analysis results output by the large language model. Then, corresponding data is extracted from the structured data based on the analysis results to obtain simplified data, which is the data required to execute the analysis instructions.
[0059] Figure 3 This is a flow chart of an example of data reconstruction provided in the embodiment of this specification. Figure 3 As shown, structured data, data descriptions corresponding to the structured data, and analysis instructions can be input into a large language model for instruction analysis to obtain analysis results output by the large language model. Then, based on the analysis results, a reconstruction function that meets the requirements is selected from a plurality of predefined reconstruction functions (that is, a target reconstruction function is selected). Then, based on the selected reconstruction function, the results of data extraction in the previous step (that is, streamlined data) can be reconstructed to obtain reconstructed data. Through reconstruction, the reconstructed data includes the structured information required for the analysis instruction.
[0060] Figure 4This is a schematic diagram of the structure of a device for performing a data analysis task provided in an embodiment of this specification. The device for performing a data analysis task (hereinafter referred to as "task execution device 1") can be implemented as all or part of an electronic device through software, hardware, or a combination of both. According to some embodiments, the task execution device 1 includes a first acquisition module 11 and a second acquisition module 12.
[0061] a first obtaining module 11, configured to input structured data to be analyzed and an analysis instruction corresponding to the structured data into a first large language model, and obtain an instruction analysis result output by the first large language model, wherein the instruction analysis result indicates a portion of the structured data required for the analysis instruction;
[0062] The second obtaining module 12 is configured to extract data from the structured data according to the instruction analysis result to obtain simplified data, so as to execute a data analysis task on the serialized simplified data through the analysis instruction.
[0063] In some embodiments, the task execution device 1 is also used to: input the structured data to be analyzed into the second largest language model, and obtain the summary description text corresponding to the structured data output by the second largest language model; wherein, the first acquisition module 11 is used to: input the structured data, the summary description text and the analysis instructions corresponding to the structured data into the first largest language model, and obtain the instruction analysis results output by the first largest language model.
[0064] In some embodiments, the task execution device 1 is also used to: input the structured data and the analysis instructions into a third language model to obtain a reconstruction analysis result corresponding to the structured data output by the third language model; determine a target reconstruction function from a plurality of preset reconstruction functions based on the reconstruction analysis result, and use the target reconstruction function to reconstruct the streamlined data to obtain reconstructed data containing structured information, so as to perform a data analysis task on the serialized reconstructed data through the analysis instructions.
[0065] In some embodiments, determining a target reconstruction function from a plurality of preset reconstruction functions according to the reconstruction analysis results includes: based on a plurality of preset reconstruction functions and an application scenario corresponding to each reconstruction function, determining a target application scenario from a plurality of application scenarios according to the reconstruction analysis results, and using the reconstruction function corresponding to the target application scenario as the target reconstruction function.
[0066] In some embodiments, determining a target reconstruction function from a plurality of preset reconstruction functions according to the reconstruction analysis result includes: determining whether there is data missing in the streamlined data according to the reconstruction analysis result; if not, determining a target reconstruction function from a plurality of preset reconstruction functions according to the reconstruction analysis result.
[0067] In some embodiments, the task execution device 1 is also used to: if it is determined that there is data missing in the simplified data, re-extract data from the structured data based on the data missing analysis result corresponding to the simplified data and the instruction analysis result, wherein the data missing analysis result is used to indicate the part of the simplified data where there is data missing.
[0068] The above device embodiments correspond to the aforementioned method embodiments. For detailed descriptions, please refer to the description of the method embodiments, which will not be repeated here. The device embodiments are obtained based on the corresponding method embodiments and have the same technical effects as the corresponding method embodiments. For detailed descriptions, please refer to the corresponding method embodiments.
[0069] The embodiments of this specification also provide a computer storage medium, which can store multiple instructions, and the instructions are suitable for being loaded by a processor and executing the method of the embodiments of this specification.
[0070] An embodiment of the present specification further provides a computer program product, which stores at least one instruction, and the at least one instruction is loaded by the processor to execute the method of the embodiment of the present specification.
[0071] The embodiments of this specification also provide Figure 5 The structural diagram of the electronic device shown in FIG. Figure 5 At the hardware level, the electronic device includes a processor, an internal bus, a network interface, memory, and non-volatile storage, and may also include other hardware required for its operations. The processor reads the corresponding computer program from the non-volatile storage into the memory and then runs it to implement the above method.
[0072] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
[0073] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0074] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0075] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.
[0076] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.
[0077] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.
[0078] This specification may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.
[0079] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.
[0080] The foregoing is merely an example of the present invention and is not intended to limit the present invention. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be included within the scope of the claims of the present invention.
Claims
1. A method for performing a data analysis task, comprising: Inputting structured data to be analyzed and analysis instructions corresponding to the structured data into a first large language model, obtaining an instruction analysis result output by the first large language model, wherein the instruction analysis result is used to indicate a portion of the structured data required for the analysis instruction; Data extraction is performed on the structured data according to the instruction analysis result to obtain simplified data, so as to execute a data analysis task on the serialized simplified data through the analysis instruction.
2. The method according to claim 1, further comprising: Inputting the structured data to be analyzed into the second language model to obtain a summary description text corresponding to the structured data output by the second language model; The step of inputting the structured data to be analyzed and the analysis instructions corresponding to the structured data into the first large language model to obtain the instruction analysis results output by the first large language model includes: The structured data, the summary description text, and the analysis instructions corresponding to the structured data are input into a first large language model to obtain an instruction analysis result output by the first large language model.
3. The method according to claim 1, further comprising: Inputting the structured data and the analysis instruction into a third language model to obtain a reconstruction analysis result corresponding to the structured data output by the third language model; According to the reconstruction analysis result, a target reconstruction function is determined from a plurality of preset reconstruction functions, and the target reconstruction function is used to reconstruct the simplified data to obtain reconstructed data containing structured information, so as to perform a data analysis task on the serialized reconstructed data through the analysis instruction.
4. The method according to claim 3, wherein determining a target reconstruction function from a plurality of preset reconstruction functions according to the reconstruction analysis result comprises: Based on a plurality of preset reconstruction functions and an application scenario corresponding to each reconstruction function, a target application scenario is determined from the plurality of application scenarios according to the reconstruction analysis result, and the reconstruction function corresponding to the target application scenario is used as a target reconstruction function.
5. The method according to claim 3, wherein determining a target reconstruction function from a plurality of preset reconstruction functions according to the reconstruction analysis result comprises: Determining whether there is data missing in the simplified data according to the reconstruction analysis result; If not, a target reconstruction function is determined from a plurality of preset reconstruction functions according to the reconstruction analysis result.
6. The method according to claim 5, further comprising: If it is determined that the simplified data has data missing, data extraction is performed again on the structured data according to the data missing analysis result corresponding to the simplified data and the instruction analysis result, wherein the data missing analysis result is used to indicate the part of the simplified data that has data missing.
7. A device for performing a data analysis task, comprising: a first obtaining module, configured to input structured data to be analyzed and an analysis instruction corresponding to the structured data into a first large language model, and obtain an instruction analysis result output by the first large language model, wherein the instruction analysis result indicates a portion of the structured data required for the analysis instruction; The second obtaining module is used to extract data from the structured data according to the instruction analysis result to obtain simplified data, so as to perform a data analysis task on the serialized simplified data through the analysis instruction.
8. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
9. An electronic device, characterized in that: include: A processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the steps of the method according to any one of claims 1 to 6.
10. A computer program product having at least one instruction stored thereon, characterized in that: When the at least one instruction is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.