Data analysis method and device based on large model and computer equipment

Abstract

The invention relates to a data analysis method and device based on a large model and computer equipment. Comprising the steps that table information extraction can be automatically conducted on any input single table, input questions are automatically disassembled to form answer thinking chain data, data analysis codes are generated through a generation algorithm module based on a large language model, and a data analysis conclusion is finally formed through optimization and post-processing. Therefore, a large amount of repeated work in the traditional data analysis process is optimized through modularized and automatic workflow, the data analysis work efficiency is greatly improved, meanwhile, the threshold of data analysis work is lowered, and data can be directly questioned through natural languages.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a data analysis method, device and computer equipment based on a large model. Background Art

[0002] In order to process natural language and understand disordered data, large language models came into being. In production and life, large language models can play a vital role.

[0003] At present, existing technologies only apply big language models to some links in the traditional data analysis workflow, and have not yet enabled the big language model for the entire data analysis process. There are still some links that require a lot of customized development and pre-definition work, such as the need to develop tool APIs for data processing, data analysis and other links, or to perform a lot of pre-processing and pre-definition work on data in specific fields and scenarios before they can be called for analysis through a big language model. Summary of the invention

[0004] Based on this, the purpose of this application is to provide a large model-based data analysis method, device and computer equipment that lowers the technical threshold of data analysis to solve the technical problems mentioned in the above background technology.

[0005] In the first aspect, the present application provides a data analysis method based on a large model, including:

[0006] Determine problem data and a data table corresponding to the problem data, and extract background knowledge from the data table;

[0007] Decomposing the problem data to obtain thinking chain data;

[0008] Combining the background knowledge with the thought chain data, and converting the combined result into a target data code based on the big model;

[0009] The running result of the target data code is processed to obtain the analysis result of the problem data.

[0010] In one embodiment, the problem data is decomposed to obtain thinking chain data, including: decomposing multiple module problems in the problem data; constructing sub-thinking chain data corresponding to each module problem; and associating each sub-thinking chain data according to the logical relationship between each module problem to obtain thinking chain data.

[0011] In one embodiment, sub-thinking chain data corresponding to each of the module questions is constructed, including: determining the user display requirements and the field conditions corresponding to the user display requirements; extracting multiple target keywords in the module questions according to the field conditions; and constructing the multiple target keywords into sub-thinking chain data according to the logical relationship between each of the target keywords.

[0012] In one embodiment, the background knowledge includes multiple marked keywords; the thought chain data includes multiple target keywords; the background knowledge and the thought chain data are combined, including: matching each target keyword with multiple marked keywords in the background knowledge respectively to obtain at least one information data corresponding to the target keyword; processing the information data according to the processing algorithm associated with the target keyword.

[0013] In one embodiment, after converting the combined result into the target data code based on the large model, it also includes: when an error occurs when running the target data code, calling new thinking chain data associated with the problem data, and returning to the process of combining the background knowledge and the new thinking chain data to continue; the new thinking chain data is obtained by decomposing the problem data based on different decomposition ideas.

[0014] In one embodiment, the analysis results include text results and picture results; after obtaining the analysis results of the problem data, it also includes: determining different interfaces associated with different display systems, and calling and displaying at least one of the text results and the picture results through different interfaces.

[0015] In a second aspect, the present application also provides a data analysis device based on a large model, including:

[0016] A data source module, used to determine problem data and a data table corresponding to the problem data;

[0017] An information extraction module, used to extract background knowledge from the data table;

[0018] A problem analysis module, used for decomposing the problem data to obtain thought chain data;

[0019] An intelligent generation module, used for combining the background knowledge with the thought chain data, and converting the combined result into a target data code based on a large model;

[0020] The analysis and optimization module is used to process the running results of the target data code to obtain the analysis results of the problem data.

[0021] In one embodiment, the above-mentioned device also includes: a code execution module, which runs the target data code, and when an error occurs when running the target data code, triggers the problem analysis module to call new thinking chain data associated with the problem data; and is also used to obtain the analysis result of the problem data by triggering the analysis optimization module when the target data code runs successfully.

[0022] In a third aspect, the present application further provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned large model-based data analysis method when executing the computer program.

[0023] In a fourth aspect, the present application further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned large-model-based data analysis method are implemented.

[0024] The above-mentioned data analysis method, device, computer equipment and readable storage medium based on the big model, by determining the problem data and the data table corresponding to the problem data, and extracting the background knowledge in the data table, then decomposing the problem data to obtain the thinking chain data; after combining the background knowledge and the thinking chain data, the combined result can be converted into the target data code based on the big model, so that the running result of the target data code can be processed to obtain the analysis result of the problem data. The technical solution of the present application has high universal adaptability. For any single table input, the table information can be automatically extracted, and the input question can be automatically disassembled to form the answer thinking chain data. The data analysis code is generated by the generation algorithm module based on the big language model, and the data analysis conclusion is finally formed after optimization and post-processing. Therefore, a large amount of repetitive work in the traditional data analysis process is optimized through modular and automated workflows, which greatly improves the efficiency of data analysis work, and also lowers the threshold of data analysis work, and can directly ask questions to the data through natural language. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A diagram of an application environment of a data analysis method based on a large model in an embodiment;

[0026] Figure 2 A schematic diagram of a flow chart of a data analysis method based on a large model in one embodiment;

[0027] Figure 3 A schematic diagram of the contents of a target data code in one embodiment;

[0028] Figure 4 A schematic diagram of a flow chart of a data analysis method based on a large model in another embodiment;

[0029] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0031] The data analysis method based on the large model provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, it is realized through the collaborative work between the terminal 102 and the server 104. Among them, the terminal 102 communicates with the server 104 through the network. The terminal 102 sends the problem data obtained from the user to the server 104, and the server 104 determines the data table corresponding to the problem data and extracts the background knowledge in the data table; the problem data is decomposed to obtain the thinking chain data. The server 104 is also used to combine the background knowledge and the thinking chain data, and convert the combined results into target data codes based on the large model; the running results of the target data code are processed to obtain the analysis results of the problem data, so that the terminal 102 displays the analysis results to the user. Among them, the terminal 102 can be the end system where the user is located, which can be but not limited to various personal computers, laptops, smart phones, tablet computers, drone equipment, smart car-mounted equipment and portable wearable devices. Server 104 can be implemented with an independent server or a server cluster composed of multiple servers. It can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0032] In one embodiment, Figure 2 As shown, a data analysis method based on a large model is provided, and the method is applied to a computer system as an example for explanation. The computer system includes Figure 1 The terminal and server in the embodiment include the following steps:

[0033] Step 202, determining the problem data and the data table corresponding to the problem data, and extracting the background knowledge in the data table.

[0034] The data table includes but is not limited to data in a table format, and may also be specific data in other different formats such as document data.

[0035] Specifically, the computer system includes a terminal that can be presented to the user and is configured with an input module adapted to different systems. When the user needs to simply and quickly implement the data analysis task, the user uploads the proposed problem data and the data table corresponding to the problem data through the terminal respectively, so that the input module receives the problem data and the data table, such as uploading the data table "Number of civil cars owned in each province in China (10,000 vehicles).xlsx", and the proposed problem data is "What is the gap between the number of civil cars owned in Beijing between 2002 and 2020 and the number of civil cars owned in Shanghai between 2002 and 2020".

[0036] The computer system also includes an information extraction module. When the computer system determines the problem data and the data table corresponding to the problem data, the information extraction module reads the local information of the data table and combines the semantic understanding function to extract information from the input data table to form background knowledge for data analysis. Among them, the background knowledge includes multiple tagged keywords, such as time, province and other keywords. Therefore, the user only needs to input a simple data table or problem data, without professional programming skills or data analysis knowledge, and does not need to limit the input to specific data, to achieve subsequent complex data analysis tasks for any data. This fully takes into account the versatility of the data analysis function, greatly reduces the threshold for data analysis, and improves the user experience.

[0037] In one embodiment, the computer system reads multiple column names in a data table, identifies the data in the multiple column names, and performs different keyword tags. For example, after extracting all column names of a table, the first k rows of the table are analyzed for data examples, and special processing tags are added to special columns, such as marking the "time" keyword on the time column name, so that the associated algorithm is called for processing when the data corresponding to the keyword is subsequently processed.

[0038] In one embodiment, multiple data tables are pre-stored in a data source module configured in a computer system. When the computer system obtains problem data, it can automatically identify and match the data table corresponding to the problem data without the user having to upload the data table one by one for different problem data each time, thereby making the data analysis response process more efficient and convenient.

[0039] Step 204, decompose the problem data to obtain thought chain data.

[0040] Specifically, the computer system can use a model that combines few-shot learning and thought chain prompts (also known as Few-shot CoT technology) to decompose the problem data. The computer system also includes a problem analysis module, and the above-mentioned Few-shot CoT model is pre-configured in the problem analysis module. Therefore, the computer device decomposes the problem data input by the user into a series of smaller, easy-to-manage sub-problems in the form of thought chains, that is, thought chain data, through the Few-shot CoT technology, and finally solves the entire problem data by gradually solving these sub-problems. When solving each sub-problem, the reasoning process is clearly recorded and displayed, which helps to understand the logic and judgment of each step and improve the transparency and explainability of the entire process.

[0041] Among them, thought chain technology is an advanced technology in the field of natural language processing, which is used to enhance the reasoning and logical capabilities of artificial intelligence models. It breaks down complex tasks into a series of smaller and easier to manage steps or thinking processes, enabling the model to solve problems more effectively.

[0042] In one embodiment, the model combining the above-mentioned few-sample learning and thought chain prompts is pre-trained in a computer system, including: the computer system determines the demand tasks that users usually need to provide and the data set containing a small number of labeled samples; designs a series of reasoning chain prompts based on the characteristics of the demand tasks to guide the model to reason step by step; trains the model through the data set of a small number of labeled samples and reasoning chain prompts, and uses the actual user needs as a test set to evaluate the performance of the model.

[0043] Step 206, combining the background knowledge and the thinking chain data, and converting the combined result into a target data code based on the big model.

[0044] Specifically, data analysis tasks can usually be implemented through programming languages, data analysis software, big data platforms, cloud services, etc., among which the implementation through programming languages ​​has the characteristics of high scalability, excellent integration, rich tool libraries, etc., and is more widely used. The computer system also includes an intelligent generation module. This application optimizes the process of implementing data analysis with programming languages. Therefore, the computer device combines the above background knowledge and thinking chain data through the intelligent generation module, and calls the large language model to understand the two aspects of information, and converts the combined results into target data code through a given generation template. The programming languages ​​of the target data code include but are not limited to python, SQL, and R language.

[0045] Among them, large models refer to machine learning models with large-scale parameters and complex computing structures, which are usually built by deep neural networks and have billions or even hundreds of billions of parameters. The purpose of designing large models is to improve the model's expressiveness and predictive performance, and to be able to handle more complex tasks and data.

[0046] Therefore, by combining the powerful understanding and generation capabilities of the large language model, the intelligent generation module can automatically and efficiently generate high-quality target data code. It also supports multiple programming languages, especially the optimization of the Python language, which improves the efficiency and scalability of code generation.

[0047] In one embodiment, taking the generation of Python code as an example, the target data code obtained by converting the combined result based on the large model is as follows: Figure 3 shown.

[0048] Step 208, processing the running result of the target data code to obtain the analysis result of the problem data.

[0049] The computer system also includes a code execution module and an analysis and optimization module.

[0050] Specifically, when the computer device obtains the target data code based on different programming languages, the target data code is run through the code execution module. When the running result does not report an error, the computer device analyzes and optimizes the running result of the code through the analysis and optimization module in combination with background knowledge. The obtained result is divided into a text part and a picture part.

[0051] Then, for the text part, the computer equipment organizes the brief data analysis results into complete and easy-to-read data analysis conclusions, and provides corresponding analysis evidence to obtain the final text results; for the image part, the computer equipment further interprets and describes the information reflected in the image, obtains the final image results, and combines the text results and image results to form a complete and well-founded analysis result. Because the analysis and optimization module can further analyze and optimize the code execution results, generate concise and easy-to-read text conclusions and detailed chart interpretations, the professionalism and diversity of the output results are ensured.

[0052] In one embodiment, the computer system can be applied to a variety of products such as financial data analysis software, marketing effectiveness analysis platform, and manufacturing production efficiency management tools. Through automated data processing and analysis processes, the system can quickly extract relevant data, decompose complex problems, generate and execute efficient analysis codes, and optimize results, thereby generating detailed analysis reports and optimization suggestions to help users make more accurate and efficient decisions.

[0053] In the above-mentioned data analysis method based on the big model, by determining the problem data and the data table corresponding to the problem data, and extracting the background knowledge in the data table, the problem data is decomposed to obtain the thought chain data, and the data analysis code is generated through the generation algorithm module based on the big language model, and the data analysis conclusion is finally formed after optimization and post-processing. Therefore, a large amount of repetitive work in the traditional data analysis process is optimized through modular and automated workflows, which greatly improves the efficiency of data analysis work, and also lowers the threshold of data analysis work, and directly asks questions to the data through natural language.

[0054] Since this method of automatically generating analysis results based on data tables and problem data input by users can be widely used in multiple fields, including business intelligence, market research, financial services, medical and health management, education, e-commerce, government public management, and customer relationship management, it can help various industries efficiently process and analyze data, thereby improving work efficiency, optimizing decision-making processes, and enhancing competitiveness.

[0055] In one embodiment, the problem data is decomposed to obtain thinking chain data, including: decomposing multiple module problems in the problem data; constructing sub-thinking chain data corresponding to each module problem; and associating each sub-thinking chain data according to the logical relationship between each module problem to obtain the thinking chain data.

[0056] Specifically, since each thinking step in the thinking chain should be logically related, they should be connected to each other to form a complete thinking process. Therefore, the computer device decomposes the problem data into multiple module problems based on a certain decomposition idea. Taking the above problem data "What is the gap between the number of civilian cars owned by Beijing between 2002 and 2020 and the number of civilian cars owned by Shanghai between 2002 and 2020" as an example, it can be decomposed into module problem A as "What is the number of civilian cars owned by Beijing between 2002 and 2020", module problem B as "What is the number of civilian cars owned by Shanghai between 2002 and 2020", and module problem C as "What is the gap between module problem A and module problem B". Then the computer device constructs the sub-thinking chain data corresponding to each module problem, and associates each sub-thinking chain data according to the logical relationship between each module problem to obtain the thinking chain data.

[0057] It is easy to understand that logical relationships represent the logical relationships that exist when thinking about how to execute the steps of a problem. For example, in the above example, after executing module problem A and module problem B and obtaining results, the logical process of executing module problem C based on the results of the previous step is required. Therefore, after associating the corresponding sub-thinking chain data according to the logical relationship, a coherent path to solve the problem data can be constructed.

[0058] In one embodiment, the computer device considers the data features of the problem data and determines whether the problem data needs to be decomposed into multiple module problems. If not, the computer device directly constructs the thinking chain data corresponding to the problem data.

[0059] In this embodiment, complex problems are decomposed into manageable modular sub-problems and solved step by step, while the reasoning process is clearly recorded and displayed. This not only enhances the transparency and interpretability of data analysis, but also facilitates users to understand the logic and judgment of each step. Compared with the traditional situation where some links require a lot of custom development and predefined work, this application directly decomposes the execution process of the problem in a modular form, greatly shortening the time for subsequent large language model call analysis.

[0060] In one embodiment, sub-thinking chain data corresponding to each module question is constructed, including: determining the user display requirements and the field conditions corresponding to the user display requirements; extracting multiple target keywords in the module question according to the field conditions; and constructing the multiple target keywords into sub-thinking chain data according to the logical relationship between each target keyword.

[0061] The user display requirements include but are not limited to the need to be achieved by drawing a bar chart, etc.

[0062] Specifically, taking the above module question A as an example, the process of creating sub-thinking chain data by the computer device is as follows: First, when it is determined that the user's display requirement is to present a bar chart, the corresponding field situation can be determined as the x-axis and y-axis fields. Secondly, the computer device can extract multiple target keywords in the module question according to the field situation, such as "Beijing" and "2002-2020", and find the record with the value of "Beijing". Then, based on the logical relationship between each target keyword, that is, on the logical basis of determining the record of "Beijing", the "civilian car ownership" of "2002-2020" is extracted from these records respectively. Finally, use the bar() function of matplotlib, take the year as the x-axis, and "civilian car ownership" as the y-axis, draw a bar chart, and call the show() function for image display.

[0063] In this embodiment, by determining the user's display requirements and the corresponding field conditions, multiple target keywords in the module question can be more accurately extracted to help the system accurately understand the user's needs. By analyzing the logical relationship between each target keyword, a more complete and organized sub-thinking chain data can be constructed, which helps to deeply understand the logic and association behind the problem. The constructed sub-thinking chain data can help the computer system to perform data analysis more comprehensively in the future and improve the accuracy and effectiveness of data analysis.

[0064] In one embodiment, the background knowledge includes multiple marked keywords; the thought chain data includes multiple target keywords; combining the background knowledge and the thought chain data includes: matching each target keyword with multiple marked keywords in the background knowledge respectively to obtain at least one information data corresponding to the target keyword; processing the information data according to a processing algorithm associated with the target keyword.

[0065] Specifically, the computer device will match multiple tagged keywords in the background knowledge through the logical relationship between the target keywords. For example, the tagged keyword corresponding to "Beijing" is first matched as "province", so the information data corresponding to the target keyword includes various data related to "Beijing" in the background knowledge, such as "time", "civilian car ownership", etc. Based on the logical basis of determining the information data related to "Beijing", the tagged keyword corresponding to "2002-2020" is matched as "time", and so on. The intelligent generation module in the computer device determines the processing algorithm associated with the target keyword, such as the time processing algorithm and the ownership data processing algorithm, and then the processing algorithm associated with the target keyword processes the information data.

[0066] In this embodiment, by matching each target keyword with multiple marked keywords in the background knowledge, at least one information data corresponding to each target keyword can be accurately obtained, which helps the system to more accurately understand the user's needs and problem background. By accurately extracting and effectively processing information data, the system can provide a more in-depth and comprehensive analysis for each target keyword, thereby enhancing its problem-solving ability and speed.

[0067] In one embodiment, after converting the combined results into target data code based on the large model, it also includes: when an error occurs when running the target data code, calling new thinking chain data associated with the problem data, and returning to the process of combining background knowledge and new thinking chain data to continue.

[0068] Among them, the new thinking chain data is obtained by decomposing the problem data based on different decomposition ideas.

[0069] Specifically, the computer device re-decomposes multiple module problems in the problem data based on different decomposition ideas until new thinking chain data is obtained according to the new logical relationship between the new module problems. The new thinking chain data can be pre-stored. When the target data code reports an error, the computer device directly calls the new thinking chain data associated with the problem data.

[0070] In this embodiment, when the target data code runs into an error, calling the new thought chain data associated with the problem data can help the system recover quickly and continue to process the problem. And using the new thought chain data, the system can provide more perspectives and analysis for each specific problem, thereby enhancing its problem processing capabilities and improving the efficiency and accuracy of subsequent data analysis.

[0071] In one embodiment, after obtaining the analysis result of the question data, the method further includes: determining different interfaces associated with different display systems, and calling and displaying at least one of the text results and the image results through the different interfaces.

[0072] Specifically, the analysis results include text results and picture results. By configuring the output modules in the computer system that are compatible with different systems and packaging the output modules into the form of API interfaces, when different interfaces associated with different display systems are determined, the computer system can output the text results and picture results of the optimization conclusions through different interfaces. Therefore, it is ensured that the analysis results can be flexibly called by different systems and front-end projects.

[0073] In one embodiment, Figure 4 As shown, another data analysis method based on a large model is provided, including: the input module of the computer device temporarily stores the received problem data and data table to the data source module of the computer device. The background knowledge in the data table is obtained by the information extraction module of the computer device, and the problem analysis module of the computer system decomposes the problem data to obtain the thinking chain data. The intelligent generation module in the computer device combines the background knowledge and the thinking chain data, and converts the combined result into the target data code based on the large model. The code execution module in the computer device runs the target data code, and when the execution is successful, the analysis and optimization module in the computer system analyzes and optimizes the execution of the code operation result. The output module of the computer device outputs the analysis result of the problem data. When an error is reported when running the target data code, since the code execution module needs to check the full text and unify the code running caliber, it is necessary to re-trigger the problem analysis module, and then call the new thinking chain data associated with the problem data through the problem analysis module, so that the intelligent generation module can continue to combine the background knowledge with the new thinking chain data.

[0074] In this embodiment, the present invention realizes automated processes such as data extraction, problem decomposition, code generation and execution, and result optimization through an intelligent data analysis system based on a large language model, which significantly improves the efficiency and accuracy of data analysis. In particular, when processing large-scale data and complex analysis tasks, the system shows significant advantages and greatly improves the efficiency and quality of data analysis. In addition, the entire process greatly shortens the time of data analysis, reduces manual intervention, and lowers the technical threshold of data analysis work through automated and intelligent means.

[0075] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0076] Based on the same inventive concept, the embodiment of the present application also provides a large-model-based data analysis device for implementing the large-model-based data analysis method involved above. The implementation solution provided by the device to solve the problem is similar to the implementation solution recorded in the above method, so the specific limitations in one or more large-model-based data analysis device embodiments provided below can refer to the limitations of the large-model-based data analysis method above, and will not be repeated here.

[0077] In one embodiment, a data analysis device based on a large model is provided, comprising: a data source module, for determining problem data and a data table corresponding to the problem data;

[0078] Information extraction module, used to extract background knowledge from data tables;

[0079] The problem analysis module is used to decompose the problem data and obtain the thinking chain data;

[0080] Intelligent generation module, used to combine background knowledge and thinking chain data, and convert the combined results into target data code based on the big model;

[0081] The analysis and optimization module is used to process the running results of the target data code to obtain the analysis results of the problem data.

[0082] In one embodiment, the above-mentioned device also includes: a code execution module, which is used to run the target data code, and when an error occurs when running the target data code, it triggers the problem analysis module to call new thinking chain data associated with the problem data; and is also used to obtain the analysis results of the problem data by triggering the analysis optimization module when the target data code runs successfully.

[0083] Each module in the above-mentioned large model-based data analysis can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0084] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store analysis results of problem data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a data analysis method based on a large model is implemented.

[0085] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0086] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0087] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0088] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program includes computer instructions, the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in the above-mentioned method embodiments.

[0089] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited thereto.

[0090] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0091] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A data analysis method based on a large model, characterized in that: The method comprises: Determine problem data and a data table corresponding to the problem data, and extract background knowledge from the data table; Decomposing the problem data to obtain thinking chain data; Combining the background knowledge with the thought chain data, and converting the combined result into a target data code based on the big model; The running result of the target data code is processed to obtain the analysis result of the problem data.

2. The method according to claim 1, characterized in that The step of decomposing the problem data to obtain thought chain data includes: Decomposing multiple module problems in the problem data; Construct sub-thinking chain data corresponding to each of the module problems; According to the logical relationship between each of the module problems, each of the sub-thinking chain data is associated to obtain the thinking chain data.

3. The method according to claim 2, characterized in that The sub-thinking chain data corresponding to each of the module problems is constructed, including: Determine user display requirements and field conditions corresponding to the user display requirements; Extract multiple target keywords in the module question according to the field conditions; According to the logical relationship between each of the target keywords, multiple target keywords are constructed into sub-thinking chain data.

4. The method according to claim 1, characterized in that: The background knowledge includes a plurality of marked keywords; the thought chain data includes a plurality of target keywords; and combining the background knowledge and the thought chain data includes: Matching each target keyword with a plurality of marked keywords in the background knowledge respectively to obtain at least one information data corresponding to the target keyword; The information data is processed according to a processing algorithm associated with the target keyword.

5. The method according to claim 1, characterized in that After converting the combined result into the target data code based on the large model, it also includes: When an error occurs when running the target data code, new thinking chain data associated with the problem data is called, and the process returns to the process of combining the background knowledge and the new thinking chain data; the new thinking chain data is obtained by decomposing the problem data based on different decomposition ideas.

6. The method according to claim 1, characterized in that The analysis results include text results and picture results; after obtaining the analysis results of the problem data, it also includes: Different interfaces associated with different display systems are determined, and at least one of the text results and the image results is called and displayed through the different interfaces.

7. A data analysis device based on a large model, characterized in that: The device comprises: A data source module, used to determine problem data and a data table corresponding to the problem data; An information extraction module, used to extract background knowledge from the data table; A problem analysis module, used for decomposing the problem data to obtain thought chain data; An intelligent generation module, used for combining the background knowledge with the thought chain data, and converting the combined result into a target data code based on a large model; The analysis and optimization module is used to process the running results of the target data code to obtain the analysis results of the problem data.

8. The device according to claim 7, characterized in that The device also includes: A code execution module is used to run the target data code, and when an error occurs during the running of the target data code, trigger the problem analysis module to call new thinking chain data associated with the problem data; and is also used to obtain the analysis result of the problem data by triggering the analysis optimization module when the target data code runs successfully.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable 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.

Images