Intelligent Analysis Method for Data Intelligence Based on LLM Large Language Model

Abstract

The present invention provides a method for intelligent analysis of data intelligence based on the LLM large language model. The present invention provides a method for intelligent analysis of data intelligence based on the LLM large language model. Through the LLM large language model, based on the parent and child prompt words in the binary tree of prompt words for data intelligence analysis, intelligence analysis and mining are carried out on the tree structure. By traversing each intelligence analysis prompt word with a parent-child relationship on the input binary tree, the corresponding data intelligence with a tree structure is retrieved, analyzed, and mined. It can rely on the LLM large language model to quickly generate an intelligence network binary tree with a sense of hierarchy, which not only improves the efficiency of data intelligence mining and analysis, but also does not involve a large number of sample training and learning of machine learning algorithm models, reducing the complexity of data analysis. And finally, a data intelligence network with a tree structure can be generated, allowing users to conduct intelligence analysis and access from different levels, reducing the difficulty of intelligence analysis and understanding.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and particularly to an intelligent data intelligence analysis method based on a large language model (LLM), an intelligent data intelligence analysis device based on an LLM, an electronic device, and a computer-readable storage medium. Background Art

[0002] Data intelligence analysis is a process of collecting, processing, and analyzing a large amount of data to extract valuable information and insights. It combines methods in fields such as data science, statistics, and business intelligence, aiming to provide support for decision-making.

[0003] The key steps of data intelligence analysis include data collection, data cleaning, data analysis, data visualization, and decision support:

[0004] 1. Data collection: It is the basis for obtaining reliable data and the first step in data intelligence analysis. By collecting data from multiple channels and dimensions, the comprehensiveness and diversity of the data can be ensured, laying a solid foundation for subsequent analysis. Data collection mainly includes the acquisition of internal data and external data. Internal data usually comes from an enterprise's business systems, such as CRM, ERP, etc.; external data includes public data, third-party data, etc.

[0005] 2. Data cleaning: It is an important link in data intelligence analysis, aiming to ensure the accuracy and consistency of the data. Problems such as missing, duplicate, incorrect, and inconsistent data may occur during the data collection process, which will affect the subsequent analysis results. Data cleaning includes steps such as data deduplication, data completion, data verification, and data conversion.

[0006] 3. Data analysis: It is the core link of data intelligence analysis, aiming to mine valuable information from the data. Data analysis includes different types such as descriptive analysis, diagnostic analysis, predictive analysis, and prescriptive analysis.

[0007] 4. Data visualization: It is an important link to help users better understand and interpret data. Data visualization visually displays complex data in the form of charts, graphs, etc., making the data easier to understand and analyze. Data visualization can help users quickly discover the patterns and trends in the data, improving the efficiency and effectiveness of data analysis.

[0008] 5. Decision support: It is to use the results of data analysis to provide a scientific basis for an enterprise's strategic planning, marketing, product development, etc. Decision support needs to comprehensively consider various factors, such as the market environment, competitors, customer needs, etc., and conduct a comprehensive analysis and evaluation.

[0009] There are various methods for data intelligence analysis, and usually traditional data intelligence analysis methods mainly include the following:

[0010] Descriptive analysis: It is a basic analysis method aimed at summarizing past data and understanding what has happened.

[0011] Diagnostic analysis: Focuses on why a specific event occurs. Through in-depth data mining, analysts can identify the factors affecting a specific event.

[0012] Predictive analysis: Uses historical data to predict future trends. By building statistical models or using machine learning algorithms, future situations can be predicted to make better strategic decisions.

[0013] Prescriptive analysis: Aims to provide action recommendations. It not only tells enterprises what may happen in the future but also provides suggestions on how to respond.

[0014] Real-time analysis: In the digital age, real-time data analysis has become increasingly important. It allows enterprises to analyze data while it is being generated, enabling quick responses to market changes.

[0015] However, in the process of data analysis using the above traditional data intelligence analysis methods, the traditional analysis methods mainly rely on data statistical tools to assist in analysis. By classifying useful data intelligence, it is impossible to record and statistically analyze it to identify factors affecting specific events, etc. And for conventional statistical tools or components, the efficiency of mining data intelligence is very low, almost requiring manual operation. Therefore, the analysis and collection efficiency of data intelligence is very slow; using machine learning algorithms for predictive analysis involves a large amount of preparation of training samples and takes a long time for model training, with a huge workload and being relatively complex; existing data intelligence analysis cannot generate a hierarchical data intelligence network and cannot provide a well-structured intelligence dataset. Summary of the Invention

[0016] To solve the technical problems existing in the prior art, the present invention provides the following technical solutions:

[0017] On the one hand, a data intelligence intelligent analysis method based on the LLM large language model is provided. This method is implemented by an electronic device and includes:

[0018] S1. Construct and configure a prompt word binary tree for data intelligence analysis, where the prompt word binary tree includes a parent node and several child nodes, and the parent node and the child nodes are respectively configured with intelligence analysis prompt words with corresponding permissions;

[0019] S2, traversing the prompt word binary tree based on a recursive traversal algorithm, reading the intelligence analysis prompt words on the parent and child nodes in sequence and inputting them into a preset LLM large language model, performing a database search and analysis based on the intelligence analysis prompt words through the LLM large language model, and mining to obtain the data intelligence analysis results corresponding to the parent node or the child node;

[0020] S3, according to the relationship between the parent and child nodes, based on the information volume and information importance sorting, orderly counting the data intelligence analysis results of each node and generating a corresponding intelligence network binary tree;

[0021] S4. Traverse and verify the intelligence network binary tree, and output and save the intelligence network binary tree after verification.

[0022] Preferably, the step S1, constructing and configuring a binary tree of prompt words for data intelligence analysis, includes:

[0023] Import binary tree;

[0024] Traversing the parent node and each of the child nodes on the binary tree;

[0025] According to the analysis dimensions and contents of the data intelligence to be analyzed, respectively construct analysis data keywords, analysis scope keywords and analysis logic on the corresponding analysis dimensions on the parent node and each of the child nodes;

[0026] Combining the analysis data keywords, the analysis scope keywords and the analysis logic on the corresponding analysis dimension, constructing the intelligence analysis prompt words corresponding to the parent node and each of the child nodes respectively, and configuring the intelligence analysis prompt words at the corresponding parent node or the child node;

[0027] After the configuration is completed, the prompt word binary tree is obtained.

[0028] Preferably, S2, based on a recursive traversal algorithm, traverses the prompt word binary tree, reads the intelligence analysis prompt words on the parent and child nodes in sequence and inputs a preset LLM large language model, performs a database search and analysis based on the intelligence analysis prompt words through the LLM large language model, and mines and obtains the data intelligence analysis results corresponding to the parent node or the child node, including:

[0029] Starting from the parent node on the prompt word binary tree, the intelligence analysis prompt words configured on the parent node and the child node are read in sequence, wherein the recursive traversal algorithm :

[0030] ,

[0031] in:

[0032] is the target node to be traversed currently,

[0033] is the parent node of the current node N, initially represents the root node, that is, there is no parent node,

[0034] are the left child node and the right child node of the current node N respectively,

[0035] is the operation sequence operator, indicating "execute the left operation first and then the right operation",

[0036] is to process the parent-child relationship, output or record the information of the parent node P and the child node N,

[0037] is the root node, indicating that the node does not exist, and the traversal terminates at this time;

[0038] Input the read intelligence analysis prompt word into the preset LLM large language model, so that the LLM large language model performs a database search in the data intelligence database based on the intelligence analysis prompt word, analyzes and mines the data intelligence analysis results matching the intelligence analysis prompt word and outputs them;

[0039] Write and save the output data intelligence analysis results to the corresponding parent node or child node to obtain the data intelligence analysis results of the corresponding parent node or child node.

[0040] Preferably, in S3, according to the relationship between the parent and child nodes, based on the amount of information and the importance of information for sorting, orderly count the data intelligence analysis results of each node and generate the corresponding intelligence network binary tree, including:

[0041] Calculate the amount of information and the importance of information of each node and sort them, and traverse and count the data intelligence analysis results at the parent node and each child node based on the sorting results; among them, the amount of information and the importance of information are respectively expressed as and :

[0042] ,

[0043] where:

[0044] N: the current node,

[0045] A i : the i-th attribute of node N,

[0046] k: The total number of attributes of node N,

[0047] w i : The weight of the i-th attribute,

[0048] Complexity(A i ): The complexity of the i-th attribute, calculated as follows:

[0049] If the attribute value is numerical: Complexity(A i ) = log2(1 + |A i |);

[0050] If the attribute value is categorical: Complexity(A i ) = Entropy(A i ), where Entropy(A i ) is the entropy value of the categorical attribute;

[0051] ,

[0052] Where:

[0053] Depth(N): The depth of node N in the tree, where the depth of the root node is 0,

[0054] Degree(N): The degree of connection of node N,

[0055] Centrality(N): The centrality of node N, where the centrality is betweenness centrality or closeness centrality,

[0056] : The weight coefficient, used to adjust the contribution of each item to the importance;

[0057] Import a new binary tree A, and configure the corresponding parent and child nodes of the new binary tree A according to the relationship between the parent and child nodes on the binary tree of the said prompt words;

[0058] Write the data intelligence analysis results of the obtained parent node and each child node through traversal extraction onto the new binary tree A to obtain the intelligence network binary tree.

[0059] Preferably, in S4, traverse and verify the intelligence network binary tree, and output and save the intelligence network binary tree after passing the verification, including:

[0060] Pre-configure corresponding data intelligence verification rules for the said parent node and each child node;

[0061] Use the large language model (LLM) to extract the verification keywords, verification scope keywords, and corresponding verification logic in the data intelligence verification rules;

[0062] Combine the verification keywords, the verification scope keywords, and the corresponding verification logic to construct intelligence verification prompts for the corresponding parent node and each child node respectively, and configure the intelligence verification prompts at the corresponding parent node or child node;

[0063] Traverse the data intelligence analysis results and the intelligence verification prompts at the parent node or child node of the intelligence network binary tree, and input the intelligence verification prompts into the preset LLM;

[0064] Through the LLM, based on the intelligence verification prompts at the parent node or child node, perform data intelligence verification on the data intelligence analysis results at the parent node or child node:

[0065] If the verification passes, mark the parent node or child node as the passed status;

[0066] Otherwise, mark it as the warning status;

[0067] After the verification is completed, output and save the intelligence network binary tree with status marks.

[0068] On the other hand, a data intelligence intelligent analysis device based on the LLM is provided. The data intelligence intelligent analysis device based on the LLM is used to implement the above-mentioned data intelligence intelligent analysis method based on the LLM. The device includes:

[0069] A prompt construction module for constructing and configuring a prompt binary tree for data intelligence analysis, where the prompt binary tree includes a parent node and several child nodes, and the parent node and the child nodes are respectively configured with intelligence analysis prompts with corresponding permissions;

[0070] An intelligence mining module for traversing the prompt binary tree based on a recursive traversal algorithm, sequentially reading the intelligence analysis prompts on the parent and child nodes and inputting them into the preset LLM, and performing database analysis through the LLM based on the intelligence analysis prompts to mine the data intelligence analysis results corresponding to the parent node or child node;

[0071] An intelligence statistics module for sorting based on the amount of information and information importance according to the relationship between the parent and child nodes, and orderly statistics the data intelligence analysis results of each node and generating a corresponding intelligence network binary tree;

[0072] An intelligence verification module is used to traverse and verify the intelligence network binary tree, and after successful verification, output and save the intelligence network binary tree.

[0073] On the other hand, an electronic device is provided, which includes: a processor; a memory, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by the processor, any one of the methods in the above-mentioned data intelligence intelligent analysis method based on the LLM large language model is implemented.

[0074] On the other hand, a computer-readable storage medium is provided, in which at least one instruction is stored, and the at least one instruction is loaded and executed by a processor to implement any one of the methods in the above-mentioned data intelligence intelligent analysis method based on the LLM large language model.

[0075] The beneficial effects brought by the technical solutions provided in the embodiments of the present invention at least include:

[0076] The present invention provides a data intelligence intelligent analysis method based on the LLM large language model. Through the LLM large language model, intelligence analysis and mining are carried out on the parent and child prompt words in the prompt word binary tree for data intelligence analysis, and by traversing each intelligence analysis prompt word with a parent-child relationship on the input binary tree, the corresponding data intelligence with a tree structure is retrieved, analyzed, and mined. It can rely on the LLM large language model to quickly generate a hierarchical intelligence network binary tree, which not only improves the efficiency of data intelligence mining and analysis, but also does not involve a large number of sample training and learning of machine learning algorithm models, reducing the complexity of data analysis. And finally, a data intelligence network with a tree structure can be generated, allowing users to conduct intelligence analysis and access from different levels, reducing the difficulty of intelligence analysis and understanding. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0078] Figure 1 is a flowchart of a data intelligence intelligent analysis method based on the LLM large language model provided by the embodiments of the present invention;

[0079] Figure 2 is a schematic diagram of the application structure of a binary tree provided by the embodiments of the present invention;

[0080] Figure 3 is a block diagram of a data intelligence intelligent analysis device based on the LLM large language model provided by the embodiments of the present invention;

[0081] Figure 4 It is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. Specific implementation manners

[0082] The following describes the technical solutions in the present invention with reference to the accompanying drawings.

[0083] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as an "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either of them can be selected.

[0084] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same. "(of)", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same.

[0085] In the embodiments of the present invention, sometimes subscripts such as W1 may be miswritten as non-subscript forms such as W1. When the difference is not emphasized, the meanings they express are the same.

[0086] To make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0087] The LLM large language model of this embodiment can be self-designed or by calling a third-party LLM large language model such as the Wenyan Yixin model of Baidu, etc., and this embodiment does not make any limitations.

[0088] An embodiment of the present invention provides a data intelligence intelligent analysis method based on an LLM large language model. This method can be implemented by an electronic device, and this electronic device can be a terminal or a server. As Figure 1 shown in the flowchart of the data intelligence intelligent analysis method based on the LLM large language model, the processing flow of this method can include the following steps:

[0089] S1. Construct and configure a prompt word binary tree for data intelligence analysis. Among them, the prompt word binary tree includes a parent node and several child nodes, and the parent node and the child nodes are respectively configured with intelligence analysis prompt words with corresponding permissions;

[0090] S2. Traverse the prompt word binary tree based on the recursive traversal algorithm, sequentially read the intelligence analysis prompt words on the parent and child nodes and input them into the preset large language model (LLM). Through the LLM, perform database query analysis based on the intelligence analysis prompt words, and mine the data intelligence analysis results corresponding to the parent node or the child node;

[0091] S3. According to the relationship between the parent and child nodes, sort based on the amount of information and the importance of information, orderly count the data intelligence analysis results of each node, and generate a corresponding intelligence network binary tree;

[0092] S4. Traverse and verify the intelligence network binary tree, and output and save the intelligence network binary tree after passing the verification.

[0093] In this embodiment, the structure of the binary tree is not described here. A binary tree has at least one parent node, and each parent node has several child nodes (the lower-level child nodes may serve as the parent nodes at this level, and there are several child nodes at the next lower level). The specific node levels can be configured according to needs.

[0094] The present invention defines and configures binary trees for various scenarios, thereby constructing corresponding prompt word binary trees or intelligence network binary trees, etc. By constructing a large language model prompt word in a tree structure, the parent and child nodes are configured with prompt words according to the scope and type of intelligence prompt words, and then input into the LLM to let the model perform data intelligence mining on the corresponding nodes (permissions and data ranges) based on the prompt words. Finally, analyze and generate a tree-structured data intelligence analysis result, which is clear at a glance and enables users to understand the analysis results with different levels of hierarchy.

[0095] During specific operations, the LLM can define each data analysis logic based on the keywords in the prompt words, and perform corresponding data mining, analysis, and verification operations to match and analyze the data mining results that conform to the prompt words. The steps for the LLM to perform data analysis based on the prompt words can refer to the following application descriptions of the LLM.

[0096] The present invention combines the LLM (Large Language Model) for intelligent data mining. The method of constructing intelligence prompt words to guide the model to perform multi-round intelligence data analysis and identify the final intelligence data mainly includes the following steps:

[0097] 1. Data collection and preprocessing

[0098] ‌Data collection‌: Collect data intelligence related to a specific topic from various data sources (such as databases, online repositories, documents, etc.) and save it in a database (data intelligence database, and several data intelligences can be saved and configured according to the business scope, etc. The specific database is specified by the user).

[0099] Data preprocessing: Clean the collected data, filter out noise, correct errors, and organize it into a structured format suitable for analysis. High-quality and clean data is crucial for ensuring model performance.

[0100] 2. Construct intelligence prompts

[0101] Define intelligence requirements: Clearly define the types and targets of intelligence to be mined.

[0102] Construct prompts: Based on the intelligence requirements, construct a series of intelligence prompts (construct a tree-structured large language model prompt, and configure the parent and child nodes according to the scope and type of the intelligence prompts). These prompts should be able to guide the LLM model to focus on the key information in the data and promote the model to conduct multi-round analysis.

[0103] 3. Application of LLM model

[0104] Select a suitable LLM model: Select a pre-trained large language model, such as GPT-3, GPT-4, etc. These models have been pre-trained on large-scale text data and possess powerful language understanding and generation capabilities.

[0105] Multi-round data analysis:

[0106] First-round analysis: Input the preprocessed data and the constructed intelligence prompts into the LLM model. The model will conduct a preliminary analysis of the data based on the prompts and generate the first-round analysis results.

[0107] Subsequent-round analysis: According to the results of the first-round analysis, adjust or supplement the intelligence prompts and input them into the model again for the next-round analysis. This process can be repeated multiple times until the final intelligence data is identified.

[0108] 4. Intelligence identification and extraction

[0109] Result evaluation: Evaluate the analysis results generated by the LLM model and identify the key information related to the intelligence requirements.

[0110] Intelligence extraction: Extract the final intelligence data from the analysis results and organize it into a format that is easy to understand and use. Here, the LLM model can be allowed to identify the data format of the corresponding nodes and automatically call the visualization component for data visualization processing, facilitating the display of the analysis results of each node in a tree structure.

[0111] Therefore, the present invention performs intelligence analysis and mining on the tree structure through the parent and child prompt words in the prompt word binary tree for data intelligence analysis by the LLM large language model. By traversing each intelligence analysis prompt word with a parent-child relationship on the input binary tree, the corresponding data intelligence with a tree structure is retrieved, analyzed, and mined. It can rely on the LLM large language model to quickly generate an intelligence network binary tree with a sense of hierarchy, which not only improves the efficiency of data intelligence mining and analysis, but also does not involve a large number of sample training and learning of machine learning algorithm models, reducing the complexity of data analysis. And finally, it can generate a data intelligence network with a tree structure, allowing users to conduct intelligence analysis and access from different levels, reducing the difficulty of intelligence analysis and understanding.

[0112] The application principle of the present invention will be further described below.

[0113] Preferably, the step S1 of constructing and configuring a prompt word binary tree for data intelligence analysis includes:

[0114] Import the binary tree;

[0115] Traverse the parent node and each child node on the binary tree;

[0116] According to the analysis dimension and content of the data intelligence to be analyzed, respectively construct the analysis data keywords, analysis range keywords, and analysis logic on the corresponding analysis dimension on the parent node and each child node;

[0117] Combining the analysis data keywords, the analysis range keywords, and the analysis logic on the corresponding analysis dimension, respectively construct the intelligence analysis prompt words corresponding to the parent node and each child node, and configure the intelligence analysis prompt words at the corresponding parent node or child node;

[0118] After the configuration is completed, the prompt word binary tree is obtained.

[0119] In the present invention, data intelligence analysis is analyzed in layers according to the hierarchy of parent-child nodes. Therefore, intelligence analysis prompts of corresponding levels are configured on the parent node and the corresponding child nodes under the parent node, so that the greater than model is based on the intelligence analysis prompts configured by each node, and the corresponding intelligence analysis scope and dimension and analysis logic level data intelligence analysis are performed on the corresponding node. For example, on a certain parent node, it is necessary to let the parent node analyze a certain format data intelligence, such as searching and analyzing the parent node data intelligence of the customer's additional purchase behavior of a certain commodity, and the data intelligence of the next analysis dimension of the child node below it can be for women or men to analyze the additional purchase behavior of the certain commodity, or within a certain month, a certain year, the repurchase rate analysis of the commodity, etc., so as to form a hierarchical analysis logic and scope. Therefore, the data intelligence analysis of the parent node and the child node is different in the analysis dimension and analysis scope, as well as the key data keywords.

[0120] Therefore, corresponding keywords and analysis logic on the corresponding hierarchical dimensions can be constructed here for the parent and child nodes respectively. Subsequently, the keywords and analysis logic of the corresponding nodes are combined to form intelligence analysis prompt words, and corresponding intelligence analysis prompt words are constructed and configured for each parent node or child node. Subsequently, the preset LLM large language model can be input to learn the keywords and analysis logic in the prompt words, and data intelligence analysis and mining of the corresponding prompt words are performed for each parent node or child node according to the intelligence analysis prompt words, so as to generate data intelligence analysis results with corresponding analysis scopes, analysis keywords and logic for each parent node and child node.

[0121] Preferably, S2, based on a recursive traversal algorithm, traverses the prompt word binary tree, reads the intelligence analysis prompt words on the parent and child nodes in sequence and inputs a preset LLM large language model, performs a database search and analysis based on the intelligence analysis prompt words through the LLM large language model, and mines and obtains the data intelligence analysis results corresponding to the parent node or the child node, including:

[0122] Starting from the parent node on the prompt word binary tree, the intelligence analysis prompt words configured on the parent node and the child node are read in sequence, wherein the recursive traversal algorithm :

[0123] ,

[0124] in:

[0125] is the target node to be traversed currently.

[0126] is the parent node of the current node N, initially Represents the root node, i.e., it has no parent node,

[0127] Are the left and right child nodes of the current node N respectively,

[0128] Is an operation sequence operator, indicating "execute the left operation first, and then the right operation",

[0129] Is to process the parent-child relationship and output or record the information of the parent node P and the child node N,

[0130] Is the root node, indicating that the node does not exist, and at this time the traversal terminates;

[0131] Input the read intelligence analysis prompt word into the preset LLM large language model, so that the LLM large language model performs a database search in the data intelligence database based on the intelligence analysis prompt word, analyzes and mines the data intelligence analysis results matching the intelligence analysis prompt word and outputs them;

[0132] Write and save the output data intelligence analysis result to the corresponding parent node or child node to obtain the data intelligence analysis result of the corresponding parent node or child node.

[0133] And here, the recursive traversal algorithm is used , traverse the parent node and each child node on the binary tree, let the large language model traverse the binary tree, so as to read the intelligence analysis prompt words configured on each node, and perform database mining analysis based on the input intelligence analysis prompt value, so as to mine the data intelligence matching the information in the prompt word and obtain the corresponding analysis result based on the analysis logic, and output and save it on the corresponding node.

[0134] The recursive traversal algorithm :

[0135] ,

[0136] Among them:

[0137] Is the target node to be traversed currently,

[0138] Is the parent node (the parent node of the current node N, initially Indicates that the root node has no parent node),

[0139] Are the left and right child nodes of the current node N respectively,

[0140] is an operation sequence operator, indicating "execute the left operation first and then the right operation".

[0141] To process the parent-child relationship, output or record the information of the parent node P and the child node N. When accessing a node, not only the current node is accessed, but also its parent node is recorded. For example, when traversing, the information of the parent node is passed, and the relationship between the parent node and the child node is processed simultaneously when accessing. Here, it is necessary to determine the processing relationship (processing the parent-child relationship) between the parent node and the child node according to the analysis logic in the "intelligence analysis prompt words" configured on the parent node and the child node respectively. For example, the child node contains "additional purchase behavior analysis", and the parent node contains "sales data of commodity formats", and the "additional purchase behavior analysis" needs to rely on the "sales data of commodity formats" for analysis and calculation. Therefore, at this time, the access can record the relationship between the parent node and the child node, and clarify that the additional purchase behavior needs to be obtained after the sales data of commodity formats;

[0142] Is the root node, indicating that the node does not exist and the traversal terminates;

[0143] Recursive traversal algorithm , the logic is as follows:

[0144] 1. Input: the current node N and its parent node P;

[0145] 2. Termination condition: If the current node N is empty( ), then return directly;

[0146] 3. Process the parent-child relationship

[0147] Call , record the relationship between the parent node P and the child node N (for example, print P N). If P= , it means that N is the root node, and at this time, the root node information can be processed separately; specifically, the processing relationship of the parent-child relationship is carried out in combination with the keyword logic of the accessed node.

[0148] 4. Recursively traverse the subtree:

[0149] First, recursively traverse the left subtree: ;

[0150] Then, recursively traverse the right subtree: .

[0151] Features of this algorithm

[0152] 1. Explicit parent-child relationship: Each time a node is accessed, the relationship between its parent node and the current node is recorded;

[0153] 2. Depth - first traversal: Implement the logic of first visiting the parent node and then sequentially traversing the left and right sub - trees through recursion.

[0154] The large - language model can perform database retrieval based on prompt words: First, mine the data intelligence that matches the analysis dimensions, analysis data keywords, and analysis scope keywords in the prompt words, and further perform in - depth analysis on the data intelligence based on the analysis logic therein to generate the data intelligence analysis results under this dimension and permissions. Then, write the mined data intelligence and the corresponding data intelligence analysis results to the corresponding parent node or child node.

[0155] In the background database, corresponding data storage addresses can be assigned to each parent node or child node. After the large - language model retrieves and analyzes the database to obtain the corresponding data intelligence and the corresponding analysis results, the results can be saved to the storage address of this node.

[0156] Here, a distributed storage system can also be used, such as the storage method of the HDFS distributed system, to store the data intelligence and its analysis results of each node, which is convenient for distributed management and storage.

[0157] Preferably, S3, according to the relationship between parent and child nodes, sorts based on the amount of information and information importance, and orderly counts the data intelligence analysis results of each node and generates the corresponding binary intelligence tree, including:

[0158] Calculate the amount of information and information importance of each node and sort them. Based on the sorting results, traverse and count the data intelligence analysis results at the parent node and each child node; among them, the amount of information and information importance are respectively expressed as and :

[0159] ,

[0160] Among them:

[0161] N: The current node,

[0162] A i : The i - th attribute of node N,

[0163] k: The total number of attributes of node N,

[0164] w i : The weight of the i - th attribute,

[0165] Complexity(A i ): The complexity of the i - th attribute, and the calculation method is as follows:

[0166] If the attribute value is of numerical type: Complexity(Ai ) = log2(1 + |A i |);

[0167] If the attribute value is of the categorical type: Complexity(A i ) = Entropy(A i ), where Entropy(A i ) is the entropy value of the categorical attribute;

[0168] ,

[0169] where:

[0170] Depth(N): The depth of node N in the tree, with the root node having a depth of 0,

[0171] Degree(N): The degree of connection of node N,

[0172] Centrality(N): The centrality of node N (betweenness centrality or closeness centrality),

[0173] : The weight coefficient, used to adjust the contribution of each item to the importance;

[0174] Import a new binary tree A and configure the corresponding parent and child nodes of the new binary tree A according to the relationship between the parent and child nodes on the binary tree of the prompt words;

[0175] Write the data intelligence analysis results of the extracted parent node and each child node to the new binary tree A to obtain the intelligence network binary tree.

[0176] After having the large language model traverse the prompt words of each node on the binary tree of the prompt words and perform database query analysis, and output the corresponding data intelligence analysis results, in order to output the data intelligence analysis results of each node in an orderly manner and perform statistics. A new binary tree structure is separately constructed here. This new binary tree structure will be separately used for visualizing and displaying the data intelligence analysis results of each node extracted above. Because the previous binary tree of the prompt words is mainly used to construct the prompt words for the model to traverse, if it is used for display again, there will be an alternation of data tasks and possible chaos. Here, the traversal retrieval analysis and result display of the prompt words are respectively operated using binary trees, which can avoid direct data mixing between binary trees and reduce the structural complexity of each binary tree and the degree of data redundancy (because the data analysis of one business form contains data results of several hierarchical structures and the data volume is huge).

[0177] Therefore, a new binary tree A can be imported, and according to the relationship between the parent and child nodes on the binary tree of the above prompt words, the relationship between the parent and child nodes of the new binary tree can be configured. By adopting the above method, a new binary tree A can be quickly constructed, and the data intelligence analysis results corresponding to the nodes can be located, copied and saved on each node of the new binary tree according to the relationship between the parent and child nodes on its original prompt word binary tree for analysis result display, and an intelligence network binary tree for visualizing the data intelligence analysis results can be obtained. This intelligence network binary tree directly displays the analysis results of the corresponding nodes and is displayed through a hierarchical data structure, enabling users to quickly analyze the data intelligence at each level.

[0178] Among them, in order to effectively quantify and statistically analyze the information characteristics of each node, and conduct an orderly statistical analysis of the intelligence information of the parent and child nodes to avoid subsequent data management chaos, the data volume and information importance of each node are combined here to orderly statistically analyze the data intelligence of each node. Each node can be marked according to the amount of information and information importance of each node, which is convenient for users to visually analyze and identify the attributes (amount of information and importance) of the data intelligence information of the parent and child nodes. And based on the amount of information and importance, convenient extraction of node information can be carried out, which can also facilitate system classification statistics and storage.

[0179] Combining the information amount size and information importance formula can be used in the traversal algorithm:

[0180] 1. Calculate the information amount: Use to calculate the information amount of each node.

[0181] 2. Calculate the importance: Use to calculate the importance of each node.

[0182] 3. Sort and extract: Extract important information according to importance, and then extract other information according to the information amount.

[0183] The above 、 example is as follows:

[0184] Suppose node N has the following attributes:

[0185] Attribute A1 (numerical type): The value is 10.

[0186] Attribute A2 (categorical type): The value is {"High","Low","Medium"}, and the distribution is P ={0.5,0.3,0.2}

[0187] Calculation process:

[0188] 1. Complexity(A1)=log(1 + 10)≈3.46;

[0189] 2. Entropy(A2) =

[0190] ≈ 1.57;

[0191] 3. Assume the weights w1 = 1, w2 = 2;

[0192] 4. Volume(N) = 1·3.46 + 2·1.57 = 6.60.

[0193] Assume the following information of node N:

[0194] Depth(N) = 2,

[0195] Degree(N) = 3,

[0196] Centrality(N) = 0.8,

[0197] Weight coefficients α = 0.5, β = 0.3, γ = 0.2,

[0198] Importance calculation process:

[0199] .

[0200] In this embodiment, the weight coefficients w i , α, β, γ can be adjusted according to the specific scenario.

[0201] The present invention also performs data verification on the generated data intelligence and its results in a binary tree structure. Based on the above-mentioned method of traversing each node of the binary tree by the large language model for data analysis, a data intelligence verification prompt is set here, allowing the large language model to traverse and verify the correctness of the data intelligence and its analysis results of each node. If a node fails the verification, the node is marked as abnormal, facilitating the user to quickly locate the nodes with abnormal data intelligence and results, and perform node data intelligence analysis to quickly find the cause.

[0202] Preferably, in S4, traversing and verifying the binary tree of the intelligence network, and after passing the verification, outputting and saving the binary tree of the intelligence network, includes:

[0203] Pre-configuring corresponding data intelligence verification rules for the parent node and each child node;

[0204] Using the LLM large language model to extract the verification keywords, verification scope keywords and corresponding verification logic in the data intelligence verification rules;

[0205] Construct intelligence verification prompt words corresponding to the parent node and each child node respectively by combining the verification keyword, the verification scope keyword, and the corresponding verification logic, and configure the intelligence verification prompt words at the corresponding parent node or child node;

[0206] Traverse the data intelligence analysis results and the intelligence verification prompt words at the parent node or child node of the intelligence network binary tree, and input the intelligence verification prompt words into the preset large language model (LLM);

[0207] Through the LLM, based on the intelligence verification prompt words at the parent node or child node, conduct data intelligence verification on the data intelligence analysis results at the parent node or child node:

[0208] If the verification passes, mark the parent node or child node as the passed status;

[0209] Otherwise, mark it as the warning status;

[0210] After the verification is completed, output and save the intelligence network binary tree with status marks.

[0211] Here, it is necessary to use the large language model to conduct intelligence verification on the data intelligence and its analysis results of each parent node and child node on the binary tree. Therefore, an intelligence network binary tree can be generated for verification according to the construction method of the above prompt word binary tree. Based on the above intelligence network binary tree, corresponding data intelligence verification rules can be configured for each parent node and child node. The construction method of the above prompt word binary tree can be referred to, and verification rules for corresponding ranges can be configured for each node, and further use data mining algorithms to extract the verification keywords, verification scope keywords, and corresponding verification logic. Further combine the verification keywords, verification scope keywords, and corresponding verification logic to generate corresponding intelligence verification prompt words, and configure them at the corresponding parent node or child node. Because the corresponding intelligence verification prompt words, corresponding data intelligence, and corresponding data analysis results have been configured and saved on the intelligence network binary tree, subsequently, the large language model can be allowed to traverse the binary tree, extract the intelligence analysis results and intelligence verification prompt words of the corresponding nodes, input the intelligence verification prompt words into the large language model, and let the large language model conduct data intelligence verification on the data intelligence and its analysis results at each node based on the information in the intelligence verification prompt words.

[0212] For the data intelligence verification rules of each node, corresponding intelligence verification prompt words can be constructed based on the data processing scope, logic, keywords, etc. set for each node in the early stage. Therefore, based on the above intelligence network binary tree, this department combines the keywords and logic constructed for each node to generate corresponding intelligence verification prompt words, enabling the model to perform data intelligence verification on the data intelligence analysis results of each node based on the intelligence verification prompt words of each node. For the analysis results of nodes that pass the verification, a passed status mark can be added to the node; for the analysis results of nodes that fail the verification, an alarm status mark can be added. When performing visual display later, the intelligence network binary tree is used to perform corresponding status visual display, quickly locating the nodes in abnormal status.

[0213] Specific visualization techniques can call corresponding visualization components or programs, etc., to achieve corresponding visualization rendering and generation.

[0214] Figure 3 It is a block diagram of a data intelligence intelligent analysis device based on the LLM large language model shown according to an exemplary embodiment. This device is used for the data intelligence intelligent analysis method based on the LLM large language model. Refer to Figure 3 This device includes a prompt word construction module 310, an intelligence mining module 320, an intelligence statistics module 330, and an intelligence verification module 340. Among them:

[0215] The prompt word construction module is used to construct and configure a prompt word binary tree for data intelligence analysis. Among them, the prompt word binary tree includes a parent node and several child nodes, and the parent node and the child nodes are respectively configured with intelligence analysis prompt words with corresponding permissions;

[0216] The intelligence mining module is used to traverse the prompt word binary tree based on the recursive traversal algorithm, sequentially read the intelligence analysis prompt words on the parent and child nodes and input them into the preset LLM large language model, and perform database analysis through the LLM large language model based on the intelligence analysis prompt words to mine the data intelligence analysis results corresponding to the parent node or the child node;

[0217] The intelligence statistics module is used to sort based on the amount of information and information importance according to the relationship between the parent and child nodes, and orderly count the data intelligence analysis results of each node and generate a corresponding intelligence network binary tree;

[0218] The intelligence verification module is used to traverse and verify the intelligence network binary tree, and output and save the intelligence network binary tree after passing the verification.

[0219] Please understand the above-mentioned modules in combination with the corresponding steps in the above method.

[0220] Figure 4 This is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. As Figure 4 shown, the electronic device may include the above-mentioned Figure 3 data intelligence intelligent analysis device based on the LLM large language model as shown. Optionally, the electronic device 410 may include a first processor 2001.

[0221] Optionally, the electronic device 410 may further include a memory 2002 and a transceiver 2003.

[0222] Among them, the first processor 2001 is connected to the memory 2002 and the transceiver 2003, and can be connected through a communication bus, for example.

[0223] Next, in combination with Figure 4 each component of the electronic device 410 will be specifically introduced:

[0224] Among them, the first processor 2001 is the control center of the electronic device 410, and can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or can also be an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. For example: one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs).

[0225] Optionally, the first processor 2001 can execute various functions of the electronic device 410 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0226] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 4 the CPU0 and CPU1 shown in

[0227] In a specific implementation, as an embodiment, the electronic device 410 may also include multiple processors, such as Figure 4The first processor 2001 and the second processor 2004 shown in []. Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here can refer to one or more devices, circuits, and / or processing cores for processing data (such as computer program instructions).

[0228] Among them, the memory 2002 is used to store the software program for implementing the solution of the present invention and is controlled by the first processor 2001 for execution. The specific implementation manner can refer to the above method embodiments and will not be elaborated here.

[0229] Optionally, the memory 2002 can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic storage medium or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 2002 can be integrated with the first processor 2001 or exist independently and is coupled to the first processor 2001 through the interface circuit of the electronic device 410 ( Figure 4 not shown in []). The embodiments of the present invention do not make specific limitations on this.

[0230] The transceiver 2003 is used to communicate with a network device or with a terminal device.

[0231] Optionally, the transceiver 2003 can include a receiver and a transmitter ( Figure 4 not shown separately in []). Among them, the receiver is used to implement the receiving function, and the transmitter is used to implement the sending function.

[0232] Optionally, the transceiver 2003 can be integrated with the first processor 2001 or exist independently and is coupled to the first processor 2001 through the interface circuit of the electronic device 410 ( Figure 4 not shown in []). The embodiments of the present invention do not make specific limitations on this.

[0233] It should be noted thatFigure 4 The structure of the electronic device 410 shown does not constitute a limitation on the router. The actual knowledge structure recognition device may include more or fewer components than shown, or combine certain components, or have a different component arrangement.

[0234] In addition, the technical effects of the electronic device 410 can refer to the technical effects of the data intelligence intelligent analysis method based on the LLM large language model described in the above method embodiments, and will not be elaborated here.

[0235] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), and this processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0236] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0237] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more collections of available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0238] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood with reference to the context before and after.

[0239] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0240] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0241] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled professionals may use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0242] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the devices, apparatuses, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0243] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0244] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0245] In addition, the functional units in each embodiment of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0246] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0247] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A data intelligence intelligent analysis method based on LLM large language model, characterized in that: The method comprises: S1. Construct and configure a binary tree of prompt words for data intelligence analysis, wherein the binary tree of prompt words includes a parent node and a plurality of child nodes, and the parent node and the child node are respectively configured with intelligence analysis prompt words of corresponding authority; S2, traversing the prompt word binary tree based on a recursive traversal algorithm, reading the intelligence analysis prompt words on the parent and child nodes in sequence and inputting them into a preset LLM large language model, performing a database search and analysis based on the intelligence analysis prompt words through the LLM large language model, and mining to obtain the data intelligence analysis results corresponding to the parent node or the child node; S3, according to the relationship between the parent and child nodes, based on the amount of information and the order of information importance, the data intelligence analysis results of each node are counted in order and a corresponding intelligence network binary tree is generated; S4, traversing and verifying the intelligence network binary tree, and outputting and saving the intelligence network binary tree after verification, including: Pre-configure corresponding data intelligence verification rules for the parent node and each of the child nodes; Utilize the LLM large language model to extract verification keywords, verification scope keywords and corresponding verification logic in the data intelligence verification rules; Combining the verification keyword, the verification scope keyword and the corresponding verification logic, constructing intelligence verification prompt words corresponding to the parent node and each of the child nodes, and configuring the intelligence verification prompt words at the corresponding parent node or the child node; Traversing the data intelligence analysis results and the intelligence verification prompt words at the parent node or the child node on the intelligence network binary tree, and inputting the intelligence verification prompt words into the preset LLM large language model; The data intelligence analysis result at the parent node or the child node is verified by the LLM large language model based on the intelligence verification prompt word at the parent node or the child node: If the verification is successful, the parent node or the child node is marked as a passed state; Otherwise it is marked as an alarm state; After verification is completed, the intelligence network binary tree with status marks is output and saved.

2. The data intelligence intelligent analysis method based on the LLM large language model according to claim 1 is characterized in that: The step S1, constructing and configuring a binary tree of prompt words for data intelligence analysis, includes: Import binary tree; Traversing the parent node and each of the child nodes on the binary tree; According to the analysis dimensions and contents of the data intelligence to be analyzed, respectively construct analysis data keywords, analysis scope keywords and analysis logic on the corresponding analysis dimensions on the parent node and each of the child nodes; Combining the analysis data keywords, the analysis scope keywords and the analysis logic on the corresponding analysis dimension, constructing the intelligence analysis prompt words corresponding to the parent node and each of the child nodes respectively, and configuring the intelligence analysis prompt words at the corresponding parent node or the child node; After the configuration is completed, the prompt word binary tree is obtained.

3. The data intelligence intelligent analysis method based on the LLM large language model according to claim 1 is characterized in that: S2, traversing the prompt word binary tree based on a recursive traversal algorithm, reading the intelligence analysis prompt words on the parent and child nodes in sequence and inputting a preset LLM large language model, performing a database search and analysis based on the intelligence analysis prompt words through the LLM large language model, and mining to obtain data intelligence analysis results corresponding to the parent node or the child node, including: Starting from the parent node on the prompt word binary tree, the intelligence analysis prompt words configured on the parent node and the child node are read in sequence, wherein the recursive traversal algorithm : in: is the target node to be traversed currently. is the parent node of the current node N, initially Represents the root node, that is, no parent node, are the left and right child nodes of the current node N respectively, is the operation sequence symbol, which means "execute the left operation first, then the right operation". To process the parent-child relationship, output or record the information of the parent node P and the child node N. is the root node, indicating that the node does not exist and the traversal terminates; Input the read intelligence analysis prompt word into the preset LLM large language model, so that the LLM large language model searches the data intelligence database based on the intelligence analysis prompt word, analyzes and mines the data intelligence analysis results that match the intelligence analysis prompt word, and outputs them; The output data intelligence analysis result is written and saved to the corresponding parent node or the child node to obtain the data intelligence analysis result corresponding to the parent node or the child node.

4. The data intelligence intelligent analysis method based on the LLM large language model according to claim 3 is characterized in that: S3, according to the relationship between the parent and child nodes, based on the information volume and information importance sorting, orderly counting the data intelligence analysis results of each node and generating a corresponding intelligence network binary tree, including: The amount of information and the importance of information of each node are calculated and sorted, and the data intelligence analysis results at the parent node and each of the child nodes are traversed and counted based on the sorting results; wherein the amount of information and the importance of information are respectively expressed as and : ， in: N: current node, A i : The i-th attribute of node N, k: the total number of attributes of node N, w i : The weight of the i-th attribute, Complexity(A i ): The complexity of the i-th attribute is calculated as follows: If the attribute value is a numeric type: Complexity (A i ) = log2(1 +|A i |); If the attribute value is of type Complexity (A i ) = Entropy(A i ), where Entropy(A i ) is the entropy value of the categorical attribute; ， in: Depth(N): The depth of node N in the tree, where the depth of the root node is 0. Degree(N): the connectivity of node N, Centrality(N): The centrality of node N, where centrality is betweenness centrality or closeness centrality. : Weight coefficient, used to adjust the contribution of each item to importance; Import a new binary tree A, and configure corresponding parent and child nodes of the new binary tree A according to the relationship between the parent and child nodes on the prompt word binary tree; The data intelligence analysis results of the parent node and each of the child nodes obtained by traversal extraction are written into the new binary tree A to obtain the intelligence network binary tree.

5. A data intelligence intelligent analysis device based on the LLM large language model, the data intelligence intelligent analysis device based on the LLM large language model is used to implement the data intelligence intelligent analysis method based on the LLM large language model as claimed in any one of claims 1 to 4, characterized in that: The device comprises: A prompt word construction module is used to construct and configure a prompt word binary tree for data intelligence analysis, wherein the prompt word binary tree includes a parent node and a plurality of child nodes, and the parent node and the child node are respectively configured with intelligence analysis prompt words of corresponding authority; An intelligence mining module is used to traverse the prompt word binary tree based on a recursive traversal algorithm, read the intelligence analysis prompt words on the parent and child nodes in turn and input them into a preset LLM large language model, perform a database search and analysis based on the intelligence analysis prompt words through the LLM large language model, and mine and obtain the data intelligence analysis results corresponding to the parent node or the child node; An intelligence statistics module is used to sequentially count the data intelligence analysis results of each node and generate a corresponding intelligence network binary tree according to the relationship between parent and child nodes and based on the information volume and information importance sorting; The intelligence verification module is used to traverse and verify the intelligence network binary tree, and output and save the intelligence network binary tree after verification.

6. An electronic device, characterized in that: The electronic device comprises: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program codes, which can be called by a processor to execute the method according to any one of claims 1 to 4.

Images