Intelligent data generation method, system and equipment based on large model and medium

Abstract

The invention provides an intelligent data generation method, system and device based on a large model and a medium, and the method comprises the steps: obtaining dialogue information, processing the dialogue information based on an LLM language model to determine key information, and determining a retrieval intention type according to the key information; a plurality of retrieval tasks related to the retrieval intention type are executed based on a multi-path recall algorithm, each retrieval task retrieves from a preset database to obtain a retrieval result related to the key information, and the retrieval results are reordered to obtain target data; and calling a tool component associated with the retrieval intention type, and processing the target data through the tool component to generate a data report related to the target data. According to the method, the accuracy and the efficiency of data retrieval can be improved, so that flexible and changeable business requirements are met.

Description

Technical Field

[0001] The present invention relates to the field of data processing and artificial intelligence technology, and in particular to a large model-based intelligent data generation method, system, device and medium. Background Art

[0002] In today's highly competitive business environment, the fund management of enterprises is crucial. Effective fund management can not only ensure the normal operation of the enterprise, but also provide strong support for the development of the enterprise. In order to achieve efficient fund management, enterprises need to build business reports, data views and data warehouses of various dimensions to deeply explore the value of treasury data assets and provide accurate and timely information support for fund management decisions.

[0003] However, with the continuous development of corporate business and the sharp increase in data volume, traditional treasury report generation methods face many challenges. First, traditional methods usually rely on manual operations and fixed templates, which are not only inefficient but also difficult to meet flexible and changing business needs. For example, when an enterprise needs to generate customized reports based on specific business scenarios, manual operations often take a lot of time and energy, and are prone to errors and omissions. Secondly, due to the complexity and diversity of treasury data, manual screening and analysis of data are prone to errors and omissions, affecting the accuracy and timeliness of decision-making. In addition, traditional retrieval methods make it difficult to quickly and accurately find the required information in massive data, reducing work efficiency. Summary of the invention

[0004] The embodiment of the present invention provides an intelligent data generation method based on a large model to solve the problems existing in the related technologies. The technical solution is as follows:

[0005] In a first aspect, an embodiment of the present invention provides a method for generating intelligent data based on a large model, comprising:

[0006] Obtain conversation information, process the conversation information based on the LLM language model to determine key information, and determine the type of search intent based on the key information;

[0007] Based on the multi-way recall algorithm, multiple retrieval tasks related to the types of retrieval intentions are executed. Each retrieval task retrieves retrieval results associated with key information from a preset database, and re-ranks the retrieval results to obtain target data;

[0008] Call the tool component associated with the search intent type, process the target data through the tool component, and generate a data report related to the target data.

[0009] In one embodiment, the retrieval intent includes table file retrieval, graph file retrieval, real-time data retrieval, specific data indicator query intent, data range limitation retrieval intent and data comparison intent.

[0010] In one embodiment, it further includes:

[0011] Based on the data paging algorithm, treasury data is obtained in batches. The treasury data includes financial data, business data, and transaction data.

[0012] Determine the outliers in the treasury data based on the anomaly detection algorithm, clean the outliers, and obtain valid data;

[0013] Identify the characteristic information of valid data, label and classify the valid data according to the characteristic information, and obtain the target treasury data;

[0014] The target treasurer data is stored in a database to obtain a preset database.

[0015] In one embodiment, it further includes:

[0016] When the search intent type is real-time data search, connect to the real-time data source;

[0017] Get the latest data related to key information from real-time data sources for display.

[0018] In one embodiment, the multi-way recall algorithm includes a sparse BM25 recall algorithm, a dense vector recall algorithm, and a large model rewrite recall algorithm.

[0019] In one embodiment, it further includes:

[0020] Determine the annotation requirements based on the conversation information, and determine the report nodes and annotation contents that need to be annotated based on the annotation requirements;

[0021] The corresponding report nodes in the data report are automatically marked with corresponding annotation contents to obtain an updated data report.

[0022] In one embodiment, the tool component includes a graphic component, which is used to run a specified graphic drawing algorithm to draw a graphic file related to the target data, and the graphic file includes a line chart, a bar chart, and a pie chart.

[0023] In a second aspect, an embodiment of the present invention provides an intelligent data generation method based on a large model.

[0024] In a third aspect, an embodiment of the present invention provides an electronic device, the device comprising: a memory and a processor. The memory and the processor communicate with each other through an internal connection path, the memory is used to store instructions, the processor is used to execute the instructions stored in the memory, and when the processor executes the instructions stored in the memory, the processor executes the method in any one of the above-mentioned embodiments.

[0025] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a computer, the method in any one of the above-mentioned embodiments is executed.

[0026] The advantages or beneficial effects of the above technical solution include at least:

[0027] The present invention adopts the natural interactive form of dialogue to intelligently analyze the types of retrieval intent, enters a multi-round and multi-channel data file recall process according to the types of retrieval intent to complete the corresponding retrieval task, obtains the target data related to the dialogue keywords, and then generates corresponding data reports based on the target data through tool components for users to view, thereby improving the accuracy and efficiency of data retrieval to meet flexible and changeable business needs.

[0028] The above summary is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments and features described above, further aspects, embodiments and features of the present invention will be readily apparent by reference to the accompanying drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the multiple drawings represent the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments disclosed in the present invention and should not be regarded as limiting the scope of the present invention.

[0030] Figure 1 It is a flow chart of the intelligent data generation method based on a large model of the present invention;

[0031] Figure 2 The structure block diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.

[0033] In today's highly competitive business environment, the fund management of enterprises is crucial. Effective fund management can not only ensure the normal operation of the enterprise, but also provide strong support for the development of the enterprise. In order to achieve efficient fund management, enterprises need to build business reports, data views and data warehouses of various dimensions to deeply explore the value of treasury data assets and provide accurate and timely information support for fund management decisions.

[0034] However, with the continuous development of corporate business and the sharp increase in data volume, traditional treasury report generation methods face many challenges. First, traditional methods usually rely on manual operations and fixed templates, which are not only inefficient but also difficult to meet flexible and changing business needs. For example, when an enterprise needs to generate customized reports based on specific business scenarios, manual operations often take a lot of time and energy, and are prone to errors and omissions. Secondly, due to the complexity and diversity of treasury data, manual screening and analysis of data are prone to errors and omissions, affecting the accuracy and timeliness of decision-making. In addition, traditional retrieval methods make it difficult to quickly and accurately find the required information in massive data, reducing work efficiency.

[0035] In order to solve the above problems, an embodiment of the present invention provides an intelligent data generation method based on a large model, which can improve the accuracy and efficiency of data retrieval and meet the needs of enterprises to quickly obtain required information from massive data.

[0036] like Figure 1 As shown, the intelligent data generation method based on the large model in this embodiment includes:

[0037] Step S1: Acquire dialogue information, process the dialogue information based on the LLM language model, determine key information, and determine the type of retrieval intent based on the key information.

[0038] It should be noted that conversation information refers to voice / text conversations captured based on LLM (Large Language Models). Voice / text conversations can be voice data or text data in multiple rounds of conversations. Through multiple rounds of conversations, users can gradually add details to a search requirement, and ultimately obtain key information that better meets user needs and accurately understand the user's search intent.

[0039] For example, in the first round of conversation, the user inputs by voice or text: I want the financial data report of a certain province or city. Intelligent data retrieval returns the first round of search results. If the user wants to further search for the financial data report of a certain district in the province or city based on the previous results, he can continue to input by voice or text: only look at a certain district. This directly starts the second round of search and obtains the second round of search results. And so on, it can support multiple rounds of search recall.

[0040] Based on the pre-built LLM language model, natural language processing is performed on the user's conversation information to filter out important elements such as nouns, verbs, and time words in the conversation information, and the corresponding key information is extracted based on the important elements. The key information mainly refers to keywords. Since the user's input is diverse, in order to retrieve data efficiently in the future, the user's input must first be analyzed for retrieval intent.

[0041] This embodiment determines the type of user's search intent based on the extracted keywords. The types of search intent include chatting, table file retrieval, graph file retrieval, real-time data retrieval, specific data indicator query intent, data range limitation retrieval intent and data comparison intent.

[0042] For example, if the keywords obtained by natural language processing include words related to weather, greetings, daily topics, etc., such as "today", "how are you", "hello", etc., it can be preliminarily judged as a chat intent. If words such as "chart", "graph", "bar chart" appear, it may be a picture file retrieval intent; if words such as "table", "data table", "data list" appear, it may be a table file retrieval intent; if words such as "real-time", "current", "latest" appear, it may be a real-time data retrieval intent. For specific data indicator query intent, keywords of specific data indicator names will appear; data range-limited retrieval intent will have clear time range, geographical range and other expression keywords; data comparison intent will have keywords such as "comparison" and "compared with...".

[0043] If the user's input is short and vague, you can ask more information to determine the exact type of search intent. For example, when the user enters "data", you can ask "Do you need tabular data, graphical data, or real-time data? Or specific data indicators?", and combine the conversation information in multiple rounds of conversations with the context to further confirm the user's search intent.

[0044] Among them, for the intention of chatting, the prompt instruction of the LLM language model can be activated to explain to the user the purpose and boundaries of the retrieval module, and guide the user to correctly input relevant retrieval requirements.

[0045] Step S2: Execute multiple retrieval tasks related to the types of retrieval intentions based on the multi-way recall algorithm. Each retrieval task retrieves retrieval results associated with key information from a preset database, and re-sorts the retrieval results to obtain target data.

[0046] This embodiment improves the data recall rate by using a hybrid multi-way recall algorithm based on sparse BM25, dense vector (Embedding) and large model rewriting of retrieval statements.

[0047] Among them, the main advantages of BM25 are high efficiency and high keyword hit recall rate. The advantage of Embedding is that it can capture semantic information and facilitate matching semantically similar results. The big model rewrites the search statement, which uses the big model technology to rewrite the search statement entered by the user, remove irrelevant words, and highlight the focus of the search.

[0048] Specifically, the sparse BM25 recall algorithm is used to calculate the relevance score between the keywords in the conversation information and each data document in the preset database, and the document with the highest score is recalled from the preset database.

[0049] The dense vector recall algorithm converts the keywords in the conversation information and the data documents in the preset database into dense vectors (Embedding), and recalls semantically similar documents in the preset database by calculating the similarity between the keyword vector and the document vector (such as cosine similarity).

[0050] The big model rewrite recall algorithm uses the big model to rewrite the user's query statements, optimize the query statements, use the rewritten query statements to search in the preset database, and recall relevant documents.

[0051] It should be noted that the sparse BM25 recall algorithm, dense vector recall algorithm and large model rewrite recall algorithm have been disclosed in the prior art, and the principles and operating logic of each algorithm will not be described in detail here.

[0052] This embodiment uses a multi-way recall algorithm that integrates the sparse BM25 recall algorithm, the dense vector recall algorithm, and the large model rewrite recall algorithm to simultaneously execute multiple retrieval tasks. The corresponding retrieval results (document list) can be obtained by executing the retrieval task corresponding to each recall algorithm. Finally, the retrieval results obtained by executing each retrieval task are re-ranked through the rerank algorithm to obtain the target data.

[0053] Among them, the method of re-ranking the search results is:

[0054] The lists obtained by the above three recalls are merged into a large candidate set, and features are extracted for each document in the candidate set. The features include BM25 score, Embedding similarity score, large model rewrite score, etc.

[0055] Use historical data to train a rerank model, which can predict the relevance of documents based on features. For each document in the candidate set, combine the BM25 score, Embedding similarity score, and large model rewrite score, and use the rerank model to calculate the comprehensive score of each document.

[0056] The documents in the candidate set are sorted according to the comprehensive scores, and the top N documents with the highest scores are selected as the final recall results.

[0057] This embodiment uses the multi-way recall combined with the rerank re-ordering method to fully utilize the advantages of different recall technologies, improve retrieval accuracy and efficiency, and ultimately output target data that better meets user needs.

[0058] It should be noted that the preset database needs to be built in advance, specifically:

[0059] Based on the data paging algorithm, various treasury data are obtained from the enterprise's treasury system in batches. The treasury data includes financial data, business data, transaction data, etc.; based on the anomaly detection algorithm, the outliers of the treasury data are determined, and the outliers are cleaned to obtain valid data; the characteristic information of the valid data is identified, and the valid data is labeled and classified according to the characteristic information to obtain the target treasury data; the target treasury data is stored in the database to obtain the preset database.

[0060] Data collection can be achieved through interface calls (such as using RESTful API technology), database connections (such as using JDBC technology to connect to relational databases), or file imports (using Java's I / O stream or Python's related libraries). During the collection process, a large amount of data is divided into multiple smaller, manageable data sets using a data paging algorithm, which are processed and transmitted in batches to obtain a large amount of treasury data and avoid memory overflow. Efficient data collection is achieved by writing interface call code, configuring database connection parameters, and reading and parsing the format of imported files.

[0061] After collecting the treasury data, the collected data is cleaned, converted and integrated to remove noise and outliers. Outliers can be obtained using anomaly detection algorithms such as the statistical 3σ rule. The processed data is formatted in a unified manner to obtain valid data, which can be suitable for subsequent model processing. At the same time, the supervised learning algorithm in machine learning can be used to annotate and classify the processed valid data, providing a basis for subsequent model training and retrieval, and improving data quality.

[0062] Step S3: Call the tool component associated with the search intent type, process the target data through the tool component, and generate a data report related to the target data.

[0063] When the search intent type is table file search intent or graph file search intent, the corresponding tool component is the text graph component. The text graph component is used to run the specified graph drawing algorithm to draw graph files related to the target data. The graph files include line graphs, bar graphs, and pie charts.

[0064] When it is determined that the user's search intention is to search for table files, the system searches for relevant table files in the preset database based on the keywords or data descriptions provided by the user. Matching can be performed by keywords such as file name, table title, column name, etc. For example, if the user enters "2023 sales data table", the system will search the database for table files with "2023 sales" in the file name or relevant expressions in the table title, and generate corresponding graphic files based on the table files.

[0065] If multiple related table files are found, they can be filtered according to the user's further needs, such as sorting and displaying by time range, data range, etc.

[0066] For the image file retrieval intent, the database is also searched for relevant image files, such as bar graphs, line graphs, pie charts, etc., based on the keywords provided by the user. Matching can be performed by file name, graphic title, legend, etc. For example, if the user enters "market share bar graph", the system will search for files with "market share bar graph" in the file name or with related expressions in the graphic title.

[0067] When it is determined that the user's intention is to retrieve real-time data, the specified real-time data source is connected, such as sensor data, trading system data, etc. According to the keywords entered by the user, the latest relevant data is obtained from the real-time data source, and then processed and displayed. For example, if the user enters "current stock price", the system will immediately connect to the stock trading system, obtain the latest stock price information from the stock trading system, and display it to the user.

[0068] Similarly, other search intents are also displayed by searching for corresponding files from the preset database based on keywords, which are not explained here one by one.

[0069] Furthermore, in order to achieve better interactive effects, the interactive module of this embodiment provides a friendly interactive interface, and obtains the required treasury reports, views, real-time data and other target data through the intelligent retrieval function of the previous step based on the user input, and passes them to the report generation module to generate corresponding graphic files. The interactive module uses front-end frameworks such as Vue.js or React to build a friendly interface, and by designing the layout and interactive elements of the interactive interface, writing the function of demand parsing, and realizing the communication interface with the report generation module, the user experience is improved.

[0070] The intelligent report generation module uses large model Agent technology to automatically generate treasury reports or views based on the user's demand description, retrieved data and various tool components configured in the backend. When the user enters requirements related to graphic generation, the Wenshengtu component uses advanced image generation algorithms, combined with the characteristics of treasury data and the user's specific requirements, to quickly generate high-quality graphics for display. The displayed graphic files can also provide functions such as zooming in, zooming out, and annotation to facilitate users to view details.

[0071] For example, when a user inputs "help me generate a line graph showing the company's capital flow trend", the Wenshengtu component first analyzes the key information in the user's needs, including "company capital flow" and "line graph". Then, it extracts data related to capital flow from the retrieved data, and determines the coordinate axis range and scale of the line graph based on the time series and numerical changes of these data. Next, the Wenshengtu component uses professional graphics drawing algorithms to draw a clear and accurate line graph to show the company's capital flow trend. During the graphics generation process, the Wenshengtu component can also be adjusted and optimized according to the user's further requirements. For example, determine whether there is a need for annotation from the user's dialogue information. If the user requires annotations to be added to the line graph to explain important capital flow nodes or events, the report nodes that need to be annotated and the annotation content can be determined; at this time, the Wenshengtu component can automatically identify the relevant data points and add corresponding annotations.

[0072] In addition, if the user has specific requirements for the color and style of the graphics, the Wenshengtu component can also be personalized to meet the user's aesthetic needs. Through this technology of intelligently producing graphics, users can more intuitively understand the key information in the treasury data and provide stronger support for fund management decisions. For example, the user inputs: Help me generate an illustration of the Mid-Autumn Festival. The intelligent report generation module will use Agent technology to call the Wenshengtu component to generate different graphs for user selection. For example, the user inputs: Help me analyze the flow data in the xxx table file and draw a pie chart. The intelligent report generation module will use Agent technology to call the data analysis component to first analyze the table file, then draw a data pie chart and return it to the user.

[0073] This embodiment uses LLM's multi-round and multi-path hybrid method to recall various static and dynamic search data, realize comprehensive and accurate search of treasury data, and meet the needs of users in different scenarios. At the same time, it innovatively adopts large model Agent technology, calls different tool components through dialogue, and collaboratively and intelligently generates report files, realizing intelligent and personalized report generation, and improving the quality and practicality of reports.

[0074] Another embodiment of the present invention provides a large model-based intelligent data generation system, which executes the above-mentioned large model-based intelligent data generation method. The overall architecture of the system and the collaborative working mechanism between modules ensure the efficient operation and stable performance of the system, and through reasonable architecture design and close cooperation between modules, rapid data processing and automatic generation of reports are achieved.

[0075] It should be noted that the functions of each module in the system of the embodiment of the present invention can be found in the corresponding description in the above method, and will not be repeated here.

[0076] Another embodiment of the present invention provides an electronic device, Figure 2 FIG. 2 shows a structural block diagram of an electronic device according to an embodiment of the present invention. Figure 2 As shown, the electronic device includes: a memory 100 and a processor 200, wherein the memory 100 stores a computer program that can be run on the processor 200. When the processor 200 executes the computer program, the intelligent data generation method based on the large model in the above embodiment is implemented. The number of the memory 100 and the processor 200 can be one or more.

[0077] The electronic device also includes:

[0078] The communication interface 300 is used to communicate with external devices and perform data exchange transmission.

[0079] If the memory 100, the processor 200 and the communication interface 300 are implemented independently, the memory 100, the processor 200 and the communication interface 300 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0080] Optionally, in a specific implementation, if the memory 100, the processor 200 and the communication interface 300 are integrated on a chip, the memory 100, the processor 200 and the communication interface 300 can communicate with each other through an internal interface.

[0081] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which implements the method provided in the embodiment of the present invention when executed by a processor.

[0082] An embodiment of the present invention further provides a chip, which includes a processor for calling and executing instructions stored in the memory from the memory, so that a communication device equipped with the chip executes the method provided by the embodiment of the present invention.

[0083] An embodiment of the present invention also provides a chip, including: an input interface, an output interface, a processor and a memory, wherein the input interface, the output interface, the processor and the memory are connected via an internal connection path, and the processor is used to execute the code in the memory. When the code is executed, the processor is used to execute the method provided by the embodiment of the invention.

[0084] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. It is worth noting that the processor may be a processor supporting the advanced RISC machines (ARM) architecture.

[0085] Further, optionally, the above-mentioned memory may include a read-only memory and a random access memory, and may also include a non-volatile random access memory. The memory 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 include a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may include a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct rambus RAM (DR RAM).

[0086] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it 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. When the computer program instructions are loaded and executed on a computer, the process or function according to the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium.

[0087] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0088] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0089] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of various changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for generating intelligent data based on a large model, characterized in that: include: Acquire dialogue information, process the dialogue information based on the LLM language model to determine key information, and determine the type of search intent based on the key information; Execute multiple retrieval tasks related to the retrieval intent type based on a multi-way recall algorithm, each of the retrieval tasks retrieves retrieval results associated with the key information from a preset database, and re-ranks the retrieval results to obtain target data; A tool component associated with the search intent type is called, the target data is processed by the tool component, and a data report related to the target data is generated.

2. The method for generating intelligent data based on a large model according to claim 1, characterized in that: The retrieval intentions include table file retrieval, graph file retrieval, real-time data retrieval, specific data indicator query intention, data range limitation retrieval intention and data comparison intention.

3. The method for generating intelligent data based on a large model according to claim 1, characterized in that: Also includes: Acquire treasury data in batches based on a data paging algorithm, wherein the treasury data includes financial data, business data, and transaction data; Determine the abnormal values ​​of the treasury data based on an abnormality detection algorithm, and perform data cleaning on the abnormal values ​​to obtain valid data; Identify characteristic information of the valid data, and mark and classify the valid data according to the characteristic information to obtain target treasury data; The target treasurer data is stored in a database to obtain the preset database.

4. The method for generating intelligent data based on a large model according to claim 1, characterized in that: Also includes: When the search intention type is real-time data search, connecting to a real-time data source; The latest data related to the key information is obtained from the real-time data source for display.

5. The method for generating intelligent data based on a large model according to claim 1, characterized in that: The multi-way recall algorithm includes a sparse BM25 recall algorithm, a dense vector recall algorithm, and a large model rewrite recall algorithm.

6. The method for generating intelligent data based on a large model according to claim 1, characterized in that: Also includes: Determine the annotation requirement according to the dialogue information, and determine the report node to be annotated and the annotation content according to the annotation requirement; The corresponding report nodes in the data report are automatically marked with the corresponding marked content to obtain an updated data report.

7. The method for generating intelligent data based on a large model according to claim 1, characterized in that: The tool component includes a Wensheng chart component, and the Wensheng chart component is used to run a specified graphic drawing algorithm to draw a graphic file related to the target data, and the graphic file includes a line chart, a bar chart, and a pie chart.

8. An intelligent data generation system based on a large model, characterized in that: Execute the large model-based intelligent data generation method as described in any one of claims 1 to 7.

9. An electronic device, characterized in that: include: A processor and a memory, wherein the memory stores instructions, and the instructions are loaded and executed by the processor to implement the large model-based intelligent data generation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for generating intelligent data based on a large model as described in any one of claims 1 to 7 is implemented.

Images