Generative BI method and device based on voice dialogue and storage medium

Abstract

The invention relates to a method and equipment for generating BI based on voice conversation and a storage medium, and the method comprises the steps: collecting and preprocessing voice, then carrying out voice recognition and transcription, and carrying out the intention understanding and semantic analysis of a transcribed text in combination with a natural language processing technology; automatically generating a corresponding data query statement or code according to a semantic analysis result; generating a visualization scheme according to the visualization requirements obtained by semantic analysis and the characteristics of the data analysis result; establishing connection with a plurality of data sources, integrating and preprocessing data acquired from different data sources, and setting data cache and management; and according to the generated visualization scheme, generating a visualization chart by using a visualization library, realizing an interaction function of the visualization chart, and realizing multi-device adaptation of visualization display. According to the method, a user can interact with the system through a natural language without learning complex query grammar and operation process, so that the convenience of data analysis and the user experience are greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device and storage medium for generating BI based on voice dialogue. Background Art

[0002] Traditional BI systems typically require users to possess specialized data analysis knowledge and skills during data processing and analysis, such as writing complex query statements (such as SQL) or mastering the operation of specific analytical tools. This high barrier to entry for ordinary business personnel limits the widespread application of BI systems in corporate decision-making and the effective mining of data value. Furthermore, traditional BI systems' interactive methods are mostly based on graphical interfaces, which are inefficient and cannot meet the rapidly changing needs of business analysis.

[0003] The rapid development of speech recognition, natural language processing, and generative AI technologies has provided new avenues for improving BI systems. Voice interaction allows users to communicate with the system more naturally and conveniently, while generative technologies can automatically generate data analysis logic and visualization solutions based on user needs, improving the efficiency and flexibility of data analysis. However, a comprehensive technical solution for the deep integration of voice interaction and generative BI is currently lacking. Summary of the Invention

[0004] The present invention proposes a method, device and storage medium for generating BI based on voice dialogue, which can solve at least one of the technical problems in the background technology.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for generating BI based on voice dialogue, comprising the following steps:

[0007] Speech collection and preprocessing, followed by speech recognition and transcription, combined with natural language processing technology to understand the intent and parse the semantics of the transcribed text;

[0008] Automatically generate corresponding data query statements or codes based on the results of semantic analysis;

[0009] Generate a visualization solution based on the visualization requirements obtained from semantic analysis and the characteristics of data analysis results;

[0010] Establish connections with multiple data sources, integrate and pre-process data obtained from different data sources, and set up data caching and management;

[0011] According to the generated visualization scheme, the visualization library is used to generate visualization charts, and the interactive functions of the visualization charts are realized, and the multi-device adaptation of the visualization display is realized.

[0012] Further, speech recognition and transcription include using a speech recognition engine to convert the preprocessed speech signal into text form based on deep neural network and hidden Markov model technologies;

[0013] When training the speech recognition model, a large amount of speech data covering business domain vocabulary and expressions is used to improve the recognition ability of business-related speech commands.

[0014] Further, intent understanding and semantic parsing include combining natural language processing technologies to perform intent understanding and semantic parsing on the transcribed text;

[0015] By constructing a domain ontology model and a semantic role annotation algorithm, the data analysis intent, data objects, qualification conditions, and expected visualization forms of the user are identified.

[0016] Further, when collecting speech in a noisy environment, adaptive filtering technology is used to track and eliminate environmental noise in real time; taking the least mean square (LMS) adaptive filtering algorithm as an example, the filter coefficients are adjusted according to the least mean square error criterion;

[0017] Let the input signal be x(n), the desired response be d(n), and the filter coefficient vector be w(n). Then the filter output y(n) = wT(n)x(n); the error signal e(n) = d(n) - y(n). According to the LMS algorithm, the filter coefficient update formula is w(n + 1) = w(n) + 2μe(n)x(n), where μ is the step size parameter that controls the convergence speed and stability; by continuously iterating and updating w(n), the filter can adaptively match the characteristics of environmental noise, effectively remove noise, and improve the quality of the speech signal.

[0018] Further, the double-threshold energy detection method is used in speech collection to determine the start and end points of speech segments; first, calculate the short-time energy of the speech signal E(n) = ∑i = n - N + 1nx2(i), where N is the analysis frame length; set a high threshold Th and a low threshold Tl. When E(n) > Th, it is determined that the speech segment starts; if E(n) < Tl and lasts for a certain period of time, it is determined that the speech segment ends.

[0019] Further, corresponding data query statements or codes are automatically generated. For relational databases, SQL query statements are generated, including selecting appropriate tables and columns, constructing query conditions, and join operations;

[0020] For non-relational databases or other data storage forms, corresponding query API calls or data access logics are generated; when generating query statements, the index structure and storage characteristics of the data are considered to improve query efficiency.

[0021] Furthermore, connections are established with multiple data sources, including relational databases, non-relational databases, data files in cloud storage, and internal enterprise file systems, and unified access to data is achieved by configuring data source connection parameters and drivers.

[0022] Furthermore, user management and permission control: establish a user management system to register, log in and manage permissions for users; assign different system access rights based on user roles to ensure data security and compliance with the system;

[0023] Model update and optimization: Collect user voice commands and feedback on analysis results, and use this data to update and optimize speech recognition models, semantic parsing models, data analysis algorithms, and visualization solutions;

[0024] System monitoring and performance optimization: Real-time monitoring of the system's operating status, including server resource usage, data query and processing time, and concurrent user access; based on the monitoring results, the system's performance is optimized to ensure efficient and stable operation of the system.

[0025] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the above method.

[0026] On the other hand, the present invention further discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0027] As can be seen from the above technical solutions, the method of generating BI based on voice dialogue of the present invention, as an emerging data analysis method, has many significant features and advantages, as follows:

[0028] Its unique feature is the combination of generative BI and voice dialogue, which breaks the limitations of traditional data analysis tools on users' technical knowledge and operational experience. Users can interact with the system through natural language without having to learn complex query syntax and operational processes, greatly improving the convenience of data analysis and user experience.

[0029] The first advantage is personalized data analysis. By understanding user preferences and historical query records, the system can provide users with tailored data analysis and insights, helping them quickly find key information relevant to their business or interests and improving decision-making efficiency. Secondly, it breaks down data silos. Modern enterprise data is scattered across multiple systems and platforms, making it difficult to integrate with traditional tools. However, generative BI based on voice conversations can integrate data from different sources, including structured and unstructured data, providing enterprises with a comprehensive data analysis perspective.

[0030] In terms of application results, the use of big models in the financial sector at the data processing and analysis levels has improved data processing efficiency and analysis accuracy, shortening project delivery cycles. In the political and legal fields, iFLYTEK's legal big models and related applications have enhanced the informatization of trial processes, achieving intelligent, comprehensive coverage of trial scenarios, and providing an effective solution for collaborative work between the political and legal sectors.

[0031] In short, conversational-based generative BI demonstrates tremendous potential and advantages in the field of data analysis, bringing significant application results to businesses and society. It not only improves the efficiency and accuracy of data analysis but also provides users with a more convenient and efficient way to interact with data, helping businesses better utilize data, make more accurate decisions, and enhance their competitiveness. In the future, with the continuous development and improvement of artificial intelligence technology, conversational-based generative BI is expected to be widely applied in even more fields, playing a greater role in promoting the development and progress of various industries. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0034] The present invention aims to provide a generative BI system and method based on voice dialogue, which enables users to quickly and accurately obtain data analysis results through voice commands without the need for complex operations and professional knowledge. It also improves the efficiency and quality of data analysis and promotes scientific and timely enterprise decision-making. The specific details are as follows:

[0035] 1. Voice interaction module

[0036] Voice capture and preprocessing: A high-quality microphone or other voice input device is used to capture user voice commands. The captured voice signals are then preprocessed, such as through noise reduction and enhancement, to improve speech recognition accuracy. For example, adaptive filtering techniques are used to remove ambient noise, and voice activity detection algorithms are used to determine the start and end points of speech segments.

[0037] Speech Recognition and Transcription: Utilizing an advanced speech recognition engine, based on technologies such as Deep Neural Networks (DNNs) and Hidden Markov Models (HMMs), we convert pre-processed speech signals into text. The speech recognition model is trained using a large amount of speech data covering business vocabulary and expressions to improve its ability to recognize business-related voice commands.

[0038] Intent understanding and semantic parsing: Combined with natural language processing technology, the transcribed text is subjected to intent understanding and semantic parsing. By building a domain ontology model and a semantic role labeling algorithm, the user's data analysis intention (such as query, statistics, prediction, etc.), data objects (such as customers, products, orders, etc.), restrictions (such as time range, region, quantity threshold, etc.), and desired visualization form (such as bar charts, line charts, pie charts, etc.) are identified. For example, for the voice command "Query the sales trend of products with sales exceeding 1 million last year, and display it with a line chart", it can accurately parse that the intention is query and visualization trend analysis, the data object is product, the restriction is that sales exceeded 1 million last year, and the visualization form is a line chart.

[0039] Adaptive filtering technology: When collecting speech in noisy environments, adaptive filtering technology can track and eliminate ambient noise in real time. Taking the most commonly used least mean square (LMS) adaptive filtering algorithm as an example, its core principle is to adjust the filter coefficients based on the minimum mean square error criterion. Let the input signal be x(n), the desired response be d(n), and the filter coefficient vector be w(n). Then the filter output is y(n) = wT(n)x(n). The error signal e(n) = d(n) - y(n). According to the LMS algorithm, the filter coefficient update formula is w(n+1) = w(n) + 2μe(n)x(n), where μ is the step size parameter, controlling convergence speed and stability. By continuously iteratively updating w(n), the filter can adaptively match the characteristics of the ambient noise, effectively removing noise and improving speech signal quality.

[0040] Voice Activity Detection Algorithm: The double-threshold energy detection method is used to determine the start and end points of speech segments. First, calculate the short-time energy of the speech signal E(n) = ∑i=n-N+1nx2(i), where N is the analysis frame length. Set the high threshold Th and the low threshold Tl. When E(n) > Th, it is determined that the speech segment starts; if E(n) < Tl and lasts for a certain period of time, it is determined that the speech segment ends. This algorithm can effectively distinguish speech and non-speech signals, reduce the processing of invalid data, and improve the efficiency of speech recognition.

[0041] 2. Generative Data Analysis Module

[0042] Data Query Generation: According to the results of semantic parsing, automatically generate corresponding data query statements or codes. For relational databases, generate SQL query statements, including selecting appropriate tables, columns, constructing query conditions, and join operations. For non-relational databases or other data storage forms (such as NoSQL databases, data files in file systems, etc.), generate corresponding query API calls or data access logics. When generating query statements, consider the index structure and storage characteristics of the data to improve query efficiency. For example, according to the user's instruction to query customer information in a specific area, the generated SQL query statement will involve the join operation of the customer table and the area table, and add area filtering conditions.

[0043] Data Analysis Algorithm Selection and Generation: Based on the user's data analysis intention and data characteristics, select a suitable algorithm from a predefined data analysis algorithm library and generate the corresponding algorithm implementation code. The algorithm library covers various types of algorithms such as statistical analysis (e.g., mean and variance calculation), data mining (e.g., clustering and classification algorithms), and predictive analysis (e.g., time series prediction and regression analysis). If the user's instruction is to classify customers, the system may select the K-Means clustering algorithm and generate code including steps such as data normalization, clustering calculation, and result output.

[0044] Visualization Scheme Generation: Generate a visualization scheme according to the visualization requirements obtained from semantic parsing and the characteristics of the data analysis results. Determine appropriate chart types, layouts, color schemes, data mapping relationships, etc. For example, if comparing the sizes of different category data, select a bar chart and determine the width, height, and color of the bars according to the number of data categories and the data value range. At the same time, consider the interactivity of the visualization, such as adding data tooltips, zooming, filtering, etc., to facilitate users to further explore the data.

[0045] 3. Data Processing and Storage Module

[0046] Multi-data source connections: Establish connections to a variety of data sources, including but not limited to relational databases (such as MySQL, Oracle, and SQL Server), non-relational databases (such as MongoDB, Redis, and Cassandra), data files in cloud storage (such as CSV and JSON files in AWS S3 and Google Cloud Storage), and internal enterprise file systems (such as Excel and XML files). By configuring data source connection parameters and drivers, you can achieve unified access to data.

[0047] Data integration and preprocessing: Data collected from different data sources is integrated and preprocessed. This includes data cleaning (such as removing duplicate data, handling missing values, and correcting erroneous data), data conversion (such as data type conversion, date format unification, and data encoding conversion), and data fusion (such as linking related data from different tables or files). For example, basic customer information and purchase history data collected from different databases can be linked to form a complete customer profile.

[0048] Data Caching and Management: To improve system performance, establish a data caching mechanism. Cache frequently accessed data and intermediate analysis results, and set reasonable cache update policies, such as those based on time and data update frequency. Also, optimize data storage management, including data partitioning and index optimization, to increase data query and processing speed.

[0049] 4. Visual display module:

[0050] Chart Generation and Rendering: Based on the generated visualization solution, use visualization libraries (such as D3.js, ECharts, and Highcharts) to generate visualization charts. Accurately map data to various chart elements, such as axes, data points, and graph areas, and then render and display the charts. For example, map customer sales data to the height and color of a bar chart, and map time data to the horizontal axis.

[0051] Interactive functionality: Implement interactive features for visualizations, such as hovering to display detailed data information, clicking to drill down, zooming to view data in different ranges, and filtering data under specific conditions. Implementing interactive logic through front-end technologies like JavaScript allows users to deeply explore data and discover hidden information and trends.

[0052] Multi-device adaptation: Considering that users may access the system from different devices (such as desktop computers, tablets, and smartphones), the visualization display is adapted to multiple devices. The chart size, layout, and interaction method are automatically adjusted according to the device screen size, resolution, and interaction method to ensure a good user experience.

[0053] 5. System management and optimization module

[0054] User Management and Permission Control: Establish a user management system to manage user registration, login, and permissions. Assign different system access permissions based on user roles (e.g., administrator, analyst, general business personnel, etc.), such as data query scope, analysis function usage rights, and data modification rights, to ensure data security and system compliance.

[0055] Model Update and Optimization: We collect user voice commands and feedback on analysis results, and use this data to update and optimize the speech recognition model, semantic parsing model, data analysis algorithm, and visualization solution. For example, we can improve speech recognition accuracy by adding new training data, and adjust the default parameters for chart generation based on user feedback on visualization effects.

[0056] System Monitoring and Performance Optimization: Real-time monitoring of the system's operating status, including server resource usage (such as CPU, memory, disk I / O, network bandwidth, etc.), data query and processing time, and concurrent user access. Based on the monitoring results, system performance optimization is performed, such as adjusting server configuration, optimizing query algorithms, and increasing cache capacity, to ensure efficient and stable system operation.

[0057] The following examples illustrate:

[0058] At a certain city's urban operations management command center, case handlers are responsible for handling various complaint cases to improve urban governance and residents' quality of life. Complaint data comes from a wide range of sources and is stored in various systems, including relational databases (such as MySQL, Oracle, and DM) for formally filed complaint records and documents in cloud storage (such as detailed complaint descriptions sent by citizens via email).

[0059] The data analyst opened the generative BI system based on voice dialogue and said into the microphone: "Analyze the handling of noise pollution complaints over the past month. Compare the number of complaints and the average processing time by region. Use a bar chart to compare the number of complaints by region, and use a line chart to show the trend of the average processing time for each region."

[0060] After the system's high-quality microphones capture voice commands, it uses adaptive filtering technology to remove ambient office noise and employs a double-threshold energy detection method to accurately determine the start and end points of the voice segments. The speech recognition engine utilizes deep neural networks and hidden Markov model technology to convert the speech into text. Next, using a constructed domain ontology model and semantic role labeling algorithm, the system accurately interprets the data analyst's intent: to analyze the handling of noise pollution complaints. The data object is noise pollution complaint handling, with the constraint of the past month. The desired visualization format is a bar chart showing the comparison of regional complaint numbers and a line chart showing the changing trends in regional average handling times.

[0061] Based on the semantic analysis results, the system generates complex SQL queries for relational databases. It selects records from the table storing complaint handling information that have complaints filed within the past month regarding noise pollution, and then associates fields such as processing time and region. For relevant data in non-relational databases and cloud storage, the system generates corresponding query API calls or data access logic to retrieve the complete complaint data. When generating queries, the database's index structure is fully considered to improve query efficiency.

[0062] After acquiring the data, the system selects an appropriate statistical algorithm from a predefined data analysis algorithm library based on the data analysis intent. The system calculates the number of complaints and average processing time for each region. Based on the regional complaint number comparison, a bar chart is used to display the data, mapping the region to the horizontal axis and the number of complaints to the vertical axis. Different colors are used to distinguish between regions. For the average processing time trend, a line chart is selected, with time on the horizontal axis (dividing the past month into weeks) and the average processing time on the vertical axis. Data prompt interaction is added to the bar chart and line chart to facilitate data analysts to view specific data.

[0063] The system establishes connections with various data sources, configures corresponding connection parameters and drivers, and enables unified access to various complaint data. Data obtained from different data sources is integrated and pre-processed, such as data cleansing to remove duplicate complaint records and invalid data; data conversion to standardize the format of processing time to facilitate calculation of average duration; and data fusion to merge relevant information on the same complaint from different systems.

[0064] The system uses the ECharts visualization library to generate bar charts and line charts based on the generated visualization scheme. The processed data is accurately mapped onto chart elements. In the bar chart, the height of the bars accurately reflects the number of complaints in each region; in the line chart, the trend of the line clearly shows the changes in the average processing time for each region.

[0065] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the above method.

[0066] On the other hand, the present invention further discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0067] In another embodiment provided by the present application, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute any of the methods for generating BI based on voice dialogue in the above embodiments.

[0068] It is understandable that the system, device and storage medium provided in the embodiments of the present invention correspond to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant contents can refer to the corresponding parts of the above methods.

[0069] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When software is used for implementation, 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 described in the embodiment of the present application 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. For example, the computer instructions can be transmitted from a website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0070] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0071] Each embodiment in this specification is described in a related manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiment is generally similar to the method embodiment, so the description is relatively simple. For related parts, refer to the description of the method embodiment.

[0072] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for generating BI based on voice dialogue, characterized in that: including the following steps Speech acquisition and preprocessing, followed by speech recognition and transcription, and combining natural language processing techniques to perform intent understanding and semantic parsing on the transcribed text; According to the results of semantic parsing, automatically generate corresponding data query statements or code; Generate a visualization plan based on the visualization requirements obtained from semantic parsing and the characteristics of data analysis results; Establish connections with multiple data sources, integrate and preprocess the data obtained from different data sources, and set up data caching and management; According to the generated visualization plan, use a visualization library to generate visualization charts, implement the interactive functions of the visualization charts, and achieve multi-device adaptation for visualization display.

2. The method for generating BI based on voice dialogue according to claim 1, characterized in that: Speech recognition and transcription include using a speech recognition engine to convert the preprocessed speech signal into text form based on deep neural network and hidden Markov model technologies; When training the speech recognition model, use a large amount of speech data covering business domain vocabulary and expressions to improve the recognition ability for business-related speech commands.

3. The method for generating BI based on voice dialogue according to claim 2, characterized in that: Intent understanding and semantic parsing include combining natural language processing techniques to perform intent understanding and semantic parsing on the transcribed text; By constructing a domain ontology model and a semantic role annotation algorithm, identify the user's data analysis intent, data objects, limiting conditions, and expected visualization forms.

4. The method for generating BI based on voice dialogue according to claim 3, characterized in that: When collecting speech in a noisy environment, use adaptive filtering technology to track and eliminate environmental noise in real time; taking the least mean square (LMS) adaptive filtering algorithm as an example, the filter coefficients are adjusted according to the least mean square error criterion; Let the input signal be x(n), the desired response be d(n), and the filter coefficient vector be w(n), then the filter output y(n) = wT(n)x(n); the error signal e(n) = d(n) - y(n), and according to the LMS algorithm, the filter coefficient update formula is w(n + 1) = w(n) + 2μe(n)x(n), where μ is the step size parameter that controls the convergence speed and stability; by continuously iteratively updating w(n), the filter can adaptively match the characteristics of environmental noise, effectively remove noise, and improve the quality of the speech signal.

5. The method for generating BI based on voice dialogue according to claim 4, characterized in that: The double-threshold energy detection method is used in speech acquisition to determine the start and end points of speech segments; first, calculate the short-time energy of the speech signal E(n) = ∑i=n-N+1nx2(i), where N is the analysis frame length; set a high threshold Th and a low threshold Tl, when E(n) > Th, it is determined that the speech segment starts; if E(n) < Tl and lasts for a certain period of time, it is determined that the speech segment ends.

6. The method for generating BI based on voice dialogue according to claim 5, characterized in that: Automatically generate corresponding data query statements or code. For relational databases, generate SQL query statements, including selecting appropriate tables and columns, constructing query conditions, and join operations; For non-relational databases or other data storage forms, generate corresponding query API calls or data access logics; when generating query statements, consider the index structure and storage characteristics of the data to improve query efficiency.

7. The method for generating BI based on voice dialogue according to claim 6, characterized in that: Establish connections with multiple data sources, including relational databases, non-relational databases, data files in cloud storage, and internal enterprise file systems, and achieve unified access to data by configuring data source connection parameters and drivers.

8. The method for generating BI based on voice dialogue according to claim 7, characterized in that: Also includes: User management and authority control: Establish a user management system to register, log in and manage permissions for users; Assign different system access rights based on user roles to ensure data security and compliance with the system; Model update and optimization: Collect user voice commands and feedback on analysis results, and use this data to update and optimize speech recognition models, semantic parsing models, data analysis algorithms, and visualization solutions; System monitoring and performance optimization: Real-time monitoring of the system's operating status, including server resource usage, data query and processing time, and concurrent user access; Based on the monitoring results, the system performance is optimized to ensure efficient and stable operation of the system.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 8.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the computer program is executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 8.

Images