Visual report generation method and device based on AI dialogue mode

Abstract

The invention relates to the technical field of big data and artificial intelligence, and particularly provides a visual report generation method and device based on an AI dialogue mode, and the method comprises the following steps: S1, data processing; s2, performing dialogue type intelligent query; s3, generating and displaying a graph; and S4, report generation and display. Compared with the prior art, the method has the advantages that report generation efficiency can be improved, dialogue type interaction between a user and a system is achieved through a large model technology, and the user can quickly generate a visual report meeting requirements without compiling complex query statements or codes.

Description

Technical Field

[0001] The present invention relates to the technical field of big data and artificial intelligence, and specifically provides a method and device for generating a visualized report based on AI dialogue. Background Art

[0002] Data visualization reports are tools that express data through visual methods such as graphics and charts. They can convert abstract data into specific graphics and charts, and intuitively display information such as data distribution, trends, and associations. They help people quickly understand the content of the data, and understand the relationships and trends between the data, thereby supporting more informed decision-making and analysis, helping to more effectively convey information and communicate ideas to others, and reducing misunderstandings and communication barriers.

[0003] With the continuous advancement of AI and ML technologies, data visualization reports will be able to process and analyze data more intelligently. For example, AI algorithms can automatically select the most suitable visualization chart type, or ML models can be used to predict data trends and generate predictive reports.

[0004] With the rapid development of big data and artificial intelligence technology, data visualization has become an important tool for corporate decision-making and data analysis. Traditional report generation methods often require users to have professional programming or data analysis knowledge, which is cumbersome and inefficient. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a highly practical method for generating visual reports based on AI dialogue.

[0006] A further technical task of the present invention is to provide a reasonably designed, safe and applicable AI-based conversational visualization report generation device.

[0007] The technical solution adopted by the present invention to solve its technical problem is:

[0008] A method for generating a visual report based on AI conversational style, comprising the following steps:

[0009] S1. Data processing;

[0010] S2, conversational intelligent query;

[0011] S3, graphic generation and display;

[0012] S4. Report generation and display.

[0013] Further, in step S1, a data processing module, a database management module, a semantic modeling module and a prompt word management module are included;

[0014] The data processing module performs data access and preprocessing operations;

[0015] The database management module performs data connection import;

[0016] The semantic modeling module models and constructs business data, defines semantic information, and incorporates data semantics into prompt words;

[0017] The prompt word management module sets prompt words for the model after the semantic modeling is completed.

[0018] Furthermore, the prompt words are divided into two types, single model prompt word configuration and global prompt word unified configuration;

[0019] The knowledge base management module periodically extracts relevant pattern information from the semantic model to build a dictionary and index for subsequent pattern mapping. The pattern mapper matches the natural language text in the knowledge base to provide relevant information for semantic analysis;

[0020] The semantic parser understands user queries and extracts semantic information to generate semantic query statements, which are then translated into SQL statements that can be executed on the physical data model by the semantic translator.

[0021] Further, in step S2, a business semantics module, a question-answering dialogue module, and a rule semantics parsing module are included;

[0022] The business semantic module uses rule semantic parsing or big model semantic parsing technology to parse user queries. When a user enters the question-and-answer conversational module, he needs to create a new question-and-answer topic, associate semantic models with the rule semantic parsing and big model semantic parsing of the topic, and manually add example questions commonly used in the topic.

[0023] Furthermore, in the Q&A conversational interface, select a Q&A topic, enter a question to ask, or quickly create a question in the example questions or recommended questions to ask. After the background receives the user's query question, it will first perform intent analysis based on the rule semantic analysis module and the configured semantic modeling information, and finally return the intent result. If a computational analysis of indicator splitting is performed, the background will call the large model semantic analysis to perform intent analysis and present the analysis results.

[0024] Furthermore, in step S3, the corresponding visualization charts are intelligently generated based on the results of the conversational intelligent analysis. The charts displayed for each question and answer will also provide automatically generated data summaries, helping users quickly obtain a comprehensive and accurate data overview through language-descriptive statistical analysis;

[0025] For charts with time series indicators, you can view the year-on-year information directly in the generated chart. The backend will automatically calculate the year-on-year value, growth value, and growth rate.

[0026] For indicators that define time series in the model, users can specify the year-on-year and year-on-year results they want to view in the Q&A and modify the display of the year-on-year analysis in the query conditions of the Q&A results, allowing for flexible definition.

[0027] Further, in step S4, report generation includes two methods. One is to copy a single chart generated by the above question-and-answer method to an existing report or a new report to form a report. The report generated in this way supports multiple semantic models, and interactive analysis of multi-model charts is performed in the report.

[0028] The other is to generate reports directly through question-and-answer format, select a preset report template, and generate multiple related charts based on the questions through background analysis and big model technology to form a report.

[0029] Furthermore, the report can be directly generated in the question-and-answer mode. Users can pre-build multiple models and select report templates according to their needs, input natural language to ask questions in the question-and-answer module, and generate results by parsing the natural language based on the configured multiple models and their indicators;

[0030] The charts generated by the model with time series indicators will automatically include year-on-year analysis and attribution analysis. Users can enter the generated report and perform correlation analysis and drill-down analysis on the charts.

[0031] If one semantic model is configured, the generated results include one report. If multiple models are configured, multiple reports will be generated based on the problem analysis for the user to select and analyze.

[0032] If the generated report does not meet the requirement, multiple rounds of inquiries are conducted and in-depth report generation is performed to meet the requirement for generated report.

[0033] A visual report generation device based on AI conversational style, comprising: at least one memory and at least one processor;

[0034] The at least one memory is used to store a machine-readable program;

[0035] The at least one processor is used to call the machine-readable program to execute a method for generating a visual report based on AI conversational style.

[0036] Compared with the prior art, the AI-based conversational visualization report generation method and device of the present invention has the following outstanding beneficial effects:

[0037] The present invention improves the efficiency of report generation: by implementing a dialog-based interaction between the user and the system through the big model technology, the user can quickly generate a visual report that meets the needs without having to write complex query statements or codes.

[0038] Reduced operational difficulty: The report generation process is simplified and the operational difficulty is reduced, so that users can generate reports without having professional programming or data analysis knowledge.

[0039] Enhanced the readability and aesthetics of reports: Supports multiple chart types and custom settings. Users can beautify reports as needed to improve the readability and aesthetics of reports.

[0040] Improved report flexibility: Supports access and integration of multiple data sources. Users can select data sources to generate reports as needed, which improves report flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Attached Figure 1 It is a flowchart of a visual report generation method based on AI conversation. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the solution of the present invention, the present invention is further described in detail below in conjunction with specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] A best embodiment is given below:

[0045] like Figure 1 As shown, a method for generating a visual report based on AI conversation in this embodiment has the following steps:

[0046] S1. Data processing;

[0047] It includes data processing module, database management module, semantic modeling module and prompt word management module;

[0048] The data processing module is responsible for data access and preprocessing operations. The database management module connects and imports data; the semantic modeling module models and builds business data, defines semantic information, and incorporates data semantics, such as business terms and column values, into prompt words, so that LLM can better understand semantics to reduce illusions, improve the accuracy of the model, and offload the generation of advanced SQL syntax from LLM to the semantic layer to reduce complexity.

[0049] The prompt word management module is mainly used to set prompt words for the model after semantic modeling is completed. It supports two prompt word settings: single model prompt word configuration and global prompt word unified configuration, eliminating the tedious steps of repeating the same prompt word configuration for multiple models. Users can set prompt words according to analysis needs to improve the accuracy of subsequent semantic analysis results.

[0050] Knowledge base management mainly involves regularly extracting relevant pattern information from the semantic model to build dictionaries and indexes for subsequent pattern mapping; the pattern mapper mainly matches natural language text in the knowledge base to provide relevant information for semantic analysis; the semantic parser is mainly used to understand user queries and extract semantic information to generate semantic query statements, which are then translated into SQL statements that can be executed on the physical data model through the semantic translator.

[0051] The model supports the setting of row and column-level permissions to limit user access control to dimensions and indicators.

[0052] S2, conversational intelligent query;

[0053] Includes business semantics module, question-answering dialogue module and rule semantics parsing module;

[0054] Conversational intelligent query is a process where users interact with the system through natural language and raise query requirements. The system uses rule semantic parsing or large model language multi-parsing technology to parse user queries based on the corresponding business semantic model.

[0055] When users enter the Q&A conversational module, they need to create a new Q&A topic, associate a semantic model with the topic's rule semantic analysis and large model semantic analysis, and manually add sample questions commonly used in the topic. In the Q&A conversational interface, select the Q&A topic and enter a question to ask. You can also quickly create a question from sample questions or recommended questions to ask. After the background receives the user's query, it will first perform intent analysis based on the rule semantic analysis module and the configured semantic modeling information, and finally return the intent result. If a computational analysis of indicator splitting is performed, such as complex semantic analysis such as month-on-month and difference, the background will call the large model semantic analysis for intent analysis and present the analysis results.

[0056] The analysis results mainly include query conditions, query data, analysis conclusions, drillable dimensions, recommended related report information, etc., to facilitate users' subsequent report analysis.

[0057] In order to conduct in-depth analysis and support users to conduct multiple rounds of conversations, the system will automatically analyze and answer questions in context; it provides knowledge base management, and users can mark correct or incorrect answers based on the analysis results. Correctly marked relevant information will be incorporated into the knowledge base to enrich and enhance the knowledge base data. If the same question is asked again and the question has been marked as correct, the original answer will be automatically returned, making the model smarter with use.

[0058] S3, graphic generation and display;

[0059] The system intelligently generates corresponding visualization charts based on the results of conversational intelligent analysis. It supports a variety of chart types, such as bar charts, pie charts, line charts, tables, etc. For example, trend result data will automatically generate a line chart, and indicator data will automatically generate an indicator chart, and users can switch charts according to display needs. Each chart displayed in the question and answer will also provide an automatically generated data summary, which helps users quickly obtain a comprehensive and accurate data overview through language-descriptive statistical analysis.

[0060] The Q&A chart supports common calculation methods such as year-on-year, cumulative calculation, and attribution analysis to meet users' complex data analysis needs. For charts with time series indicators, you can directly view the year-on-year information in the generated chart, and the backend will automatically calculate the year-on-year value, growth value, growth rate, etc.

[0061] For indicators that define time series in the model, users can specify the year-on-year and year-on-year data they want to view in the Q&A. For example, this year's monthly supermarket sales and year-on-year sales will automatically calculate the year-on-year value, growth value and growth rate. At the same time, the display of the year-on-year analysis can be modified in the query conditions of the Q&A results, allowing for flexible definition.

[0062] After a single graphic is generated, you can copy and paste it into a specified report. A recommended report will be given in the generated results. The report contains model charts that are the same or similar to the model used to generate the graphic. Users can copy the graphic as needed, click on a recommended report, and paste it into the report.

[0063] S4, report generation and display;

[0064] There are two main ways to generate reports: one is to copy a single chart generated by the above-mentioned question-and-answer method to an existing report or a new report to form a report. The report generated in this way supports multiple semantic models, and multi-model chart interactive analysis can be performed in the report; the other is to directly generate a report through the question-and-answer method, select a preset report template, and the system will generate relevant multiple charts based on the problem through background analysis and large model technology to form a report.

[0065] Reports are generated directly in a question-and-answer style. Users can pre-build multiple models and select report templates as needed. They can enter natural language to ask questions in the question-and-answer module. The system generates results including one or more reports, descriptive conclusions, recommended questions, etc. by parsing natural language based on the configured multiple models and their indicators. The report consists of multiple chart components. Each model can flexibly set the number of generated charts, and two are generated by default. The charts generated by the model with time series indicators will automatically have calculation analysis such as year-on-year analysis and attribution analysis. Users can enter the generated report and perform association analysis, drill-down analysis, etc. on the chart; if a semantic model is configured, the generated results include a report. If multiple models are configured, multiple reports may be generated based on the question analysis for users to select and analyze; if the generated report does not meet the needs, multiple rounds of inquiries can be performed to generate in-depth reports to meet the needs of generating reports.

[0066] On the report editing page, users can fine-tune the settings of each chart in the report as needed, including data configuration, style configuration, calculation and analysis configuration, etc.; users can also customize and beautify the generated reports, export them, such as adjusting colors, fonts, sizes, etc., to improve the readability and aesthetics of the reports; the analyzed reports can be shared and exported; the report generates data summaries, analyzes the report as a whole, and gives the latest statistical conclusions of the analysis to quickly provide leaders with descriptive statistical analysis.

[0067] Based on the above method, a visual report generation device based on AI conversational mode in this embodiment includes: at least one memory and at least one processor;

[0068] The at least one memory is used to store a machine-readable program;

[0069] The at least one processor is used to call the machine-readable program to execute a method for generating a visual report based on AI conversational style.

[0070] The above-mentioned specific implementations are only specific cases of the present invention. The patent protection scope of the present invention includes but is not limited to the above-mentioned specific implementations. Any technical solutions that conform to the above-mentioned specific implementations of the present invention and any appropriate changes or substitutions made by ordinary technicians in the relevant technical field shall fall within the patent protection scope of the present invention.

[0071] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for generating a visual report based on AI conversation, characterized in that: The steps are as follows: S1. Data processing; S2, conversational intelligent query; S3, graphic generation and display; S4. Report generation and display.

2. According to claim 1, a method for generating a visual report based on AI dialogue is characterized in that: In step S1, a data processing module, a database management module, a semantic modeling module and a prompt word management module are included; The data processing module performs data access and preprocessing operations; The database management module performs data connection import; The semantic modeling module models and constructs business data, defines semantic information, and incorporates data semantics into prompt words; The prompt word management module sets prompt words for the model after the semantic modeling is completed.

3. The method for generating a visual report based on AI dialogue according to claim 2, characterized in that: The prompt words are divided into two types: single model prompt word configuration and global prompt word unified configuration; The knowledge base management module periodically extracts relevant pattern information from the semantic model to build a dictionary and index for subsequent pattern mapping. The pattern mapper matches the natural language text in the knowledge base to provide relevant information for semantic analysis; The semantic parser understands user queries and extracts semantic information to generate semantic query statements, which are then translated into SQL statements that can be executed on the physical data model by the semantic translator.

4. The method for generating a visual report based on AI dialogue according to claim 3, characterized in that: In step S2, a business semantics module, a question-answering dialogue module, and a rule semantics parsing module are included; The business semantic module uses rule semantic parsing or big model semantic parsing technology to parse user queries. When a user enters the question-and-answer conversational module, he needs to create a new question-and-answer topic, associate semantic models with the rule semantic parsing and big model semantic parsing of the topic, and manually add example questions commonly used in the topic.

5. The method for generating a visual report based on AI dialogue according to claim 4, characterized in that: In the Q&A conversational interface, select a Q&A topic and enter a question to ask. You can also quickly create a question from the sample questions or recommended questions to ask. After the background receives the user's query question, it will first perform intent analysis based on the rule semantic analysis module and the configured semantic modeling information, and finally return the intent result. If a computational analysis of indicator splitting is performed, the background will call the large model semantic analysis to perform intent analysis and present the analysis results.

6. The method for generating a visual report based on AI dialogue according to claim 5, characterized in that: In step S3, the corresponding visualization charts are intelligently generated based on the results of the conversational intelligent analysis. The charts displayed for each question and answer will also provide automatically generated data summaries, helping users quickly obtain a comprehensive and accurate data overview through language-descriptive statistical analysis; For charts with time series indicators, you can view the year-on-year information directly in the generated chart. The backend will automatically calculate the year-on-year value, growth value, and growth rate. For indicators that define time series in the model, users can specify the year-on-year and year-on-year results they want to view in the Q&A and modify the display of the year-on-year analysis in the query conditions of the Q&A results, allowing for flexible definition.

7. The method for generating a visual report based on AI dialogue according to claim 6, characterized in that: In step S4, report generation includes two methods. One is to copy a single chart generated by the question-and-answer method to an existing report or a new report to form a report. The report generated in this way supports multiple semantic models, and interactive analysis of multi-model charts is performed in the report. The other is to generate reports directly through question-and-answer format, select a preset report template, and generate multiple related charts based on the questions through background analysis and big model technology to form a report.

8. The method for generating a visual report based on AI dialogue according to claim 7, characterized in that: Reports are generated directly in a question-and-answer format. Users can pre-build multiple models and select report templates as needed. They can enter natural language to ask questions in the question-and-answer module. Results are generated by parsing natural language based on multiple configured models and their indicators. The charts generated by the model with time series indicators will automatically include year-on-year analysis and attribution analysis. Users can enter the generated report and perform correlation analysis and drill-down analysis on the charts. If one semantic model is configured, the generated results include one report. If multiple models are configured, multiple reports will be generated based on the problem analysis for the user to select and analyze. If the generated report does not meet the requirement, multiple rounds of inquiries are conducted and in-depth report generation is performed to meet the requirement for generated report.

9. A visual report generation device based on AI dialogue, characterized in that: include: at least one memory and at least one processor; The at least one memory is used to store a machine-readable program; The at least one processor is configured to call the machine-readable program to execute the method according to any one of claims 1 to 8.

Images