CN119271723A

CN119271723A - Data large-screen visualization method, device, equipment and medium based on large model

Info

Publication number: CN119271723A
Application number: CN202411425346.6A
Authority: CN
Inventors: 陆志鹏; 韩光; 李嘉宁; 郑曦; 郭祎萍; 国丽; 刘彬彬; 马博原; 张涛; 冯江伟
Original assignee: China Electronics Data Industry Group Co ltd
Current assignee: China Electronics Data Industry Group Co ltd
Priority date: 2024-10-12
Filing date: 2024-10-12
Publication date: 2025-01-07

Abstract

The present application discloses a method, device, equipment and medium for data large screen visualization based on a large model, and relates to the field of data large screen technology, including: obtaining externally input user voice data, performing intent analysis on the user voice data, and determining a query prompt word representing the intent from the user voice data; inputting the query prompt word into a preset large model for analysis, generating target data corresponding to the query prompt word and a target chart template matching the query prompt word; mapping the target data to the target chart template, determining the target chart, and visualizing the target chart. The present application can improve the efficiency of large screen data display.

Description

Large-model-based data large-screen visualization method, device, equipment and medium

Technical Field

The application relates to the technical field of large data screens, in particular to a large data screen visualization method, device, equipment and medium based on a large model.

Background

With the advent of the big data age, data visualization has become one of the important means for enterprises to make efficient decisions.

The data large screen is used as a way for intuitively displaying a large amount of data, and is widely applied to the fields of real-time monitoring, business analysis and the like. However, the current data large screen technology has a certain limitation, most of the current data large screens only can display preset chart types, such as a histogram, a line graph and the like, and the selection of the charts is always determined in a development stage, so that the display content of the data large screen is relatively fixed, and the displayed content and format are difficult to quickly adjust according to actual requirements. Once the data changes or new business needs occur, the code needs to be reconfigured and even rewritten to realize the update, which not only increases the workload, but also reduces the flexibility.

Therefore, how to solve the problem that the service requirement change causes that the large screen needs to be reconfigured to be visualized is a problem that needs to be solved at present.

The foregoing is provided merely for the purpose of facilitating understanding of the technical solutions of the present application and is not intended to represent an admission that the foregoing is prior art.

Disclosure of Invention

The application mainly aims to provide a large-screen data visualization method, device and equipment based on a large model and a storage medium, and aims to solve the technical problem that a large screen needs to be reconfigured for visualization due to service demand change.

In order to achieve the above object, the present application provides a large-model-based data large-screen visualization method, which includes:

Acquiring externally input user voice data, performing intent analysis on the user voice data, and determining query prompt words representing the intent from the user voice data;

Inputting the query prompting words into a preset large model for analysis, and generating target data corresponding to the query prompting words and target chart templates matched with the query prompting words;

and mapping the target data to the target chart template, determining a target chart, and visualizing the target chart.

In one embodiment, the step of performing intent analysis on the user voice data and determining query terms from the user voice data includes:

Converting the user voice data into text data;

performing intention analysis on the text data to determine the intention of a user;

Identifying entity keywords in the text data, wherein the entity keywords comprise one or more of topic keywords, time range keywords and location range keywords;

and according to the user intention and the entity keyword, converting the entity keyword from spoken language into a standardized query prompt word representing the user intention.

In an embodiment, the step of inputting the query term into a preset large model for analysis to generate target data corresponding to the query term and a target chart template matched with the query term includes:

Inputting the query prompt word into the preset large model, and acquiring database query sentences and chart template generation sentences which are output by the preset large model and matched with the query prompt word;

And acquiring the target data according to the database query statement, and generating a target chart template according to the chart template generation statement.

In an embodiment, the step of inputting the query term into the preset large model to obtain a database query sentence and a chart template generation sentence output by the preset large model includes:

Inputting the query prompt word into the preset large model to obtain a database query sentence output by the preset large model, wherein the preset large model determines a data set matched with the query prompt word in a preset database based on the query prompt word, and generates and outputs a database query sentence corresponding to the data set;

the method comprises the steps of obtaining preset chart templates, inputting each preset chart template and query prompt words into a preset large model, obtaining chart template generation sentences output by the preset large models, wherein the preset large models are matched with each preset chart template based on the query prompt words, determining the preset chart templates successfully matched, and generating chart template generation sentences of the preset chart templates successfully matched according to the query prompt words.

In an embodiment, before the step of inputting each preset chart template and the query term into the preset large model, the method further includes:

creating preset chart templates of different chart types;

and constructing a template description corresponding to each preset chart template, wherein the template description comprises applicable data types, creators, creation dates, versions, function descriptions, classification labels and application scenes.

In an embodiment, the step of inputting each preset chart template and the query term into the preset large model to obtain a chart template generation statement output by the preset large model includes:

Inputting the query prompt word into the preset large model, obtaining the chart template generation statement generated by the preset large model, wherein the preset large model determines the data type of the data set according to the data set matched with the query prompt word in a preset database, and determines successfully matched chart templates according to the data type of the data set and the template description of the query prompt word and each preset chart template, and generates the chart template generation statement of the successfully matched chart template.

In an embodiment, before the step of inputting the query term into a preset large model for analysis, the method further includes:

Generating a chart dataset based on a preset multi-modal language model;

Performing fine adjustment on the preset large model through the chart data set, and determining the fine-adjusted preset large model;

the step of inputting each preset chart template and the query prompt word into the preset large model to obtain the chart template generation statement output by the preset large model comprises the following steps:

Inputting each preset chart template and the query prompt word into a trimmed preset large model, obtaining a chart template generation statement generated by the trimmed preset large model, wherein the trimmed preset large model determines the data type of the data set according to the data set matched with the query prompt word in a preset database, determines a successfully matched target chart template according to the data type of the data set and the template description of the query prompt word and each preset chart template, determines template parameters of the target chart template according to the data type of the data set and the query prompt word, generates a chart template generation statement of the target chart template according to the template parameters, and the template parameters comprise chart types, color schemes, axis labels and legends.

In addition, in order to achieve the above object, the present application also provides a large-screen data visualization device based on a large model, the large-screen data visualization device based on a large model includes:

the input module is used for acquiring externally input user voice data, carrying out intention analysis on the user voice data, and determining query prompt words representing the intention from the user voice data;

The analysis module is used for inputting the query prompt words into a preset large model for analysis and generating target data corresponding to the query prompt words and target chart templates matched with the query prompt words;

and the mapping module is used for mapping the target data to the target chart template, determining a target chart and visualizing the target chart.

In addition, in order to achieve the aim, the application also provides a large-model-based data large-screen visualization device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the computer program is configured to realize the steps of the large-model-based data large-screen visualization method.

In addition, to achieve the above object, the present application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, which when being executed by a processor implements the steps of the large-model-based data large-screen visualization method as described above.

Furthermore, to achieve the above object, the present application provides a computer program product comprising a computer program which, when being executed by a processor, implements the steps of a large model based data large screen visualization method as described above.

The method comprises the steps of obtaining user voice data input from outside, carrying out intention analysis on the user voice data, determining query prompt words representing intention from the user voice data, inputting the query prompt words into a preset large model for analysis, generating target data corresponding to the query prompt words and a target chart template matched with the query prompt words, mapping the target data to the target chart template, determining a target chart and visualizing the target chart.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and together with the description, serve to explain the principles of the application.

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the description of the embodiments or the prior art will be briefly described below, and it will be obvious to those skilled in the art that other drawings can be obtained from these drawings without inventive effort.

FIG. 1 is a schematic flow chart of a large-screen visualization method based on a large model according to an embodiment of the present application;

FIG. 2 is a schematic flow chart of a data large screen visualization method based on a large model according to an embodiment of the present application;

FIG. 3 is a schematic flow chart of a second embodiment of a large-screen visualization method for data based on a large model according to the present application;

FIG. 4 is a schematic block diagram of a large-screen data visualization device based on a large model according to an embodiment of the present application;

fig. 5 is a schematic diagram of a device structure of a hardware operating environment related to a large-screen data visualization method based on a large model in an embodiment of the present application.

The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

It should be understood that the specific embodiments described herein are merely illustrative of the technical solution of the present application and are not intended to limit the present application.

For a better understanding of the technical solution of the present application, the following detailed description will be given with reference to the drawings and the specific embodiments.

It should be noted that, the execution body of the embodiment may be a computing service device with functions of data processing, network communication and program running, such as a tablet computer, a personal computer, a mobile phone, or an electronic device, a data large screen, or the like capable of implementing the above functions. The present embodiment and the following embodiments will be described below with reference to a data large screen.

Based on this, the embodiment of the application provides a large-screen data visualization method based on a large model, and referring to fig. 1, fig. 1 is a flow diagram of a first embodiment of the large-screen data visualization method based on a large model.

In this embodiment, the large-model-based data large-screen visualization method includes steps S10 to S30:

step S10, user voice data input from outside is acquired, intention analysis is carried out on the user voice data, and query prompt words representing intention are determined from the user voice data;

It should be noted that, the user voice data is a voice signal captured by a microphone or other audio input device, and in various environments, such as a conference room, a home, an office, etc., the user may trigger the data large screen to perform a specific task through a spoken command. The captured user speech data may then be converted to text using automatic speech recognition (Automatic Speech Recognition, ASR) techniques. The text of the converted user speech data may be analyzed using Natural Language Processing (NLP) techniques to identify the content of the operation or query that the user wants to perform in order to understand the user's intent. Keywords or phrases are extracted from the text that is intended to be analyzed, and these keywords or phrases constitute query terms, and techniques such as word frequency statistics, TF-IDF (word frequency-inverse document frequency), named Entity Recognition (NER), etc. may be used to determine which terms can be selected as query terms.

It can be understood that through the application of a series of key technologies, the conversion from the user voice data to the operable query prompt words is realized, a foundation is provided for the subsequent data analysis and visualization, and the intelligence and the interactivity of the system are improved.

In one possible embodiment, referring to fig. 2, step S10 may include steps a10 to a40:

step A10, converting user voice data into text data;

it should be noted that, in this step, the voice signal is converted into text by means of voice recognition, where the main technical means is automatic voice recognition (Automatic Speech Recognition, ASR) technology, and the ASR system generally includes multiple components such as voice signal processing, feature extraction, acoustic model, language model, etc., and can convert the voice command of the user into a text form that can be further processed.

Optionally, noise removal, volume adjustment, etc. may be performed corresponding to the user voice data to improve the accuracy of voice recognition.

Step A20, carrying out intention analysis on the text data to determine the intention of a user;

It should be noted that the intent analysis needs to be applied to Natural Language Processing (NLP) technology. Semantic analysis is a branch of natural language processing that utilizes semantic analysis techniques to understand the meaning of words, phrases, and sentences in text. The intent recognition is the inference from text data of the specific content of an operation or query that the user wants to perform, and machine learning models can be used to recognize the user's intent.

It will be appreciated that step a20 implements a process of intent analysis of the text data to determine the intent of the user.

Step A30, identifying entity keywords in the text data, wherein the entity keywords comprise one or more of topic keywords, time range keywords and location range keywords;

It should be noted that, named entity recognition is a natural language processing task, which is used to identify a specific type of entity in text, such as a person name, a place name, an organization name, etc., and usually uses a machine learning or deep learning method to perform entity recognition. The topic keywords are words related to the main content of the text or the topic of discussion, the time range keywords are words representing time information, such as date, time period, etc., and the place range keywords are words representing geographical location information, such as city name, country name, etc.

Step A40, according to the user intention and the entity keywords, converting the entity keywords from spoken language into standardized query prompt words representing the user intention.

It should be noted that, the spoken language expression can be converted into the standardized expression by the context understanding, the synonym replacement, and the like, so as to facilitate the subsequent query operation. For example, nlp algorithm can convert the spoken prompt word of the query into standardized search word, for example, "traffic accident data" input by user is searched by semantic similarity, and then related data is searched after being converted into "traffic accident data".

The present embodiment uses Automatic Speech Recognition (ASR) technology to convert received user speech data into text data in text form, uses Natural Language Processing (NLP) technology to analyze the text data for intent, understands the actual needs of the user, uses Named Entity Recognition (NER) and other related natural language processing technologies to identify key entities in the text data, i.e., entity keywords, and combines the user intent and the identified entity keywords to construct accurate query prompt words. Through the steps, the system can accurately understand the demands of the user and execute corresponding operations according to the query prompt words, so that the man-machine interaction is more efficient and visual.

Step S20, inputting the query prompt words into a preset large model for analysis, and generating target data corresponding to the query prompt words and target chart templates matched with the query prompt words;

It should be noted that, the preset large model is used for analyzing the query term, understanding the meaning of the query term and judging the query intention behind the query term, predicting the data content and the chart content required by the user, retrieving the related data set from the database or the data source according to the analysis result of the query term, and determining the proper chart template.

Optionally, the preset large model may analyze the meaning of the query term, determine specific intent behind it, such as whether to look up the trend of sales data, compare the performance of different products, and so on, generate an indication about the data structure, type and source according to the content of the query term, and in general, may obtain a database query statement of the target data to obtain the required target data from a corresponding database or data warehouse, and select the most suitable chart template, such as a line graph, a histogram, a pie chart, a scatter chart, and so on, or a diversified chart template combined by multiple chart types according to the type of the data and the intent of the query term, and may adjust the chart style according to specific situations, such as a color scheme, a label, a title, and so on.

It can be appreciated that step S20 enables the transition from natural language query terms to specific data visualization, providing a user with an intuitive way to understand and explore data, and this automated process greatly simplifies the workflow of data visualization.

And step S30, mapping the target data to a target chart template, determining a target chart, and visualizing the target chart.

It should be noted that the mapping process includes mapping the target data generated in step S20 to a chart template matching the query term, where each data point or data series is assigned to a corresponding position of the chart, such as a coordinate axis, color, size, or other visual element. The chart may contain a variety of elements such as X-axis and Y-axis: defining the extent, scale, and label of the coordinate axes, data of how each data is represented in the chart, such as the color, width, or shape of the mark of the line, and headings and notes of adding headings and other notes of the chart to help explain the contents of the chart.

Specifically, it is necessary to determine how to map the data points to different parts of the chart, such as positions, colors, shapes, etc. on coordinate axes, and also to consider different data types and magnitudes, so as to ensure that the chart can adapt to the data display requirements under various conditions. For elements in the chart, it is necessary to set minimum values, maximum values, scale pitches, etc. of coordinate axes according to specific target data, define visual properties of each data, such as line patterns, mark shapes, etc., add legends to distinguish different data, and provide necessary comments to explain key information of the chart. In addition, the patterns of the chart, including color schemes, fonts and characters, background and filling, and the like, can be adaptively adjusted.

It can be appreciated that the mapping of the target data to the chart template, creating and displaying the final chart is realized in the step S30, which provides an intuitive method for the user to understand and explore the data, not only improves the efficiency of data visualization, but also enhances the interaction between the user and the data.

According to the method, the device and the system, the user voice data are obtained through the user voice data input from the outside, the intention analysis is conducted on the user voice data, the query prompt words are determined from the user voice data, the query prompt words are input into the preset large model to conduct analysis, the target data corresponding to the query prompt words and the chart template matched with the query prompt words are generated, the target data are mapped to the chart template, the target chart is determined, the target chart is visualized, the interaction between the user and the data is conducted through the voice recognition technology and the natural language processing technology, the voice command of the user can be effectively and rapidly converted into the visual chart display, the large-screen visual command of the data based on the large model is rapidly responded, the defect that the chart parameters are required to be reconfigured to be visualized when service requirements change is avoided, and the efficiency of data display is greatly improved.

In the second embodiment of the present application, the same or similar content as in the first embodiment of the present application may be referred to the above description, and will not be repeated. On this basis, referring to fig. 3, step S20 may include steps S21 to S22:

Step S21, inputting the query prompt words into a preset large model, and obtaining database query sentences and chart template generation sentences which are output by the preset large model and matched with the query prompt words;

It should be noted that, based on understanding the query term, the system needs to generate one or more database query statements. These query statements are typically SQL statements or specific API calls to extract the required data from the database. In addition to database query statements, the system also needs to generate instructions for creating charts from the content of the query terms. These instructions instruct the chart generation tool how to organize and present the data, wherein the chart template generation statement needs to specify the type of chart template, or a selected preset chart template, while also specifying chart elements such as color schemes, axis labels, legends, etc.

Step S22, target data is obtained according to the database query statement, and a target chart template is generated according to the chart template generation statement.

It should be noted that, the system needs to establish a connection with the target database, and then execute the query using the database query statement generated in step S21. Meanwhile, the system determines the type of the chart (e.g., line graph, bar graph, etc.) from the chart template generation statement, then sets various parameters (e.g., axis labels, color schemes, etc.) of the chart according to the configuration information in the chart template generation statement, creates the chart using an appropriate chart generation tool or library, and takes the query result as input data.

It will be appreciated that step S22 converts the database query statement and chart template generation statement obtained from step S21 into actual data and charts, converts natural language query into useful information, and can help users to better understand and analyze data by effectively executing database query and generating intuitive charts.

According to the method, query prompt words are input into a preset large model, database query sentences and chart template generation sentences output by the preset large model are obtained, natural language texts are understood and analyzed and converted into machine executable operation instructions, target data are obtained according to the database query sentences, chart templates are obtained according to the chart template generation sentences, and specific tasks are executed according to the generated instructions. The method and the device realize the automatic and efficient data acquisition and template selection process, reduce the need of manually writing SQL query and selecting chart types, and simultaneously reduce the error rate, and are suitable for scenes requiring quick response to user demands.

In a possible implementation, step S21 may include steps S211 to S212:

Step S211, inputting a query prompt word into a preset large model to obtain a database query sentence output by the preset large model, wherein the preset large model determines a data set matched with the query prompt word in a preset database based on the query prompt word, and generates and outputs the database query sentence corresponding to the data set;

it should be noted that the query term is a natural language description provided by the user after the optimization process, and is used for expressing specific information that they want to retrieve from the database. For example, "find all sales over 100 tens of thousands of products in 2023". The preset large model needs to have the capability of natural language understanding firstly to analyze key words and phrases in query prompt words and understand the intention of the user for query, and secondly needs to be capable of identifying key entities in the query prompt words, such as a time range, an amount threshold, a product category and the like, or needs to be capable of identifying relationships between entities, such as a relationship between a product and sales amount, a relationship between a time range and sales amount and the like. According to the analyzed key entities and the relation thereof, a preset large model can construct a basic structure of query, such as table names, field names, condition screening logic and the like, and query sentences can be optimized by adding proper index conditions and reducing unnecessary connection operation so as to improve the execution efficiency. The constructed query logic is converted into standard database query sentences so as to be executed in a database. The database query statement is one or a series of SQL statements used for retrieving a data set meeting the requirements of the query prompt words from a preset database.

As an example, assume that the query term entered is "find all products sold over 100 ten thousand in 2023". Then the keywords "2023 years", "sales exceeding 100 ten thousand", "products" and the identification entity "2023 years" may be identified as corresponding to a time range, "sales exceeding 100 ten thousand" as corresponding to a numerical condition, "products" as corresponding to a data table, and the relationship between the "sales" field and the condition "exceeding 100 ten thousand" in the relational product table may be extracted, as well as the relationship between the time range "2023 years" and the time field of the product table. According to the above construction query logic, the query table is determined to be a product table, the query field is sales, the condition is that sales are greater than 100 ten thousand in 2023, and meanwhile, as the optimization of the query logic, the query speed can be increased by using indexes, for example, indexes are built on the sales field of the product table. Finally, a database query statement may be generated that is SELECT FROM products WHERE SALES >1000000 AND year BETWEEN '2023-01-01' AND '2023-12-31'.

It can be understood that the method needs to be applied to a natural language processing technology in the step to analyze the query prompt words, understand the query intention, identify the entity and the relationship, and can effectively convert the query prompt words in the natural language form into accurate database query sentences, thereby realizing effective retrieval of the database.

Step S212, a preset chart template is obtained, each preset chart template and query prompt words are input into a preset large model, chart template generation sentences output by the preset large model are obtained, wherein the preset large model is matched with each preset chart template based on the query prompt words, the preset chart templates which are successfully matched are determined, and chart template generation sentences of the preset chart templates which are successfully matched are generated according to the query prompt words.

It should be noted that the preset chart template is a series of defined chart styles and layout options for the user to select or be recommended by the system. Each preset chart template is accompanied by descriptive text descriptions explaining what type of data presentation or specific business scenario the chart is applicable to, and other information of the template. The preset large model can be matched with the template description of each preset chart template according to the content of the query prompt word, and the chart template which best meets the requirement of the query prompt word is found.

According to the method and the device, the information can be automatically extracted from the query prompt words through the large model, database query sentences and chart template generation sentences are generated, and therefore efficient data visualization is achieved.

In a possible implementation manner, before step S212, the large-screen data visualization method based on the large model may further include step B10 to step B20:

step B10, creating preset chart templates of different chart types;

It should be noted that, when creating the template, different types of charts and suitable scenes thereof need to be considered. Common chart types include, but are not limited to, bar charts, which are suitable for comparing different categories of values, line charts, which show the trend of the data over time, pie charts, which represent the proportions of the parts, scatter charts, which explore the relationship between variables, area charts, which are similar to line charts, but emphasize the trend of the number of changes, radar charts, which are used for multi-dimensional comparison, thermodynamic charts, which show the density or intensity distribution of the data, maps, which visualize the geographic data, and bubble charts, which combine the scatter charts and size proportions, which represent the relationship of the three variables.

For each preset chart template, the following parameters should be included, namely, chart titles, axes, namely, axis labels, scales and the like, coordinate axes, namely, color identifications for distinguishing different data series, data labels, namely, specific numerical value display on data points, color schemes, namely, colors of different data series, and interaction functions, namely, hovering prompt, zooming and the like. These templates can be conveniently applied to different types of data sets to quickly generate a visualized chart.

And step B20, constructing a template description corresponding to each preset chart template, wherein the template description comprises applicable data types, creators, creation dates, versions, function descriptions, classification labels and application scenes.

It should be noted that each preset chart template description should contain the following elements of applicable data types describing what type of data the template is applicable to, e.g. numerical type, category type, time series, etc., e.g. for bar charts, it may be applicable to numerical comparison showing different categories, creator of record creating person or team name of the template, creation date of mark template to track version history, version of mark template version number to facilitate tracking modification record, function description, detailed description of main functions and features of the template, e.g. "this template supports dynamic data loading to update chart content in real time", classification tags to add tags to the template to facilitate classification and search, e.g. "trend analysis ] [ comparison ] [ time series ], application scenario describing the most applicable application scenario of the template, e.g. report, monitoring panel, etc., applicable to product sales trend analysis report.

As an example, assume we have a chart template named "product sales trend", the following is an example of its template description:

The applicable data types are numerical value (sales of products), time series (date);

creator, data visualization team;

Creation date 2024-08-28;

version V1.0;

The template is a dynamic line graph, which can display the sales trend of the product in a specific time period and support the data viewing according to the time granularity of day, week and month;

classification tag [ trend analysis ] [ time series ] [ sales ];

and the application scene comprises sales report, market analysis and business monitoring.

It will be appreciated that by detailing and describing, desirable chart templates can be more easily found and better understood and used. The working efficiency is improved, and the practical value of the chart template is also enhanced.

The method and the device for generating the visual chart are used for quickly generating the visual chart by creating the preset chart templates of different chart types and applying the preset chart templates to different data sets, and building the template description corresponding to each preset chart template, wherein the template description comprises applicable data types, creators, creation dates, versions, function descriptions, classification labels and application scenes, and the most suitable template can be conveniently searched and selected to meet specific visual requirements through the description information.

In one possible embodiment, step S212 may include step C10:

And step C10, inputting the query prompt words into a preset large model, and obtaining a chart template generation statement generated by the preset large model, wherein the preset large model determines the data type of the data set according to the data set matched with the query prompt words in a preset database, and determines successfully matched chart templates according to the data type of the data set and the template description of the query prompt words and each preset chart template, so as to generate the chart template generation statement of the successfully matched chart template.

It should be noted that, after defining that each preset chart template has a template description in step B20, a suitable preset chart template may be matched according to the template description. The large model firstly finds related data sets from a preset database based on query prompt words, determines the data types (such as numerical types, category types, time sequences and the like) of the data sets, and then selects one or more chart templates which are matched with the query prompt words best as target chart templates according to the data types and the template descriptions of the query prompt words and the preset chart templates.

Alternatively, the large model needs to understand the context in the query term to determine the data set type associated therewith, e.g., if a "time series" is mentioned in the query term, the large model may look for a data set of the time series type. The large model compares the key information in the query prompt words with the template description of each preset chart template, and the fields of the template description, such as 'applicable data type', 'functional description', and 'application scene', are used for judging whether the templates are matched. The large model selects the best chart template according to the matching degree, and a certain scoring mechanism can be adopted to quantify the matching degree. The generation statement contains specific instructions for creating the chart, such as the data set used, the type of chart selected, etc., and may also contain some configuration parameters, such as color schemes, label styles, etc.

According to the method, the most suitable chart template is automatically selected and generated by combining query prompt words of a user and description information of the preset chart template and utilizing the capability of the large model. The automated method greatly simplifies the chart creation process and ensures that the generated chart meets the specific requirements of users.

In another possible embodiment, before step S20, the large-model-based data large-screen visualization method may further include:

step C10, generating a chart data set based on a preset multi-modal language model;

step C20, fine-tuning the preset large model through the chart data set, and determining the fine-tuned preset large model;

It should be noted that, the preset multimodal language model CHARTLLAMA, CHARTLLAMA is a multimodal language model specific to the chart, which can extract information from the chart, and can create the chart according to data and requirements. It can understand and generate various types of charts including bar charts, pie charts, line charts, scatter charts, and the like. In the application, a high-quality data set specially designed for graph understanding is generated through CHARTLLAMA, and the preset large model is finely tuned through the generated data set, so that rich graph types and tasks can be provided, and training and evaluation of the large model are supported.

After fine tuning, the preset large model can acquire understanding capability of the chart based on fine tuning, and can identify and understand parameters of different preset chart templates, such as chart types, color schemes, axis labels, legends and the like.

Step S212 may further include step C30:

And step C30, inputting each preset chart template and query prompt words into a trimmed preset large model, and obtaining chart template generation sentences generated by the trimmed preset large model, wherein the trimmed preset large model determines the data type of the data set according to the data set matched with the query prompt words in a preset database, matches the template description of each preset chart template according to the data type of the data set and the query prompt words, determines a target chart template successfully matched, determines template parameters of the target chart template according to the data type of the data set and the query prompt words, and generates chart template generation sentences of the target chart template according to the template parameters, wherein the template parameters comprise chart types, color schemes, shaft labels and legends.

It should be noted that, in this embodiment, the preset large model is a large model that is trimmed by a data set in a specific field, and after trimming, the preset large model can improve understanding ability of the chart, so that template parameters of different preset chart templates can be identified and understood. The trimmed preset large model determines the data type of the data set according to the data set matched with the query prompt words in the preset database, wherein the data type may comprise a numerical value type, a category type, a time sequence type and the like. The model is matched with the template description of the preset chart template according to the data type of the data set and the query prompt word, a target chart template which meets the requirements best is determined, and template parameters of the target chart template, such as chart type, color scheme, axis label and legend, are determined according to the data type of the data set and the query prompt word. And generating a specific chart template generation statement according to the determined template parameters.

As an example, assuming that a query term "show sales for each quarter over the past five years" is provided, the procedure of this step is such that the model determines the time series data contained in the dataset according to the query term, the model looks for the most suitable template for the time series data from the preset chart templates, assuming a line chart template is found, the chart type is determined as a line chart according to the query term, the color scheme may be the default color scheme or the user specified color scheme, the axis label is the quarter (horizontal axis) and sales (vertical axis), and the legend may contain labels for different years. And generating a chart template generating statement according to the information, namely creating a line graph, displaying sales of each quarter in the past five years, wherein the horizontal axis is the quarter, the vertical axis is the sales, and using a default color scheme. "

The present embodiment enhances the ability of large models to generate charts by introducing additional fine tuning steps. Through generating a high-quality chart dataset in advance and fine tuning a large model, the model can more accurately understand the user intention, identify and understand parameters of different preset chart templates, and generate chart template generation sentences meeting the requirements, so that the quality of chart generation is improved, and the whole process is more automatic and efficient.

It should be noted that the foregoing examples are only for understanding the present application, and do not constitute a limitation of the large-model-based data large-screen visualization method of the present application, and it is within the scope of the present application to make more simple transformations based on this technical concept.

The application also provides a large-screen data visualization device based on a large model, referring to fig. 4, the large-screen data visualization device based on a large model comprises:

An input module 10, configured to obtain externally input user voice data, perform intent analysis on the user voice data, and determine a query term representing the intent from the user voice data;

the analysis module 20 is configured to input the query term into a preset large model for analysis, and generate target data corresponding to the query term and a target chart template matched with the query term;

The mapping module 30 is configured to map the target data to the target chart template, determine a target chart, and visualize the target chart.

The large-model-based data large-screen visualization device provided by the application can solve the technical problem that the large screen can be visualized only by reconfiguration due to service demand change by adopting the large-model-based data large-screen visualization method in the embodiment. Compared with the prior art, the large-model-based data large-screen visualization device has the same beneficial effects as the large-model-based data large-screen visualization method provided by the embodiment, and other technical features in the large-model-based data large-screen visualization device are the same as the features disclosed by the method of the embodiment, and are not repeated herein.

The application provides a large-scale data screen visualization device based on a large model, which comprises at least one processor and a memory in communication connection with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the large-scale data screen visualization method based on the large model in the first embodiment.

Referring now to FIG. 5, a schematic diagram of a large model based data large screen visualization device suitable for use in implementing embodiments of the present application is shown. The large model-based data large screen visualization device in the embodiment of the present application may include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal DIGITAL ASSISTANT: personal digital assistants), PADs (Portable Application Description: tablet computers), PMPs (Portable MEDIA PLAYER: portable multimedia players), vehicle-mounted terminals (e.g., vehicle-mounted navigation terminals), and the like, and fixed terminals such as digital TVs, desktop computers, and the like. The large model-based data large screen visualization device shown in fig. 5 is only one example and should not impose any limitation on the functionality and scope of use of embodiments of the present application.

As shown in fig. 5, the large model-based data large screen visualization apparatus may include a processing device 1001 (e.g., a central processor, a graphic processor, etc.) that may perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access Memory (RAM: random Access Memory) 1004. In the RAM1004, various programs and data required for the operation of the large-model-based data large-screen visualization apparatus are also stored. The processing device 1001, the ROM1002, and the RAM1004 are connected to each other by a bus 1005. An input/output (I/O) interface 1006 is also connected to the bus. In general, a system including an input device 1007 such as a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc., an output device 1008 including a Liquid crystal display (LCD: liquid CRYSTAL DISPLAY), a speaker, a vibrator, etc., a storage device 1003 including a magnetic tape, a hard disk, etc., and a communication device 1009 may be connected to the I/O interface 1006. The communicator 1009 may allow the large model based data large screen visualization device to communicate wirelessly or wiredly with other devices to exchange data. While a large model-based data large screen visualization device with various systems is shown in the figures, it should be understood that not all of the illustrated systems are required to be implemented or provided. More or fewer systems may alternatively be implemented or provided.

In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. The above-described functions defined in the method of the disclosed embodiment of the application are performed when the computer program is executed by the processing device 1001.

The large-model-based data large-screen visualization equipment provided by the application can solve the technical problem that the large screen can be visualized only by reconfiguration due to service demand change by adopting the large-model-based data large-screen visualization method in the embodiment. Compared with the prior art, the large-model-based data large-screen visualization device has the same beneficial effects as the large-model-based data large-screen visualization method provided by the embodiment, and other technical features in the large-model-based data large-screen visualization device are the same as the features disclosed by the method of the previous embodiment, and are not repeated herein.

It is to be understood that portions of the present disclosure may be implemented in hardware, software, firmware, or a combination thereof. In the description of the above embodiments, particular features, structures, materials, or characteristics may be combined in any suitable manner in any one or more embodiments or examples.

The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

The present application provides a computer readable storage medium having computer readable program instructions (i.e., a computer program) stored thereon for performing the large model-based data large screen visualization method of the above embodiments.

The readable storage medium provided by the application is a computer readable storage medium, and the computer readable storage medium stores computer readable program instructions (namely computer program) for executing the large-screen data visualization method based on the large model, so that the technical problem that the large screen needs to be reconfigured for visualization due to service demand change can be solved. Compared with the prior art, the beneficial effects of the computer readable storage medium provided by the application are the same as those of the large-model-based data large-screen visualization method provided by the embodiment, and the detailed description is omitted.

The application also provides a computer program product comprising a computer program which when executed by a processor implements the steps of a large model based data large screen visualization method as described above.

The computer program product provided by the application can solve the technical problem that the large screen can be visualized only by reconfiguration due to service demand change. Compared with the prior art, the beneficial effects of the computer program product provided by the application are the same as those of the large-model-based data large-screen visualization method provided by the embodiment, and are not repeated here.

The foregoing description is only a partial embodiment of the present application, and is not intended to limit the scope of the present application, and all the equivalent structural changes made by the description and the accompanying drawings under the technical concept of the present application, or the direct/indirect application in other related technical fields are included in the scope of the present application.

Claims

1. The large-screen data visualization method based on the large model is characterized by comprising the following steps of:

2. The large model based data large screen visualization method of claim 1, wherein the step of performing intent analysis on the user voice data and determining query terms from the user voice data comprises:

Converting the user voice data into text data;

3. The large-scale data visualization method according to claim 1, wherein the step of inputting the query term into a preset large model for analysis to generate target data corresponding to the query term and a target chart template matched with the query term comprises:

4. The large-scale visualization method of data based on large model of claim 3, wherein the step of inputting the query term into the preset large model to obtain the database query sentence and the chart template generation sentence output by the preset large model comprises:

5. The large-scale visualization method of data based on a large model of claim 4, wherein before the step of inputting each of the preset chart templates and the query term into the preset large model, further comprising:

creating preset chart templates of different chart types;

6. The large-scale visualization method of claim 5, wherein the step of inputting each of the preset chart templates and the query term into the preset large model to obtain the chart template generation statement output by the preset large model comprises the steps of:

7. The large-model-based data large-screen visualization method of claim 5, wherein before the step of inputting the query term into a preset large model for analysis, further comprising:

Generating a chart dataset based on a preset multi-modal language model;

8. A large-model-based data large-screen visualization device, characterized in that the large-model-based data large-screen visualization device comprises:

9. A large model based data large screen visualization apparatus, characterized in that the apparatus comprises a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the large model based data large screen visualization method according to any of claims 1 to 7.

10. A storage medium, characterized in that the storage medium is a computer-readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, implements the steps of the large model based data large screen visualization method according to any of claims 1 to 7.