CN-121071134-B - Data visual narrative generation method, device, equipment and medium based on side-by-side

CN121071134BCN 121071134 BCN121071134 BCN 121071134BCN-121071134-B

Abstract

The application discloses a data visualization narrative generation method, device, equipment and medium based on side whitening, and relates to the technical field of artificial intelligence. The method comprises the steps of generating sentence-level side features according to side contents input by a user, generating corresponding scenario information for the side features, determining data feature information corresponding to data feature names in the side features according to data input by the user, generating visual structure information of the side features according to the data feature information, generating animation structure information of the side features according to the side features and the scenario information, the data feature information and the visual structure information corresponding to the side features, obtaining data to be rendered based on the side features and the scenario information, the data feature information, the visual structure information and the animation structure information corresponding to the side features, rendering the data to be rendered, and generating a data visual narrative picture. And generating the visual narrative according to the user-defined side drive, and generating the highly personalized visual narrative content.

Inventors

HAN DONGMING
NI JUN
DING JIEYING
Han Rufei
HU YI
ZHOU CHANGJU

Assignees

杭州同花顺数据开发有限公司

Dates

Publication Date: 20260508
Application Date: 20250814

Claims (10)

1. A method of data visualization narrative generation based on paralogues, comprising: generating sentence-level bystander according to the bystander content input by the user; Generating corresponding scenario information for the bystander, wherein the scenario information comprises theme information, emotion information and lens language information; According to the data input by the user, determining data characteristic information corresponding to the data characteristic names in the side-by-side characteristics; Generating visual structure information of the bystander according to the data characteristic information; Generating animation structure information of the white-keeping feature according to the white-keeping feature, and scenario information, data feature information and visual structure information corresponding to the white-keeping feature; obtaining data to be rendered based on the side white feature and scenario information, data feature information, visual structure information and animation structure information corresponding to the side white feature, rendering the data to be rendered, and generating a data visual narrative picture; The topic information comprises one or more of topic type, topic core, topic background, topic key, topic style and topic related information, the emotion information comprises one or more of emotion type, emotion intensity and emotion change speed, and the shot language information comprises one or more of shot type, shot function, shot rhythm, shot angle, shot Jing Bie and shot style; The generation process of the theme information comprises the steps of carrying out paragraph-level semantic segmentation on the bystander content to obtain a plurality of paragraph-level bystander, generating corresponding theme information for each paragraph-level bystander according to a theme knowledge base, and determining the theme information corresponding to the bystander according to the bystander paragraph to which the bystander is belonged.
2. The bystander-based white-out of claim 1a data visualization narrative generation method of (c), it is characterized in that the method comprises the steps of, the generating sentence-level bystander according to the bystander content input by a user comprises the following steps: inputting the bystander content, a professional knowledge base, a bystander feature prompt word and the name of a data feature sample into a large language model, wherein the professional knowledge base is used for correcting the description of the bystander content; and extracting a data feature name from the bystander content according to the name of the data feature sample by using the large language model, and generating a corresponding bystander containing the data feature name for sentence-level bystander in the bystander content according to the bystander feature prompt word.
3. The method of claim 2, wherein the expertise repository contains names, descriptions, and computational principles of index nouns; The data structure of the bystander includes one or more of the position of the bystander content in the bystander content, the index noun, the time, the main body and the name of the data characteristic.
4. The method of generating a visual narrative based on data by side of claim 2, wherein said determining data characteristic information corresponding to data characteristic names in said side of features from data entered by a user comprises: the method comprises the steps of inputting the bystander features and a data feature calculation library into a large language model, wherein the data feature calculation library comprises names, descriptions and calculation codes of data feature samples, and the calculation codes are used for calculating actual values of the data features; And matching corresponding calculation codes for the data feature names in the bystander features according to the names, the names and the descriptions of the data feature samples in the data feature calculation library by using the large language model, and calculating data feature information corresponding to the data feature names by using the calculation codes.
5. The bystander-based white-out of claim 1 a data visualization narrative generation method of (c), the method is characterized in that the generating of the corresponding scenario information for the bystander comprises the following steps: Inputting the side white feature, the scenario feature prompt word, the topic knowledge base, the emotion knowledge base and the lens language knowledge base into a large language model, wherein the scenario feature prompt word is used for standardizing a data structure of scenario information and guiding the output of the model; And generating corresponding theme information, emotion information and shot language information for the bystander by using the large language model.
6. The bystander-based white-out of claim 1a data visualization narrative generation method of (c), it is characterized in that the method comprises the steps of, the generating the visual structure information of the bystander according to the data characteristic information comprises the following steps: inputting the side white feature, the data feature information, the visual knowledge base and the visual generation prompt word into a large language model, wherein the visual generation prompt word is used for standardizing the data structure of visual structural information and guiding the output of the model; and generating visual structure information for the bystander by using the large language model, wherein the visual structure information contains a chart type.
7. The method of generating a visual narrative based on a side-by-side data according to any one of claims 1 to 6, wherein said generating animated structural information of said side-by-side feature based on said side-by-side feature and scenario information, data feature information, and visual structural information corresponding to said side-by-side feature, comprises: Inputting the bystander, the scenario information, the data characteristic information, the visual structure information, an animation knowledge base and an animation structure generating prompt word into a large language model, wherein the animation structure generating prompt word is used for standardizing a data structure of the animation structure information and guiding the output of the model; and generating animation structure information corresponding to the side features by using the large language model.
8. A data visualization narrative generation apparatus based on paralogue comprising: The white-out feature generation module, generating sentence-level bystander features according to bystander contents input by a user; The system comprises a scenario information generation module, a scenario information generation module and a processing module, wherein the scenario information generation module is used for generating corresponding scenario information for the bystander, and the scenario information comprises theme information, emotion information and lens language information; The data characteristic information determining module is used for determining data characteristic information corresponding to the data characteristic names in the side white characteristic according to the data input by the user; The visual structure information generation module is used for generating visual structure information of the bystander according to the data characteristic information; the animation structure information generation module is used for generating animation structure information of the white-out feature according to the white-out feature, and scenario information, data feature information and visual structure information corresponding to the white-out feature; The rendering module is used for obtaining data to be rendered based on the side white features and scenario information, data feature information, visual structure information and animation structure information corresponding to the side white features, rendering the data to be rendered and generating a data visual narrative picture; The method comprises the steps of generating theme information, wherein the theme information comprises one or more of a theme type, a theme core, a theme background, a theme key, a theme style and theme related information, the emotion information comprises one or more of an emotion type, an emotion intensity and an emotion change speed, the shot language information comprises one or more of a shot type, a shot function, a shot rhythm, a shot angle, a shot Jing Bie and a shot style, the generating process of the theme information comprises the steps of carrying out paragraph-level semantic segmentation on the bystander content to obtain a plurality of paragraph-level bystander, generating corresponding theme information for each paragraph-level bystander according to a theme knowledge base, and determining the theme information corresponding to the bystander according to the bystander to which the bystander is belonged.
9. An electronic device, comprising: A memory for storing a computer program; a processor for executing the computer program to implement a method of spectator-based data visualization narrative generation as set forth in any one of claims 1 through 7.
10. A computer readable storage medium for storing a computer program, wherein the computer program when executed by a processor implements a method of spectator-based data visualization narrative generation according to any one of claims 1 to 7.

Description

Data visual narrative generation method, device, equipment and medium based on side-by-side Technical Field The invention relates to the technical field of artificial intelligence, in particular to a data visualization narrative generation method, device, equipment and storage medium based on side whitening. Background Data visualization and narrative generation techniques have found widespread use in recent years to aid users in better understanding and analyzing data by converting the data into intuitive charts and narrative. In the prior art, the data visualization and narrative generation schemes mainly comprise three types, namely, template-based visualization and narrative generation, artificial intelligence-based generation and visualization tools combined with user interaction, but the schemes have poor personalized generation effect and lack personalized expression of a user visual angle. Disclosure of Invention In view of the above, the present invention aims to provide a method, a device and a medium for generating a visual narrative based on data by a user, which can support the generation of the visual narrative according to the user-defined by-user drive and generate highly personalized visual narrative content. The specific scheme is as follows: In a first aspect, the application discloses a data visualization narrative generation method based on side whitening, comprising the following steps: generating sentence-level bystander according to the bystander content input by the user; Generating corresponding scenario information for the bystander, wherein the scenario information comprises theme information, emotion information and lens language information; According to the data input by the user, determining data characteristic information corresponding to the data characteristic names in the side-by-side characteristics; Generating visual structure information of the bystander according to the data characteristic information; Generating animation structure information of the white-keeping feature according to the white-keeping feature, and scenario information, data feature information and visual structure information corresponding to the white-keeping feature; And obtaining data to be rendered based on the side white feature and scenario information, data feature information, visual structure information and animation structure information corresponding to the side white feature, and rendering the data to be rendered to generate a data visual narrative picture. Optionally, the generating the sentence-level bystander according to the bystander content input by the user includes: inputting the bystander content, a professional knowledge base, a bystander feature prompt word and the name of a data feature sample into a large language model, wherein the professional knowledge base is used for correcting the description of the bystander content; and extracting a data feature name from the bystander content according to the name of the data feature sample by using the large language model, and generating a corresponding bystander containing the data feature name for sentence-level bystander in the bystander content according to the bystander feature prompt word. Optionally, the expertise base includes names, descriptions and calculation principles of index nouns; The data structure of the bystander includes one or more of the position of the bystander content in the bystander content, the index noun, the time, the main body and the name of the data characteristic. Optionally, the determining, according to the data input by the user, the data feature information corresponding to the data feature names in the side feature includes: the method comprises the steps of inputting the bystander features and a data feature calculation library into a large language model, wherein the data feature calculation library comprises names, descriptions and calculation codes of data feature samples, and the calculation codes are used for calculating actual values of the data features; And matching corresponding calculation codes for the data feature names in the bystander features according to the names, the names and the descriptions of the data feature samples in the data feature calculation library by using the large language model, and calculating data feature information corresponding to the data feature names by using the calculation codes. Optionally, the generating corresponding scenario information for the bystander includes: Inputting the side white feature, the scenario feature prompt word, the topic knowledge base, the emotion knowledge base and the lens language knowledge base into a large language model, wherein the scenario feature prompt word is used for standardizing a data structure of scenario information and guiding the output of the model; Generating corresponding topic information, emotion information and shot language information for the bystander by using the big language model, wherein the