Invention Title:

AUTOMATED GENERATION OF DATA VISUALIZATIONS AND INFOGRAPHICS USING LARGE LANGUAGE MODELS AND DIFFUSION MODELS

Publication number:

US20240185490

Publication date:
Section:

Physics

Class:

G06T11/206

Inventor:

Assignee:

Applicant:

Drawings (4 of 9)

Smart overview of the Invention

The patent application describes a system utilizing machine learning models to automate the generation of data visualizations and infographics. This system processes raw data to create candidate analytics and scenarios for visualization using summary data. The approach allows users to visualize data in a context that aligns with their intentions, enhancing the accessibility and understanding of complex data sets. By employing visualization code scaffolds, the system generates programmatic outputs that correspond to these analytics, facilitating the creation of visualizations.

Background

With the increasing availability of data, the need for effective data visualization has grown, presenting challenges in designing informative and persuasive visuals. The complexity of data visualization often requires significant effort and skill, which can limit its effectiveness. The invention aims to address these challenges by automating the generation of visualizations, thus making data more accessible and understandable without requiring extensive expertise from users.

System Functionality

The system utilizes a language model to generate summary data and visualization scenarios from raw data. This process involves creating prompts for users to identify relevant data fields, which are then used to develop a visualization scenario. The scenario specifies the type of visual representation and extracts necessary fields for rendering. Users can interact with this process, providing input that influences the final visualization output.

Visualization Code Generation

Once a visualization scenario is established, the system generates visualization code scaffolds using a generative machine learning model. This model produces programmatic outputs that define graphical elements like types and colors for the visualization. The process includes validating these outputs to ensure accuracy in rendering. The system can also utilize existing code scaffolds retrieved from various sources, enhancing flexibility and efficiency in generating visualizations.

Infographic Creation

The system extends its capabilities to generate infographics using text-to-image generation models, such as diffusion models. By combining visualizations with prompts specifying context and artistic styles, these models produce infographics that visually represent data in an engaging manner. This feature allows users to create visually appealing representations that convey information effectively across different artistic styles.