From Natural Language to Data Visualization (NL2VIS) with Large Language Model and Pattern Matching

Abstract

In the rapidly evolving domain of artificial intelligence, Large Language Models (LLMs) such as ChatGPT have made remarkable advancements, revolutionizing how users consume and generate content. The current research on LLM-based generative tools has primarily been concentrated on the generation of text, codes, or images. This paper presents an investigation into the application of LLMs for automating the creation of data visualizations from natural language queries. The proposed approach leverages LLMs for generating data queries, coupling this capability with a pattern matching strategy aimed at creating visualizations ( eventually auto-generated dashboards) based on the extracted data. Furthermore, we introduce a novel domain-specific data visualization language (DVL). This DVL can be effortlessly translated into executable code compatible with various data visualization libraries. This exploration into the confluence of visualization, NLP, and human-computer interaction, is substantiated by our work in a practical industrial context, enriching a live product. Our findings aspire to contribute to the ongoing dialogue of how NLP techniques and interactive visualizations can be harmonized to bolster data-driven communication and analytical discourse.