LEVA: Using Large Language Models to Enhance Visual Analytics

TL;DR

This work proposes LEVA, a framework that uses large language models to enhance users’ VA workflows at multiple stages: onboarding, exploration, and summarization, and demonstrates how LEVA can be integrated into existing visual analytics systems.

Abstract

Visual analytics supports data analysis tasks within complex domain problems. However, due to the richness of data types, visual designs, and interaction designs, users need to recall and process a significant amount of information when they visually analyze data. These challenges emphasize the need for more intelligent visual analytics methods. Large language models have demonstrated the ability to interpret various forms of textual data, offering the potential to facilitate intelligent support for visual analytics. We propose LEVA, a framework that uses large language models to enhance users' VA workflows at multiple stages: onboarding, exploration, and summarization. To support onboarding, we use large language models to interpret visualization designs and view relationships based on system specifications. For exploration, we use large language models to recommend insights based on the analysis of system status and data to facilitate mixed-initiative exploration. For summarization, we present a selective reporting strategy to retrace analysis history through a stream visualization and generate insight reports with the help of large language models. We demonstrate how LEVA can be integrated into existing visual analytics systems. Two usage scenarios and a user study suggest that LEVA effectively aids users in conducting visual analytics.

Figures (8)

• Figure 1: The LEVA framework proposes strategies for leveraging LLMs to enhance visual analytics workflows, starting from onboarding and exploration to summarization. The architecture (middle) connects analysts (top) and LLMs (bottom) to achieve mixed-initiative exploration through interactive interfaces and guidance strategies of LLMs.
• Figure 2: The LLM-enhanced visual analytics workflow shows how LLMs contribute to the progress of the analysis. The LLMs support visualization understanding, mixed-initiative guidance, and automatic summary while users experience onboarding, exploration, and summarization. Onboarding refers to data understanding, visualization, and interaction perception. Exploration refers to insight discovery, hypothesis formulation, and validation. Summarization refers to selective reporting.
• Figure 3: The integration overview for augmenting a VA system with LLM capabilities involves both existing extensions and LLM-powered components in LEVA’s implementation. The enhancement brings new capabilities for end-users at three stages.
• Figure 4: An implementation of LEVA comprises of four components. Users can communicate with LLMs and control the insight annotations in (a) Chat view; the recommended insights for next step analysis from LLMs are updated in (b) Original system view; Users can retrace the interaction history in (d) Interaction stream view; Once a historical analysis path is selected in (d), the generated insight report will display in (e) Report view.
• Figure 5: An onboarding tour example of the VAST challenge system. (a) Initiation via the onboarding button, (b) Introductions to data meanings of "mbdata" and "ccdata", (c) The coordination of keyword view and timeline view based on selected keywords, and (d) The visual encoding of hexagon colors representing risk levels in specific regions.