FinFlier: Automating Graphical Overlays for Financial Visualizations with Knowledge-Grounding Large Language Model

TL;DR

FinFlier tackles the challenge of automatically overlaying graphical cues on financial charts to accompany narratives. It introduces a two-stage approach: a text-data binding module powered by a knowledge-grounding LLM with output constraint, chain-of-thought, and dynamic prompting, and a graphics overlaying module that applies narrative-aware overlays and sequencing. A design-space study and a 1752-chart corpus underpin the correspondence rules between narrative vocabulary and overlay types, enabling rapid, configurable layered charts via an interactive interface. Extensive case studies, quantitative text-data binding evaluation, and user studies demonstrate improved interpretability, engagement, and user experience, with open-source data and code to spur future research.

Abstract

Graphical overlays that layer visual elements onto charts, are effective to convey insights and context in financial narrative visualizations. However, automating graphical overlays is challenging due to complex narrative structures and limited understanding of effective overlays. To address the challenge, we first summarize the commonly used graphical overlays and narrative structures, and the proper correspondence between them in financial narrative visualizations, elected by a survey of 1752 layered charts with corresponding narratives. We then design FinFlier, a two-stage innovative system leveraging a knowledge-grounding large language model to automate graphical overlays for financial visualizations. The text-data binding module enhances the connection between financial vocabulary and tabular data through advanced prompt engineering, and the graphics overlaying module generates effective overlays with narrative sequencing. We demonstrate the feasibility and expressiveness of FinFlier through a gallery of graphical overlays covering diverse financial narrative visualizations. Performance evaluations and user studies further confirm system's effectiveness and the quality of generated layered charts.

Figures (14)

• Figure 1: The narrative introduces the change in GDP growth during the 2008 Great Recession. (a) displays the side-by-side interplay without visual linking, (b) shows the side-by-side interplay with visual linking, and (c) utilizes graphical overlays.
• Figure 2: Examples of graphical overlay techniques organized by category. The four rows from top to bottom correspond to the four chart types: single-line chart, multi-line chart, single-bar chart, multi-bar chart, and the nine columns correspond to the nine categories.
• Figure 3: The statistical results of graphical overlays in our collected corpus of layered charts.
• Figure 4: The top ten trend vocabularies summarized in the collected financial narrative dataset with their visual patterns.
• Figure 5: Statistics results on the correspondence between financial vocabularies and graphical overlays. The right shows the correspondence based on both the statistics and discussions with practitioners.