Jupybara: Operationalizing a Design Space for Actionable Data Analysis and Storytelling with LLMs

2025-01-28

AI Summary

Exploratory Data Analysis (EDA) and data storytelling often yield insights lacking actionable utility—i.e., insufficient semantic precision, rhetorical persuasiveness, or practical relevance. Method: We propose the first design space for *actionable* EDA and narrative generation, integrating these three dimensions. Our approach introduces a design-space-aware structured prompting mechanism and a multi-agent LLM collaboration architecture, implemented within Jupyter to support end-to-end AI-assisted analysis and story synthesis—while preserving domain knowledge adaptation and user controllability. Contribution/Results: Expert evaluation demonstrates strong performance in usability, guidance quality, explainability, and repairability. The system significantly improves accuracy in analytical strategy selection and enhances the practical utility of generated narratives, establishing a foundation for action-oriented data intelligence.

Technology Category

• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Bridging structured and unstructured data

Abstract

Mining and conveying actionable insights from complex data is a key challenge of exploratory data analysis (EDA) and storytelling. To address this challenge, we present a design space for actionable EDA and storytelling. Synthesizing theory and expert interviews, we highlight how semantic precision, rhetorical persuasion, and pragmatic relevance underpin effective EDA and storytelling. We also show how this design space subsumes common challenges in actionable EDA and storytelling, such as identifying appropriate analytical strategies and leveraging relevant domain knowledge. Building on the potential of LLMs to generate coherent narratives with commonsense reasoning, we contribute Jupybara, an AI-enabled assistant for actionable EDA and storytelling implemented as a Jupyter Notebook extension. Jupybara employs two strategies -- design-space-aware prompting and multi-agent architectures -- to operationalize our design space. An expert evaluation confirms Jupybara's usability, steerability, explainability, and reparability, as well as the effectiveness of our strategies in operationalizing the design space framework with LLMs.

Problem

• Data Analysis
• Information Extraction
• Data Storytelling

Innovation

• Jupybara
• Advanced Language Models
• Data Storytelling