Data Therapist: Eliciting Domain Knowledge from Subject Matter Experts Using Large Language Models

TL;DR

The paper introduces Data Therapist, a system leveraging large language models to help domain experts externalize tacit knowledge about datasets, improving data visualization understanding and design.

Contribution

It presents a novel web-based system that combines iterative Q&A and interactive annotation, enabling experts to articulate implicit data knowledge using AI assistance.

Findings

Revealed common patterns in expert reasoning about data.

Demonstrated the system's potential to enhance visualization design.

Identified opportunities for AI to support domain-specific data understanding.

Abstract

Effective data visualization requires not only technical proficiency but also a deep understanding of the domain-specific context in which data exists. This context often includes tacit knowledge about data provenance, quality, and intended use, which is rarely explicit in the dataset itself. Motivated by growing demands to surface tacit knowledge, we present the Data Therapist, a web-based system that helps domain experts externalize such implicit knowledge through a mixed-initiative process combining iterative Q&A with interactive annotation. Powered by a large language model, the system automatically analyzes user-supplied datasets, prompts users with targeted questions, and supports annotation at varying levels of granularity. The resulting structured knowledge base can inform both human and automated visualization design. A qualitative study with expert pairs from Accounting, Political Science, and Computer Security revealed recurring patterns in how expert reason about their data and highlighted opportunities for AI support to enhance visualization design.