PreVue.space: A Multi-View Dashboard with LLM-Guided Support for Astronomical Discovery]

Abstract

Astronomers often contend with massive datasets and fragmented, sequential web-based workflows. Navigating multiple database pages to enter celestial coordinates, retrieve sky surveys, filter by date, and download catalogs can disrupt exploratory thinking and hinder efficient hypothesis formulation. We propose Pre-Vue, a web-based interface concept that reduces fragmentation in exploratory survey workflows by linking sky-region selection to coordinated multi-band image panels, catalog previews, and natural-language generation of comparison views within a single workspace. Our contributions are: (1) an integrated exploratory workspace that coordinates sky-region selection, multi-band image views, catalog previews, and LLM-assisted view generation from natural-language requests; and (2) expert-informed design considerations for reducing fragmentation in astronomical survey workflows, derived from iterative feedback with astronomers at the California Institute of Technology.