Supporting Relation-Finding in Neuroscientific Text Collections using Augmented Reality: A Design Exploration

Abstract

Maintaining an overview of publications in the neuroscientific field is challenging, especially with an eye to finding relations at scale (between, e.g., brain regions and diseases) - both those well-studied and nascent. To support neuroscientists in this challenge, we used a design-based research approach to investigate whether Augmented Reality (AR) could serve as a platform to make automated methods more accessible and integrated into current work practices. Building on insights from Text and Immersive Analytics, as well as two prior user studies, we identified information and design requirements (e.g., “highlight, not hide” and “augment, not replace”), which we embodied in a system design and implementation focussed on the analysis of co-occurrences in text collections. We evaluated our system using a scenario-based video survey with a diverse sample of neuroscientists and other domain experts, focusing on the quality of our design choices and participants' willingness to adopt such an AR system in their regular literature review practices. The AR-tailored epistemic and representation design of our system were generally perceived as suitable for performing complex analytics. We therefore see opportunities in pushing our generalisable interaction paradigm further in augmenting intellectual activities. We also discuss several fundamental issues with our chosen 3D visualisations, making steps towards addressing the question whether AR a suitable medium for relation-finding in document collections.