Reproduce “From 'Here' to 'There': Exploring Proximity Semantics in Multimodal Data Exploration” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dennis Bromley, Diana Wang, Vidya Setlur
Researchers from Tableau Research introduce Proximity Semantics (PS), a conceptual framework that explains how users implicitly leverage spatial, temporal, and visual proximity to disambiguate meaning when combining free-form sketching, natural language, and visual annotations for data exploration. A multimodal research probe demonstrated that users frequently employ these proximity cues, leading to an average System Usability Scale (SUS) score of 81 for the system's expressiveness.
Key insights
- Users intuitively employ spatial, temporal, and visual proximity to convey meaning and establish relationships between disparate multimodal elements in their data queries.
- Proximity Semantics (PS) is formalized as a reference-resolution framework that establishes 'proximity bindings' between elements across various measurable spaces (Euclidean, semantic, temporal), providing a mechanism for deictic disambiguation.
- PS primitives include 'proximity bindings' (elements appearing 'close' together within an interpretive threshold) and 'connectors' (like arrows that logically collapse distance), enabling the formation of PS binding chains across multiple spaces.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.