Similarity Framework for Visualization Retrieval

Effective visualization retrieval necessitates a clear definition of similarity. Despite the increasing body of work in specialized visualization retrieval systems, a systematic approach to understanding visualization similarity remains absent. We introduce the Similarity Framework for Visualization Retrieval (Safire), a conceptual model that frames visualization similarity along two dimensions: comparison criteria and representation modalities. Comparison criteria identify what aspects make visualizations similar: data, visual encoding, interaction, style, and metadata, while considering derived properties such as data-centric and human-centric measures. Safire connects what to compare with how comparisons are executed through representation modalities. We categorize existing representation approaches into four groups based on abstraction level: raster image, vector image, specification, and natural language description, guiding what is computable and comparable. We analyze several visualization retrieval approaches with Safire to demonstrate its practical value in clarifying similarity considerations. The findings reveal how specific similarity and representation aspects align in different use cases. One significant insight is that the choice of representation modality is not only specific to implementation but an important decision that shapes retrieval capabilities and constraints. Based on our analysis, we provide recommendations and discuss broader implications for multimodal learning, AI applications, and visualization reproducibility.