Collection Wide

Collection-wide analysis focuses on understanding and leveraging similarities and structures within large datasets of items, such as images, documents, or layouts. Current research emphasizes developing novel similarity measures and interactive visualization tools to explore high-dimensional data, often employing techniques like optimal transport and graph-based methods to capture complex relationships between items. This work is significant for improving the efficiency of tasks like thematic collection design, unsupervised document structure extraction, and the evaluation of generative models, ultimately enabling more effective analysis and manipulation of large, complex datasets.