31:[{"id":"51544fbfac850ad49da5843eaff00d6502ae64effaec4001ea95475b20562308ce8bc69228e2ecd69e16185d0905b99e2d7a3c1e8c992506ae5e14141a2f476a","href":"https://static.observableusercontent.com/files/51544fbfac850ad49da5843eaff00d6502ae64effaec4001ea95475b20562308ce8bc69228e2ecd69e16185d0905b99e2d7a3c1e8c992506ae5e14141a2f476a","name":"title_metadata@3.csv","type":"text/csv","lastModified":1651736933253.007,"size":115167},{"id":"8b6662c61823cf724635793afc98b1bc47fcc23bee95483d46592ea090bf5adb004416307179d068cbff132c7918a6611b728186832a4459365b1851bb4e3565","href":"https://static.observableusercontent.com/files/8b6662c61823cf724635793afc98b1bc47fcc23bee95483d46592ea090bf5adb004416307179d068cbff132c7918a6611b728186832a4459365b1851bb4e3565","name":"title_metadata@1.csv","type":"text/csv","lastModified":1651134287401.492,"size":81507},{"id":"b523ce354a88a8e6d51688a31823e974bf2b56f101b5c392af8ffe4dd54d3fc9b181277daefa460f4888985c1dd8f86587736ee3f338a361293c10e4df9b6a14","href":"https://static.observableusercontent.com/files/b523ce354a88a8e6d51688a31823e974bf2b56f101b5c392af8ffe4dd54d3fc9b181277daefa460f4888985c1dd8f86587736ee3f338a361293c10e4df9b6a14","name":"iconclass_metadata.csv","type":"text/csv","lastModified":1651134410483.032,"size":76178},{"id":"679f64ef584fb54337a2f441e9d5a4711425c707c36a8d777ca06573763669a290aa07cfa8d397b3070e12dda860ffd155525cf1f7739ca315e9988e952a6bb8","href":"https://static.observableusercontent.com/files/679f64ef584fb54337a2f441e9d5a4711425c707c36a8d777ca06573763669a290aa07cfa8d397b3070e12dda860ffd155525cf1f7739ca315e9988e952a6bb8","name":"visual_sim.csv","type":"text/csv","lastModified":1651218464129.66,"size":416098}] 2f:{"title":"ReFa Dimension Reduction and Iconography","cells":[{"id":0,"value":"# ReFa Dimension Reduction and Iconography","pinned":false,"hidden":false,"mode":"md"},{"id":406,"value":"

This notebook visualises the visual items used in the project ReFa by the UCLAB.

Divided into ${categories.length} categories, the items are displayed according to their visual and metadata similarities. From the Iconclass notation system, to the closeness of the items' titles.

\nAll items and notations used to describe the objects were converted into vectors and a UMAP was built around these values, grouping the objects by similarity in the metadata.\n

The data was processed based on the notebooks from: [@sinanatra/image-tsne](https://github.com/sinanatra/image-tsne).

Adapted from the fisheye notebook of [cpietsch](https://observablehq.com/@cpietsch/explorawolke-fisheye-view-snappy).

","pinned":false,"hidden":false,"mode":"md"},{"id":582,"value":"viewof data = Inputs.radio(categories, {value:categories[0], label: \"Categories\", format: x => x.label})","pinned":false,"hidden":false,"mode":"ojs"},{"id":1575,"value":"fullscreen = {\n const button = html`