We developed our own similarity metric (MDSSM); it combines four sub-metrics that we found interesting and relevant to the archive: visual similarity of archive items' images, similarity of objects contained in those images, their external contextual similarity as measured by comparison of text found in knowledge bases such as Wikipedia, and similarity of these items' metadata tags in SALT archives. Here in this web interface you can browse the archive to see the relations in 2D representation while we are also working on building a 3D immersive version for Extended Reality.

As part of this experience, you will be asked to select an initial archive item , and we will display this item along with other archive items that we found most relevant to it. To get more detail about how they are connected you can click on any outer item and you can click on the centered item to get detailed information about it. As you go on you will be able to center more items and explore more.

In RELATIONSCAPE we deeply search for relations between visual items in an archive and visualize these links. Some times it is not easy for an archivist to map hidden relations between archive items. In this tool, we use the combinations of different user and machine generated visual and verbal links between archive items. Users can adjust the similarity scoring algorithm depending on what similarity criteria they value more. We believe that this may help in spotting hidden connections and creating new narratives.

We developed our own similarity metric for visual archives. It combines four sub-metrics that we found interesting and relevant to the archive:

1. Visual Similarity: Visual similarity of archive items' images such as color or layout
2. Object Recognition: Similarity of objects contained in those images
3. Knowledge API: Their external contextual similarity as measured by comparison of text found in knowledge bases such as Wikipedia
4. SALT Data Tags: similarity of these items' metadata tags in SALT archives.

To get more information on how we calculate the scores you can refer to our Similarity Scoring document.

We developed our own similarity metric (MDSSM); it combines four sub-metrics that we found interesting and relevant to the archive: visual similarity of archive items' images, similarity of objects contained in those images, their external contextual similarity as measured by comparison of text found in knowledge bases such as Wikipedia, and similarity of these items' metadata tags in SALT archives. Here in this web interface you can browse the archive to see the relations in 2D representation while we are also working on building a 3D immersive version for Extended Reality.

As part of this experience, you will be asked to select an initial archive item , and we will display this item along with other archive items that we found most relevant to it. To get more detail about how they are connected you can click on any outer item and you can click on the centered item to get detailed information about it. As you go on you will be able to center more items and explore more.

In RELATIONSCAPE we deeply search for relations between visual items in an archive and visualize these links. Some times it is not easy for an archivist to map hidden relations between archive items. In this tool, we use the combinations of different user and machine generated visual and verbal links between archive items. Users can adjust the similarity scoring algorithm depending on what similarity criteria they value more. We believe that this may help in spotting hidden connections and creating new narratives.

We developed our own similarity metric for visual archives. It combines four sub-metrics that we found interesting and relevant to the archive:

1. Visual Similarity: Visual similarity of archive items' images such as color or layout
2. Object Recognition: Similarity of objects contained in those images
3. Knowledge API: Their external contextual similarity as measured by comparison of text found in knowledge bases such as Wikipedia
4. SALT Data Tags: similarity of these items' metadata tags in SALT archives.

To get more information on how we calculate the scores you can refer to our Similarity Scoring document.