Mapping data Flourish: the presence of cats in The Metropolitan Museum of Art’s Open Access Database
In following my cat theme for mapping, I pulled raw data from The Metropolitan Museum of Art’s Open Access database as part of their Creative Commons Zero initiative. Data was then filtered for artworks that were tagged as depicting a “cat.” I wanted to explore what relationships there may be across artists who depicted similar subject matter. Over 50 columns are included in the Met’s collection information: the unique identifier ObjectID could acts as my node unique ID; the artist/maker could be node label; finally, the department to which the object belonged could acts as a grouping function found in Flourish.
Flourish – similar to Gephi – uses Source and Target for links and relationships. My initial import was using department but I realized I needed nodes to be the child records (artists) instead. The color by in Flourish function allows you to select a column for grouping – with this enabled and the grouping using the department column data, my visualization took shape:
To give visual interest, I updated the theme to “Midnight”, changed the outline width, opacity, link coloring and altered the radius sizing:
I tried unsuccessfully to enable images in my popup tool with some basic CSS – I realize on reflection that my network depiction was failing to display the Met’s linked image because I was still using parent/aggregate information (artist) whereas the image data was at a lower granular level (specific artworks.) Source labels, citation, and notes were added so that a user could access the data directly from the visualization.
Still in Flourish, I updated my chart to a radial network graph to see how relationships may look differently:
On this exercise, a good revision would be to expand my data of artworks from the Met and perhaps not limit to a particular subject matter – I think narrowing by tags (known similarities) removed my ability to find new relationships. Focusing on a selection of proficient artists may be better to see how their pieces are distributed into various departments. I also plan to explore credit line information in a larger set (the cat dataset very few donors that gave more than one object.) The dataset type and structure was also more flexible in Flourish but there are limitations in this software for clustering, degrees and weighting. Finally, it’s clear to me that limiting my data set also made presenting a narrative story difficult. A Flourish network or radial network map would be a good inclusion on my final portfolio with an expanded data set when it is better supported by a story.