Beyond the Women's Building

Women Architects and World Fairs

Page 2 of 8

Network Analysis and Spatial Connections: A Digital Love Letter to the ORBIS Project from Admin

1 November 2026 This week’s classes were centered on networks and network analysis–a tool I have only recently become familiar with but have found myself enjoying immensely. True to my classical art and archaeology roots, I was most taken by Elijah Meeks and Karl Grossner’s Modeling Networks and Scholarship with ORBIS (2012) . The article […]

Connecting, Connection, Connected: Networks as Data Visualization

            At the most basic level, networks are simple to understand. If there is one connection drawn between two heterogenous data points, you have a network! Expanding this to include more data points and connections complicates this basic graph, but it also produces a useful visualization of data that can help viewers understand connections between different data points. Before diving into this subject and my own class’s experiment making a network, it is necessary to define a few terms with the help of Scott Weingart’s post “Demystifying Networks.”[1] I will also use the Dancing with the Stars network that our Digital Art History class collaboratively created with the platform Kumu as an example.

Figure 1: Image from Weingart’s “Demystifying Networks” representing a basic network.

Weingart defines networks plainly as an interlocking system of stuff and relationships, imagined in Figure 1. This “stuff” can be divided into “types” and “nodes” or “vertices.” A “type” describes the nature of the node, while the node itself is the most individual element of the network that will be fit into this web of connections. Using our Dancing with the Stars example, the individual nodes are the participants of each season. The types are the categories that the participants fall into, like “Contestant,” “Pro,” or “Judge.” These types are divided even further for the Contestant into the type of celebrity they are, like “Professional Athlete,” “Musician,” or “TV Host.” By the end of class, we also discussed the possibility of drawing further connections between nodes by indicating relationships. For example, more avid consumers of Dancing with the Stars noted that some contestants are married or in romantic relationships with each other or that some are siblings. The product of our test network resulted in a web of connections.

This “web of connections” is visually challenging to look at, a problem that Houda Lamgaddam, et al. examine in “Perceptual Effects of Hierarchy in Art Historical Social Networks.”[1] In this article, Lamgaddam et al. explain that network viewers prefer hierarchical rather than “force-directed” networks. The article concludes that hierarchical network graphs reduce the “cognitive load” on the user. I could see these forces at work within my own experience with Benjamin Zweig’s “A Universe of Artists” which he created based on the available Wikidata of artists. Zweig’s project is impressive in its scope. It names all artists with Wikipedia pages and connects them to other artists who are linked in their page.

            While we did not input nearly as much data into our shared Dancing with the Stars Kumu network, the force-directed display creates a similar “cognitive overload” to Zweig’s “A Universe of Artists.” Because the data set for the Dancing with the Stars network is significantly smaller, the user can still see the individual nodes. Looking at Zweig’s “Universe of Artists” is much more overwhelming with this massive dataset, causing it to look like an indistinguishable mass of data.

Figure 2: An image of the Kumu network Dancing with the Stars, created collaboratively in the Alt-Methods: Digital Art History class on October 29, 2026.

Figure 3: Benjamin Zweig’s A Universe of Artists. https://bzweig633.github.io/universe-of-artists/.

The hierarchical network instead organizes nodes and connections, also known as “edges,” like a family tree so that connections can be seen linearly. This hierarchical network, however, may be better suited for some data sets more effectively than others. Hierarchical networks may make the most sense when there is a downward or upward relationship between the nodes rather than multiple elements connecting different nodes. In a family tree, connections flow in one direction, from parent to child. In our Dancing with the Stars example, we made connections of nodes based on their status as a contestant, professional, or judge. We also thought it would be interesting to chart connections between individual nodes, like marriages or relationships between participants. This would create another element to complicate the network and make a hierarchical network using this dataset more difficult to accomplish.

Placing this method of data visualization within the greater themes of this semester, the creation of networks may be the most direct method of data visualization. Creating networks is a form of macroanalysis, that, by its nature, will necessarily draw the researcher’s or user’s attention to the pattern created in the data through the connections of nodes to elements.

Timelines visualize time linearly, limiting the variables that can be used. Throughout this past week, I have worked on a timeline related to a term paper that I am writing about the architect Owen Jones. You can find a link to view “The Education, Life, Work, and Travels of Owen Jones” below. This timeline charts the major life events of Jones including his education, significant travels, and most influential publications.

Link to “The Education, Life, Work, and Travels of Owen Jones”: https://cdn.knightlab.com/libs/timeline3/latest/embed/index.html?source=v2%3A2PACX-1vSXAwJ_aNWmPgGYUohiWkg7vZvGM5DIyioO1DJBr1Xq_jUPfowFH70tpA3vI8K2TjSNCw92S_HnBmiV&font=Default&lang=en&initial_zoom=2&width=100%25&height=650


[1] Houda Lamgaddam, et al., “Perceptual Effects of Hierarchy in Art Historical Social Networks,” Digital Humanities Quarterly vol. 16 no. 1 (2022).


[1] Scott Weingart, “Demystifying Networks,” Scottbot.net, December 14, 2011, https://www.scottbot.net/HIAL/index.html@p=6279.html.

Timelines in Digital Humanities

            A few weeks ago, I ventured into spatial mapping using Google’s custom maps. This week, I tried out mapping across space by creating a timeline in Timeline JS. The program is simple to use with a pre-formatted Google sheet where the user can simply plug-in the information he or she would like to plot on their timeline. Users can add a variety of media to the timeline including image and videos in the form of jpg or mp3 link addresses. Unfortunately, this means that downloaded media cannot be attached, at least in this basic version of Timeline JS.

            My first experiment with this digital tool is very brief, although a more extensive timeline is coming soon! For this first experiment, I included a few works of art featured in the Philadelphia Art Museum as part of the exhibition A Nation of Artists: 250 Years of American Art. The pre-formatted Google Sheet is user-friendly, but Timeline JS’s inability to use downloaded images was a challenge since the Philadelphia Museum of Art does not provide an image address that can be easily copied. Instead, I used a reverse Google image search to find image addresses of the works included on a different website. The result is a four-object timeline plotting four works of art in the exhibition accompanied with an image and a brief description. You can check out this preliminary timeline below.

https://cdn.knightlab.com/libs/timeline3/latest/embed/index.html?source=v2%3A2PACX-1vS0ju4fuuXWckwIizWQE6qLyMbDz5Lwl0CEjF2UMFUlFfiRrWlbxmu_9aiRF1IIWqwwwj87DoiiidSg&font=Default&lang=en&initial_zoom=2&width=100%25&height=650

Timelines in Digital Humanities

            For class this week, I also explored other digital humanities projects’ that used timelines. The timelines in these projects suited each purpose well, whether the objective was to describe a specific historical figure or event for close-reading or for creating an interactive browsing interface.

            “Mina Loy: Navigating the Avant-Garde” was a project led by scholars in English literature from Davidson College, Duquesne University, and the University of Georgia. Spanning several years with the help of a grant from the National Endowment for the Humanities, the project seeks to examine the life and work of Mina Loy, an influential figure in the avant-garde movement. The timeline specifically is successful for explaining Loy’s life and work by separating it into phases based on places that she lived. From this project, I can see how using a timeline is helpful to visualize an historical figure’s life, especially for one who moved to different places throughout his or her lifetime.

The project integrates a variety of media throughout the interactive project, including images and audio. One shortfall of this project is broken or inactive links. I found myself encountering them throughout the timeline feature when an image could not be properly displayed. This reality indicates how digital humanities projects require a sustainability plan and constant maintenance to continue to function beyond the construction of the project.

            The project leaders, Suzanne Churchill, Linda Kinnahan, and Susan Rosenbaum, also published a reflection about the project on the platform Visualizing Objects, Places, and Spaces: A Digital Project Handbook.[1] In this reflection they described the process of collaboration across institutions and disciplines, the workflow timeline, and funding from the NEH and their respective institutions. In the section “Challenges & Opportunities,” they emphasized the need to keep the project focused and to limit—to a degree—the extent of the project or a particular aspect of the project to avoid spending too much time and energy on a part that may not be necessary. Overall, the project aligns well with other blog posts and articles that we have read throughout the semester that outline the main tenets of creating a digital humanities project. The Mina Loy project, like others, required collaboration, realistic objectives, and clear strategies for achieving those objectives while also allowing flexibility and space for discovery and innovation.

            Some other examples of timelines used in a digital context include “Mapping Austro-fascism” from Campus Medius, the Metropolitan’s “Heilbrunn Timeline of Art History,” and Google Arts & Culture’s Time Explorer.

“Mapping Austrofascism” takes a close-looking approach to a specific weekend in May 1933 the Simon Ganahl, the project leader, posits was highly influential in spreading fascism in Austria. This project combines spatial mapping with a timeline, with the timeline also indicating the duration of the events. Examining each event throughout this weekend in Austria paired with text and media clarifies and paints the user a detailed picture of the context required to understand the significance of each even throughout this brief moment in history.

A screenshot of “Mapping Austrofascism” with a view of the timeline, map, and additional information and media supporting the explanation of this event.

While “Mapping Austrofascism” and “Mina Loy: Navigating the Avant-Garde” involves extremely close looking, the Metropolitan’s “Heilbrunn Timeline of Art History,” and Google Arts & Culture’s Time Explorer is better suited for the casual browser. The Metropolitan’s “Heilbrunn Timeline of Art History” is separated thematically and geographically and allows the user to browse the Met’s digital collections. Each timeline is accompanied with an essay describing the theme of the timeline, and the timeline also indicates influential events that occurred in the region. The Google Arts & Culture’s Time Explorer is even more general, spanning across geographies and digital collections to create a more comprehensive timeline complete with images, a link to the holding institution, and details about specific works of art. With Google’s Timeline, I am curious to know how works across digital collections are chosen. I assume that the images presented are meant to be for public use.

Throughout this week, I discovered how timeline are both helpful for browsing and for close-looking. Timelines can help track and divide moments throughout an historical figure’s life, like that of Mina Loy, or the specific events over the course of a day, like the “Mapping Austro-fascism” project. On the other hand, timelines can also be fun and interactive tools provided by institutions to give the user another creative way to interact and browse through their collections. Timelines, like maps and other forms of data visualization, are helpful to visualize time while also increasing user engagement with a digital humanities project.


[1] Suzanne Churchill, Linda Kinnahan, and Susan Rosenbaum, “Mina Loy: Navigating the Avant-Garde,” Visualizing Objects, Places, and Spaces: A Digital Project Handbook (PubPub, January 14, 2021), https://handbook.pubpub.org/about.

Time got away from me in more ways than one this week

So I may have gone a little overboard this week.

Inspired by exploring Mina Loy: Navigating the Avant-Garde (https://mina-loy.com/) and given the assignment to make a timeline using TimelineJS (original recipe or reimaged), I decided to fill a gap in my own scholarship. I was impressed by how Mina Loy: Navigating the Avant-Garde went beyond a timeline to incorporate multiple projects from visualizations to games to capture different aspects of Mina Loy’s life and artistic work. While I was only making a timeline, I also wanted it to fill a gap in my own scholarship while painting a complex picture of the topic.

A missing area in video game scholarship that has proved to be a major problem for my study of artist book and fake game guide Vermis I: Lost Dungeons and Forbidden Woods has been the lack of a even somewhat cohesive history of the game guide format. While there are good sources of information on that history, they are all partial focusing on only a few aspects of the topic like publisher history or magazines. I decided what was needed was a fuller timeline with overarching game guide history in the form of news and events, specific notable game guides and publisher history, console launch dates to add context to what games would have been written about at a given period, and release dates for games mentioned. I also decided from my research that a game guide history would have to cast a wide net, including not just standard printed game strategy guides but other areas where guides have been published like magazines, zines, video, and websites.

Did I do too much? Absolutely. The timeline is a mess of dates on top of each other with very little that is readable.

Do I regret it? Not really. My timeline ended up being a great example of digital humanities as process instead of product. I learned an immense amount creating the timeline about how game guides changed over time. I also picked up on interesting trends in other areas like the consoles and games release windows getting shorter as more saw global or near global releases over time while past releases were more piecemeal and slower. In the course of my research, I found game guides and information that did not even make it into the timeline that is relevant to my thesis on Vermis I.

It was also a learning experience for using the new TimelineJS reimagined (https://knightlab.northwestern.edu/2026/08/13/timelinejs-reimagined/index.html). Overall, I found being able to see my slides as I added and edited to be much more intuitive than creating a spreadsheet in envisioning how my timeline would flow. This may have also contributed to my slide bloat (from my assigned 20 slides to the final 28 slides) since the process was creating each slide individually rather than thinking of them as a whole. Grouping was easy and essential though I wish there was an option to mass change the name of groups. The date range functionally was also lacking with no way to show that a slide was ongoing into the present. Perhaps this is for the best when a timeline is not stamped with a date of creation, but it made it frustrating to see items that should show as a range showing as single point because they had no end date. These may be issues that I could have fixed had I experimented more deeply, but on the surface, there did not appear to be a way to address these pain points. The uploads also began to chug as I neared editing slide 70 which is understandable as I was also feeling overloaded at that point. Still, these are all minor issues in what was otherwise a dangerously easy process.

Despite the information density, I do believe the timeline still has some great information. I do recommend full screen for reader’s eyes and sanity though.

Timelines and Micronarratives

While broader data visualizations and maps are helpful for getting an overview of a topic, timelines are a great way to get into the minutiae. When the scope of a project is much more focused, such as with the life of an artist or a single day in Vienna, organizing information chronologically tends to be a more compelling way to communicate an idea. This week, we looked at some projects that took different approaches to the deep dive with varying levels of success.

The most expansive project, “Mina Loy: Navigating the Avant-Garde,” was truly a bit shocking to me in its scope. Aside from scholarly essay collections, it contains multiple exhibits, timelines, games, a participatory artwork, and 20 student projects under a tab called “New Frequencies.” Reading through the article written on the process of creating the site, it becomes clear why it has been constructed that way. Most importantly, a number of grants were acquired to fund this three-year project. This allowed faculty researchers to create the bones of the site, which they then used as a pedagogical tool to allow students to explore the life and work of Mina Loy in a more creative way. As a result, there are a lot of hands that touched this project. To me, this was kind of a strength and a weakness at the same time. Having more creative reinterpretations playing on the lines of an artist’s creative philosophies is engaging (and I’m sure was fun to put together), but at the same time, the way the site is organized is somewhat meandering at times. It seems like they allowed the project to grow more organically, which does fit with the theme of the manifesto on the site; ultimately, it is successful in its goal of “challeng[ing] you to read outside the norms of scholarly writing.” I very much enjoyed the “art exhibits” section, which uses works created by Mina Loy as “maps” with annotations at specific points in the image. I think this is such a great communication tool (and use for StorymapJS) for art historical research that not only walks viewers through the intricacies of the work itself, but how to even do a close reading of an image in the first place.

A screenshot of the Storymap for Mina Loy’s 1959 work Prospector 2.

While some of the organization of the second biggest project we looked at (Mapping Austrofascism) was also a bit convoluted to me, I really liked how well the linear storytelling method used fits the thesis of the project. Essentially, a play-by-play of a 24 hour period (May 13-14, 1933) tells the story of “the Nazi ‘seizure of power’ in Germany and the Austrian response to these developments.” It is clear that there was a lot of thought put into every choice about how to represent this critical turning point; project lead Simon Ganahl writes about the influence of the chronotrope as a way to capture an essence of modernity. In my experience of navigating the project, it adds an urgency and contextualization to the description of events that I hadn’t seen in a project like this before. I was also very inspired by how in-depth the descriptions associated with each of the fifteen data points were (especially in the topology section), including images and videos. I will admit, I’m not super sure what the purpose of splitting it into topology and topography is, but if I just consider that it is two different ways of looking at the same information, it makes a little more sense. Overall, I thought this project was really cool and I definitely was inspired by its compelling and clear narrative!

An expanded selection from the 24 hour timeline on Mapping Austrofascism.

I’ll also briefly touch on some of the less successful timelines we looked at in class…namely, the Time Explorer from Google Arts and Culture. I think in principle it’s something that could be functional and even potentially open up new paths for scholarship by allowing holdings from international museums to be placed next to one another, but as it is right now, it’s just a cautionary tale in metadata standardization. Which is unfortunate, because I feel like people who aren’t necessarily deep in history truly conceptualize what global events were running concurrently: there’s the classic example of samurai and cowboys existing at the same point in time in the mid/late 19th century. A tool like this has the potential to globalize art history in a thoughtful way, but clearly, no one is cleaning up the metadata here. Hokusai’s The Great Wave off Kanagawa is indexed for every museum that has a print of it, which is at least a dozen. Perhaps the most unfortunate of the museums featured is the National Folk Museum of Korea, whose undated objects show up in every single time period on this scale (and some more often than others, for some reason). I think this is such a disservice to these artifacts, and it even makes them into a bit of a nuisance to scroll through since they are so incongruous with every list that they’re included in. Archives and museums have been dreaming of an aggregative database for ages, and while this is a step in the right direction, it looks like we’re as far as we’ve ever been from realizing that dream.

This centrifuge from the National Folk Museum of Korea makes a lot of appearances. It is shown here in the section attributed to the 1500s.

After digesting all of these ways of expressing stories temporally, we got to practice it ourselves. To be quite honest, the only thing I could think of that I could get a full 20 data points out of was the story of Patience Worth. If you looked through my CollectionBuilder site on séances I touch briefly on Patience, but she’s got an interesting story worth digging into. A St.Louis woman named Pearl Curran claims to be in contact with the spirit of an Englishwoman who died in the mid-17th century named Patience Worth. This spirit is not being used for predicting the future, reading fortunes, contacting the dead, or any of the theatrics associated with Spiritualism; instead, Patience Worth wants to write and to publish, and that’s all. I encountered Patience’s work for the first time at my museum job. We were cataloging a series of 50+ diaries kept by a local woman; this woman loved poetry, and had clipped several of Patience’s poems from Reedy’s Mirror and placed them in her diary. Their highly stylized nature made me look twice at the publishing date, then immediately go to look up who this faux old English poet was. I was immediately hooked by how odd this story is. I feel like this is also a story that lends itself particularly well to being told in a linear fashion, because there was a lot of push and pull between Pearl Curran and her critics; they very directly influenced each other. (Here is a direct link in case the embed is being strange.)

I definitely have things I would do differently next time (especially aesthetically, since I spent more time on the content than the look), but overall using the new version of TimelineJS was so easy. After fighting with the Google sheets version, I caved and made an account so I could upload my own images (mainly newspaper clippings!) which made the process a breeze. I particularly appreciated the much more elegant way the new version scales the timeline to your dates.

Narrative, deception, privacy, and purpose in situated data visualization

Where do we draw the line in data visualization between what is storytelling and what is deception? According to data visualization author and expert Nathan Yau, the answer is to question the context of the creation of the chart. Who made it? What is it trying to say? Does the data actually support that narrative or does the visual have to be skewed to make it suggest the desired narrative?1 Ultimately the suggestions for assessing the honesty of a data visualization boils down to maintaining the “essence of the data” in the visualization.2 How to determine what that essence is requires different tactics depending on if you are making the visualization or trying to judge the honesty of a visualization as a viewer.

Nathan Yau suggests a few aspects to consider when creating or assessing a visualization: context, acknowledging all visualizations have a narrative or purpose, and remembering that people are the ones who create visualizations.3 Scholars Catherine D’Ignazio and Lauren Klein also address these points in “Chapter 3: On Rational, Scientific, Objective Viewpoints from Mythical, Imaginary, Impossible Standpoints” in their book Data Feminism. By charting how data visualization has historically focused on an objectiveness goal that is fundamentally impossible, they show how the humanity of creators shows in all aspects of data handling. This is regardless of if a visualization is deliberately trying to persuade or not. Every person brings themselves to the collection and interpretation of data including the viewer.4 Because we are human, when we model data we should also consider emotion as a valid data point or component of communication.5 D’Ignazio and Klein state that this isn’t a binary choice between an attempt at objectivity and “a retreat into complete relativism and a world in which alternative facts reign and everyone gets a gold medal for having an opinion.”6 They suggest instead a number of theories that all return to noticing that people make choices with data to communicate.

D’Ignazio and Klien put the human context primarily on the creator, and suggest situating as a method of adding human context. They position themselves in the introduction to Data Feminism by noting “the gender, race/ethnicity, class, ability, education, and other subject positions” they hold..7 While this is not their only suggestion, I am addressing it because it has great potential to do harm to the very people at the margins that feminist approaches try to center. Situating oneself along demographic lines and encouraging that in others ignores that these types of disclosures might be felt as an invasion of privacy or create real danger. It reminds me of well meaning fans of a television show that forced an actor to publicly disclose his sexuality to keep his job (I am not naming the show or actor deliberately). While perhaps well intended, the result was still that they forced a young person to not only label a sexuality they were still exploring but to do so in a public statement. Not only is that harmful to the very community it is trying to protect but being situated in a specific identity does not always mean you are the best person to speak on that identity. I have seen an aspect of my own identity presented by several writers both those that hold the identity and those that don’t. Of my three favorites, one shares the identity, one does not, and one is an unknown.

No approach is without issues, but I much prefer Yau’s which puts faith in creators wanting to be honest8 while also putting judging that honesty on the viewer.9 Yau couples this approach with the idea that context is valuable but can be both explicit or implicit.10 I think this approach also shows in Yau’s about page. Yau situates himself according to who he is as a person not a demographic list and appears to balance details that help the reader understand where his opinions come from without disclosing more than he wants to as an introvert. I believe this balance is a deliberate choice since his about page ends with “My work is based on these experiences.”11 which makes it clear that Yau is aware of both the need to situate himself for the reader and how to do so in a way that is contextually relevant and within the bounds of his own choices.

Another aspect of data visualization that came to mind during these readings was how the approach to honest situated data visualization changes when the purpose is for internal research instead of external display. To experiment with that aspect, I ported a small dataset from my Information Science masters paper12 into Palladio. The data used was a simplified version of my analysis of the age of video game related materials in nine academic art libraries. All three were organized by library. My original spreadsheet also included the titles of the earliest and most recent materials and details about their content. Since the spreadsheet was not created to be parsed by a service like Palladio, I decided to remove those data points for this exercise.

Two timeline bar charts and one timespan line graph. The two bar charts show the earliest and latest year of a work in a library's collection. The timespan shows the range from earliest to latest. It is difficult to distinguish detailed information from the overly simplistic visuals.
Three visualizations from Palladio based on a sliver of my Information Science masters thesis.

I’m mixed on the result. When viewed as potential ways to display my data, I feel I made a smart choice when I decided to avoid visualizations in my paper, a choice that is as editorial as the choices made when making a data visualization. I felt that including data visualizations would risk privileging the numbers when the nuances of conducting the study added invaluable context to the numbers. My spreadsheets were in the paper as support for my conclusions, not as a way to communicate anything about the landscape of video games in art libraries on their own. As an analytical tool with this dataset, I think there is a major risk of accidental self-deception. There was no way I could find to change the parameters of either the timespan or the timelines. Because the latest year x axis is divided into months and the earliest year axis is in year increments, it appears like the earliest materials and the most recent materials cover an equal range of time. In truth, the span between the latest dates is only 3 years. Meanwhile, the earliest dates range over decades (another shortcoming of the earliest chart is it’s lack of division between years making it impossible to discern exactly what the earliest date for each library is).

The timespan filter was more promising. It showed a comparatively massive timespan of materials at the University of North Carolina Sloane Art Library. When I wrote my paper, I also noted this and it lead me to more closely examine that earliest item. By taking a closer look at the types of materials that were the earliest in each collection, I discovered that there was much more variety in earliest material than in latest. While a tool like Palladio was unnecessary for this discovery, I imagine that with a much larger dataset, the timespan view would be a good tool for catching this sort of information that might other wise be overlooked.


  1. Nathan Yau, “Defense Against Dishonest Charts,” FlowingData, February 13, 2025, https://flowingdata.com/projects/dishonest-charts/.↩︎

  2. Nathan Yau, Data Points: Visualization That Means Something (John Wiley & Sons, Inc, 2013), 98.↩︎

  3. Yau, “Defense Against Dishonest Charts”; Yau, Data Points.↩︎

  4. Catherine D’Ignazio and Lauren F. Klein, “Chapter 3: On Rational, Scientific, Objective Viewpoints from Mythical, Imaginary, Impossible Standpoints,” in Data Feminism, Strong Ideas, ed. David Weinberger (MIT Press, 2020), 8, https://mitpress.mit.edu/9780262044004/data-feminism/.↩︎

  5. D’Ignazio and Klein, “Chapter 3: On Rational, Scientific, Objective Viewpoints from Mythical, Imaginary, Impossible Standpoints,” 16.↩︎

  6. Yau, Data Points, 15.↩︎

  7. D’Ignazio and Klein, “Chapter 3: On Rational, Scientific, Objective Viewpoints from Mythical, Imaginary, Impossible Standpoints,” 16.↩︎

  8. Yau, Data Points.↩︎

  9. Yau, “Defense Against Dishonest Charts.”↩︎

  10. Yau, Data Points, 112–16.↩︎

  11. Nathan Yau, “About,” FlowingData, June 13, 2007, https://flowingdata.com/about/.↩︎

  12. Mert (Meredith) Overcash, “”I’VE EXPERIMENTS TO RUN, THERE IS RESEARCH TO BE DONE”: VIDEO GAME ART WITHIN ACADEMIC ART LIBRARY COLLECTIONS” (The University of North Carolina at Chapel Hill University Libraries, 2026), https://doi.org/10.17615/MRFG-SD27.↩︎

Annotating, Captioning, Translating

Oct. 15, 2026

In “Annotating Upstream: Digital Scholars, Art History, and the Interoperable Image,” M.J. Westerby discusses the role of International Image Interoperability Framework (IIIF), an open-source image metadata system, in art historical scholarship. At the top, he makes the stakes of this technology clear: when the British Library’s databases were hacked in 2023, their data which had been shared with IIIF was broken and led to objects becoming temporarily unavailable for scholars and the public to access on multiple platforms (not just the Library’s). While IIIF has some major benefits, like how it allows “images and descriptive metadata [to] be reused and decentralized,”1 Westerby notes that there is danger in having only one set of “upstream” metadata for any given object. Basically, Westerby argues that these institutionally-generated identifiers have a not-insignificant amount of power over how an object will later be described. Indeed, these “upstream” descriptions can shape the very narratives we have about a work, or can change the way we teach objects.

I completely agree—at the graduate level, I try to have a diverse and robust amount of sources to understand objects I write about, but as a contemporary art historian I find that there are often fewer descriptive sources than I’d like—leaving me to rely on metadata generated by only one or two galleries or museums. Similarly, I’m a Teaching Assistant for a survey of a (so-called) western art history right now, and it’s clear on the back-end how important it is to have IIIF and other institutionally-driven open-source data in order to create curriculum for and teach those kinds of classes. Westerby continues by arguing that scholars shouldn’t have to rely on sparse IIIF metadata and annotations about their images. They say, “art historians and digital scholars should produce IIIF resources and not rely only on institutional data services,” and later that “Art historians and digital scholars can take ownership of their insights and storytelling outputs by practicing upstream annotation.”2 An example of one such image description project is shown below.

Screenshot showing colored annotations in Digital Mappa 2.0 linking words describing elements of a painting to a IIIF canvas with an image of Orazio Gentileschi, The Lute Player, c. 1612/1620, National Gallery of Art, 1962.8.1. Image: M.J. Westerby
Screenshot showing colored annotations in Digital Mappa 2.0 linking words describing elements of a painting to a IIIF canvas with an image of Orazio Gentileschi, The Lute Player, c. 1612/1620, National Gallery of Art, 1962.8.1. Image: M.J. Westerby

I think that art historians could practice engaging in these captioning projects as a creative practice, experimenting with different ways to describe objects past their date and medium. In this course, I’ve made my own Collection Builder site about surveillance art. When I was approaching the “description” part of the metadata, I wanted to focus on situating works in their art historical and technological context, rather than only saying what the object is. Description, translation, and annotation are all important creative mediums in and of themselves. Nonetheless, my personal take on Westerby’s proposal is that I do not have the time or real motivation to generate my own independently sourced IIIF data (or an equivalent).

It takes significant time, and the reason why museum and library institutions are the centers of these data sources is because they can and do pay people to work on IIIF and other digital archiving projects for their jobs. What he proposes sounds a lot like something I’ve done for different internships, not all of them paid unfortunately, but all of which I put on my CV, which is at least some material benefit. This feels like a proposal geared toward digitally-inclined art historians who have the time and resources to spend working on a fully independent metadata project—which isn’t any of the grad students I know!  

Though our class didn’t engage in image annotation or metadata this week, we considered another type of data annotation: audio transcription. Using Panopto, I captioned a video I had made for the Department of Art and Art History’s instagram, where I asked different professors to give a tour of the bookshelf in their office. In part, I thought this would be an interesting case study because I had already captioned the video by hand and I wanted to see how accurate Panopto was in comparison. There were a number of art historical terms that it had trouble picking up on, but for the most part it was accurate and certainly far quicker than doing it myself. I went on to use the tool for my own research. First, I had Panopto caption an hour-long talk between the contemporary artist I’m writing my thesis on, and an important scholar who’s curated their work. Though YouTube did have captions on this video, being able to pull quotes directly from the Panopto transcript is extremely useful. Second, I used Panopto to transcribe an interview I conducted with the artist I’m writing my thesis about, who I met in their New York studio about a week ago. Aside from the clear benefits of having this interview in the form of a written document, it also saved me the deep discomfort of listening to myself talk for an hour (except for when I had to fix errors in the transcription). 

I like the idea of getting creative with captions too, though that doesn’t serve my research goals right now. A number of artists (especially video artists) have treated captions as a creative medium.

  1. Westerby, M. J., (2024) “Annotating Upstream: Digital Scholars, Art History, and the Interoperable Image”, Open Library of Humanities 10(2). doi: https://doi.org/10.16995/olh.17217
    ↩︎
  2. ↩︎

Boring Bar Graphs and Mundane Maps: Bringing Emotion Back Into Data Visualization

18 October 2026 ***CW: this post contains mentions of gun violence in the United States*** The last two weeks we have focused on data visualization in class, especially visualization beyond a simple map or graph. While maps and graphs are fantastic ways to convey information, they can sometimes exist within an emotional vacuum. Simple bar […]
« Older posts Newer posts »

© 2026 Beyond the Women's Building

Theme by Anders Noren — Up ↑