You may remember my digital project proposal! The semester is winding up and said project has reached its fruition, for better or worse. The glass-half-full takeaway from my digital project is that I learned a lot and gained experience working with an interesting and versatile digital tool, Gephi. On the other hand, for reasons I’ll explain in a minute, I don’t think that the final product is a productive tool for investigating and revealing historical information.

I pulled my data set- the various connections (professional, personal, familial) between various social reformers from the early 20th century- from sources that I have been using to write a lead-up to my thesis on Progressive Era Philadelphia. Keeping track of all the people and organizations found in my research was somewhat confusing to me, and in order to visualize it, I wanted to use a digital tool: in this case, a network visualization.

Here are the results:

Screen Shot 2017-05-01 at 12.09.49 AM
 I went from this…
Screen Shot 2017-05-04 at 9.53.27 PM.png
…to this.

 

The import of the data into Gephi worked beautifully, but I ran into some issues within the parameters of the program. The original goal was to incorporate into the visualization information about each link between entities, which would have made the project useful as a historical research tool. I could have also provided citations for my data there. However, when I tried to do this, I found that the formatting of that text is restricted in Gephi so that it appears like this:

Screen Shot 2017-05-04 at 9.06.10 PM
Close up on some connections between entities; See “corresponds with, 1901” and “founds 1912”

Because the text can only be displayed on a horizontal axis, instead of appearing along the connection line, it is rendered useless. It’s generally not possible to tell which line the text applies to. I was not able to find a workaround for this problem.

As it is, the project looks pretty cool. It illustrates the numerous connections that were present between the major players of Progressive reform in Philadelphia. I was able to color code it to reflect entities that have the most connections, which underscores their relative importance or influence. I also figured out how to highlight each item’s connections when selected, like this:

Screen Shot 2017-05-04 at 9.51.07 PM
In this screenshot, connections to the “College Settlement Association in Philadelphia” are highlighted while others fade into the background

For this reason, I could see this tool would potentially be really useful in exploring a very large dataset from one particular source. But with the mixed dataset that I’m using, and the necessity of annotating connections and providing citations, it does not make the data much more usable, either for the pubic or for researchers.

I anticipate that I might be able to surmount this difficulty if I implemented sigma.js, the javascript library that makes Gephi visualizations interactive within a web environment. Or, had I slightly more time, I think that I might instead employ tools for annotating images to make it more useful from a historical interpretation standpoint.

In some ways I feel like the theme of this semester has been me trying to make sense of my really confusing thesis research. I’ve made a map of settlement locations in Philadelphia, designed a project to chart social networks between the million people and institutions involved in the settlement movement in Philadelphia, and for this crowdsourcing assignment, I decided to dive into the sparsely-populated corner of Wikipedia devoted to the American settlement movement. Not so useful from a public history point of view (at least at this juncture), but what are ya gonna do.

“The seeing eye does not invent the landscape.”
Vida Dutton Scudder

I came across this quotation by social reformer Vida Scudder that I think illustrates the complications of historical interpretation. Reality and truth are a matter of perspective, and what looks red to me might look orange to you. This is a problem that all historians face, but one that really comes to the fore in collaborative ventures like crowdsourcing and Wikipedia.

I’ve had some experience in editing Wikipedia before, and I contributed to the pilot departmental wiki creation project initiated by Gary Scales last semester. In the interest of actually contributing to and expanding the public conversation on this very niche topic in a way that I doubt my actual thesis will achieve, I decided to see what was missing from Wikipedia articles. The key to this was also that I could add the information using source material I already have and which is sitting at arm’s length in my living room.

So, yes, my thesis research has been confusing for several reasons:

  1. My topic is somewhat passe in the history field, so most of the scholarship dates from before the 1980s (and, alas! was mostly published in a bunch of out-of-print books)
  2. There are a million people involved in the American settlement movement, all of whom seem to have corresponded and worked with each other, making it sometimes confusing who was responsible for what and when it happened
  3. Most of the people I’m writing about were women, who left a whole bunch of primary source material behind but whom haven’t been featured in very much secondary research (see also reason #1)

Not surprisingly, these issues have also been challenging for the folks authoring and editing Wikipedia articles on the American settlement movement. There just isn’t very much there. Philadelphia’s settlement movement in particular has almost no presence on Wikipedia with the exception of a section of later eugenicist Katherine B. Davis’ biography page (yikes).

NPG 1746; Octavia Hill by John Singer Sargent
Portrait of Octavia Hill by John Singer Sargent (1898), The National Portrait Gallery, London (used under CC)

The first decision I needed to make was what exactly I should add or edit, considering the sparsity of information currently available on Wikipedia. Most of the women reformers and the institutions I’m looking at do not have their own pages. Even powerful national organizations like the College Settlements Association exist only in brief references on other pages- so I decided to start there. Ultimately planning to create a College Settlements Association page, I edited information relating to its founders and establishment in the articles for Vida Scudder, Octavia Hill, and Denison House of Boston. I don’t have the time to create and fully source a brand new page right now, but now that I know the state of this topic on Wikipedia, I think that this will be an ongoing project.

The lack of secondary research is a serious problem in adding and editing Wikipedia articles for this topic. Because Wikipedia is a tertiary source, they strongly discourage the usage of original research in the creation of articles. I was able to add details like dates and names from some of the monographs I have. But for the most point what I have at this point on these people is original primary research- so it seems like to make Wikipedia happy, I’ll first need to publish my primary research findings in order to create a secondary source that I can then source in an article- but even then, it’s not ideal to base an article on just one secondary source! This is a real barrier to the dissemination of knowledge about less “popular” topics on Wikipedia, as we had previously discussed in class.

On a positive note, though, I was very excited to see that a user thanked me for some of my edits- An auspicious beginning to my Wikipedia experiment! I hope that adding my (sourced) perspective will contribute to a more colorful and nuanced rendering of the “landscape” of this subject online.

Summary
For my final project, I will be working to create a prototype network visualization graph which can be supplemented or altered in conjunction with further research on my thesis topic and utilized for a final digital history component of my MA thesis project. The topic deals with the social connections and interactivity between Philadelphia Progressive Era reformers that allowed them to carry on extra-governmental social reform activity. The graph will visualize connections between locations, individuals, and institutions like colleges, aid organizations, and civic clubs.

Method and Tools

  • I will input the data on key figures, places, and institutions from my primary source material into GraphViz, an open source graph visualization toolset (chosen because I’m familiar and comfortable with this type of data input syntax). This will yield a “relationship diagram” in .dot format. (A tutorial for this method is located here)
  • The visualization software Gephi will allow me to import the .dot file and create an interactive relational web, and tweak nodes and edges as needed to achieve a streamlined and user-friendly graphic visualization.
  • In order to render the graph visible and interactive on the web, however, I will need to take one extra step. After exporting the .gexf file visualization from Gephi, I will import it into Sigmajs, a javascript library for web rendering.
Screen Shot 2017-03-26 at 4.32.28 PM.png
Behold the beauty of sigmajs

Why Digital? Why These Tools?

This visualization software was specifically developed in order to display relational characteristics in a 3D way. In doing so, it provides an alternative way to view data that is difficult to conceptualize otherwise and reveal degrees of “closeness” of relation that might not otherwise be visible. The more connections (“edges”) that each subject (“node”) has to other nodes increases the strength of the relationship. This is a method of analysis for my research that I have not been able to work with so far. I have also never worked with network visualization tools and I look forward to learning something new.

Audience

I’m struggling the most with this aspect of the project. While I strongly desire to create a stellar digital history project that will inform and impact the lives of today’s Philadelphians, I’m not sure how to do that with this relationship diagram within the confines of this project and class. Ideally, I can envision a type of community scavenger hunt that incorporates the sites and people involved, or a six-degrees-of-separation type of interactive game. But the subject matter that I’m currently studying, that of reform in Progressive Era Philadelphia, seems rather difficult to make engaging to the public in this way.

annahowardshaw
Anna Howard Shaw, professional badass, didn’t get her own historical marker until 2014.

I think that rather than use the network visualization to engage the public directly, I might use it to be able to make a case for a few new historical markers in Philadelphia. Since the crux of my argument is that women Progressives have not been acknowledged for their contributions to social welfare in the city, and because there is a huge gender disparity in Pennsylvania historic markers, being able to show the relative importance of the profiled women to Philadelphia history – and erect markers in their memory – would be a fitting outcome.

For this assignment, I decided to map out sites from some of the primary sources that I have been consulting for work on a major research paper about the Settlement Movement in Philadelphia. I’m not very familiar yet with the geography of Philadelphia, and additionally, some street names and neighborhood layouts have change in the intervening century. Visualizing these sites on a map helps me to orient my subject matter- in this case, which neighborhoods were considered most in need of social aid and where the settlements and affiliated organizations were actually located within those areas.

Screen Shot 2017-03-23 at 11.47.18 AM
Screenshot of Google Map “Gilded Age ‘Slums’ of Philadelphia” as of March 23, 2017

I’ve compiled the list of places on a Google Map, entitled “Gilded Age ‘Slums’ of Philadelphia.” It is by no means representative of all of the locations of social aid in the city, but it is helpful to spatially plot some points which may aid in later walks around the area or understandings of neighborhood boundaries. I also plotted ward boundaries for the area of Philadelphia that I’m dealing with as they existed in 1893. From a quick glance at the map, it is apparent that a lot of the settlements that I am currently studying (Namely: the Starr Centre, the College Settlement, and the Octavia Hill Association) look like they were located on the borderlines between wards. Two major studies were undertaken of “slum conditions” in areas considered most in need of humanitarian aid- a federal one in 1894, and one sponsored by the Octavia Hill Association in 1904. I think it’s interesting to note their divergence; there is no overlap between the highlighted areas. I’m not yet sure how to utilize this point, but I do think it gives me a greater insight into the subject as a whole for the purposes of writing my thesis.

In a practical sense, as I’m adding to it as I continue my research, the map is really not an ideal resource for anyone besides myself (or someone else very familiar with the institutions and various slum surveys). I have made an effort to include citation details for the sites, but there is little in the way of other interpretation that would make the map accessible to a non-specialist. For instance, I haven’t provided any contextual information about “1904 Dinwiddie sites (study funded by Octavia Hill Association).” I don’t think that Google Maps is the best tool for building a resource that is both illustrative and interpretive. Were I to develop this into a full-blown project, I would probably utilize either Fusion Tables or some other data tables application to pull out the data (coordinates, etc.) and embed it into a site where I could incorporate floating frames to do the interpretive heavy lifting.

 

 

For my data visualization, I made the chart I wish I’d had last semester while studying the 1918 Spanish Flu Epidemic in Philadelphia. This chart visualizes mortality rates in different wards of the city of the leading causes of death in the year 1918. This makes it easy to see that although pneumonia, for example, was a major cause of death, particularly in poorer wards, its impact hardly compares with deaths from epidemic influenza.
Screen Shot 2017-03-10 at 5.40.51 PM
A screenshot of the final, interactive HTML file

I chose to create the graph by creating an HTML file that calls up the Google Charts API for formatting. I did this for several reasons. First, my data is preserved in an easily accessible code-based source (anyone can click “view source” and see what numbers I’ve entered), in a stable format that works with all web browsers. I recently had a conversation with someone who remarked that visualization tools are great, but they need to be reliable in order to be really useful; if the tool breaks or the web changes and the tool no longer has development support, it’s useless. Standard CSS and HTML are reliable old standbys for at least skeletal web development, because they are still a standard markup language after more than 20 years in existence and are likely to remain so for a while longer. This translates into project longevity.

Unfortunately, it also means that I wasn’t able to embed the chart into this WordPress blog; I can’t place iframes or even upload the HTML source file for blog readers to download and open through their web browser. Even though it’s interactive (in a very basic way), I had to post a screenshot to make it visible- certainly a downside that I did not anticipate.

I got the source material from a primary document, The Annual Report of the Bureau of Health of the City of Philadelphia for 1918, which I first found as an excerpt (that the University of Michigan rather disingenuously didn’t reveal wasn’t a complete document), but then found and was able to download in its entirety on HathiTrust, courtesy of Princeton University. I “curated” the data (read: limited the amount of superfluous data entry) by making the conscious choice to only include the top four causes of death in my data. Although I recognize that that decision to do this is not ideal in terms of transparency about the mortality rates, I also thought that to include all 99 causes of death in the chart would be too visually confusing to the viewer.

 

Review- The National Jukebox: Historical Recordings from the Library of Congress

National Jukebox: Historical Recordings from the Library of Congress. https://www.loc.gov/jukebox/. Created and maintained by the Packard Campus for Audio Visual Conservation, Library of Congress, and Sony Music Entertainment, Inc., Washington, D.C., https://www.loc.gov/jukebox/about. Reviewed Feb. 2017.

About a year ago, I found myself trying to make room for a family member’s spring-cleaning collection of old vinyl in an already-full record cabinet. Caught between a desire to keep these cool old records- for posterity’s sake!- and a pragmatic understanding that I would never, ever, throw on a 2.5 minute Dee Dee Sharp single, I eventually just left the record crate in front of the cabinet, where it has lived ever since.

The National Jukebox gathers comparable cobwebs in a similar scenario, tucked into a corner of the Library of Congress website. The project (clearly ambitious) took a lot of money, time, and effort to bring to fruition, and it has great potential to serve as a valuable educational resource. Unfortunately the seemingly-abandoned site doesn’t appear destined to achieve this legacy as things stand.
(more…)