DH 2015 Sydney notes – Friday
This was an excellent conference. Many thanks to the organizers and to the University of Western Sydney for their hospitality. It’s a beautiful campus and the crystal clear weather didn’t hurt.
Long Paper Session
“Everything on Paper Will Be Used Against Me”: Quantifying Kissinger
Micki Kaufman – CUNY
Used a quote from John Ehrlichman about the record on Nixon as motivation for the work. Ehrlichman said, basically, that historians needed to listen to all of the Nixon tapes to form a complete picture.
Source material for this work comes from the Digital National Security Archive, a Chadwyck/ProQuest product. Scraping the 18,000 metadata records led to a cease and desist letter from ProQuest. There are now multiple gigabytes of Kissinger material available from his time as national security advisor.
Fortunately, the corpus used has been processed by archivists, as she noted. This means that metadata has improved on the documents, adding to and from notes, dates, etc. It’s possible to see parallels between the human-assigned topics and those generated by a Mallet topic modelling process. She’s looking for temporal elements: does correspondence mirror or reflect in some way the historical events of the time. Do topics shift and change as events unfold?
Showed a “longue duree” view of topics from 1969-1977, and stark density shifts become apparent, largely driven by Kissinger’s role changes (to Secretary of State, where he had to work on the record more). This distant reading exercise shows these shifts in clear detail.
Showed Gephi visualizations and spoke about initial work doing three-dimensional analysis. Also interested in word collocation, for example around the word bombing (subjective decision based on Kissinger’s views and practices). First attempt to visualize this was unsuccessful (x-y scatterplot), creating a fairly useless graphic. Another visualization can successfully show that the word appears in different contexts in the memoranda (memcons) and telephone conversations (telcons). Looking more closely, one sees that certain words relate to illegal or legal bombing.
Spoke a bit about the different needs of historians and linguists when looking at a corpus. Given how the historical documents are created (where conversations happen in stages and clusters), larger gaps in collocation are interesting, while linguists would find such gaps less useful.
Improving Compliance with Evolving Standards Using Computed Transformation of Digital Collections
Peter Cornwell – U of Westminster
Joint project between Westminster, Lyon, and Heidelberg around contemporary China collections, as well as various resources elsewhere. These are both public and controlled-access collections.
Showed a variety of these projects, focusing on the Chinese poster collection at Westminster, noting the standards and tools used to create them, as well as the lack of standards. The poster collection had two elements, created in 2001 and 2009 using different tools. The collections have rich metadata, but much of it predated some of the current description schemes. Showed how they took these two collections to create a more standards-compliant system.
Short Paper Session
Modeling Approaches to Library-led DH Pedagogy
Thomas George Padilla, Bobby Smiley, Sara Miller, Hailey Mooney – Michigan State U
Not all libraries have the resources of the Stanfords, to name one example, but all libraries do all tend to teach. As such, teaching is often the “low-hanging fruit” of engagement in the digital humanities space. Showed a variety of job ads that underscore this, even though, as he noted, LIS education doesn’t provide the skills being sought.
Showed outcome of a word occurrence search with a journal issue related to DH in libraries where service, support, and work are the dominant terms. Finds that troublesome. There are challenges:
- content – “teach me Gephi” isn’t an easy demand to meet; there are methods and context one must know
Worked with ACH to crosswalk the ACRL information literacy framework (IL) with other frameworks more closely related to DH work (e.g.- Data and Information Literacy – IMLS-funded effort). DIL narrows down many of the broad categories found in the ACRL framework. Found other frameworks coming from disciplines, e.g.- AHA. Goal is to knit together the various expectations and frameworks to arrive at grounded principles for instruction. Showed specific examples.
One fused IL and DIL with disciplinary expectations (DE). The IL frame was “information creation as a process.” In DIL, this is about data quality and documentation, databases and data formats, and data conversion and interoperability. This leads to an outcome that covers the various elements and makes a sensible whole. Showed another example using the IL notion of “scholarship is a conversation” and the DIL element “data processing and analysis” with the disciplinary need to understand the complex nature of the historical record.
Beyond Boundaries: Digital Humanities, Life Sciences and IT research
Claire Clivaz – U of Lausanne
Should we be afraid of doing research with the life sciences and IT? Her answer is that it’s worth doing, and her talk is meant to encourage us to do so.
Bioinformatics, for example, has much to offer in terms of data visualization. Also gave a funny example of how rigid IT structures blocked the ingest of humanities content that had an offending apostrophe.
Noted that the emerging requirements coming from journals that require scientists to submit laboratory notebook information, such as experimental and analytical design offer an opportunity for collaboration, although I missed how this takes shape.
A Plea for a Method-driven Agenda in the Digital Humanities
Jonas Kuhn, Nils Reiter – U Stuttgart
He started by noting that eHumanities focuses on objects, artefacts and that the humanities disciplines have questions about these objects. In an ideal world, we could take a large number of objects into account and marry the questions of the humanities with the tools and methods coming from computer science. This is hard. For one, data is heterogeneous and project specific, so preprocessing is difficult. Tools have to be adapted, etc.
In sum, humanities often seeks the narrow and specialized, while CS seeks more generalized methods.
Their proposal posits identifying conceptual subtasks that are prerequisites for later analysis. These subtasks can have many forms, but are possible to identify. Once these have been identified, one can find projects that have similar subtasks. These subtasks may have no connection to the research questions, of course. Isolate the subtasks independent of the research questions and work on them in that environment. Some subtasks may have no tool, yet.
Example: character mentions in novels, which is related to Natural Language Processing fields CR and NER.
How About Tools for the Whole Range of Scholarly Activities?
John Bradley – King’s College London
Started out by stating that the tools built for DH purposes have had little impact on humanities research in general. If the tools aren’t finding an audience, there may be things we can do to address that.
What’s a tool? Something that allows us “to change material objects more than [man] could without them” (Feibleman 1967). Others view them similarly, i.e.- as ways of extending human abilities or enhancing human activity. Bradley described three tool paradigms (riffing on Feibleman):
- tools for making
- tools for exploring
- tools for thinking
Ran through some quotes and ideas highlighting how each class is defined. Ramsay and Rockwell, for example, refer to tools as “telescopes for the mind” in a 2012 book chapter.
- Focus has recently been on tooks for exploring. Is this adequate to support the continued interests of DH?
- Do we want to promote tools for making as a kind of scholarly activity? How can we describe this work so it appears as plausibly DH? Does it need to appear as an academic activity?
- Can we build tools that work like mathematics (or the piano) in the humanities (tools for thinking – an area of little exploration in DH)?