The source code for this site is maintained at https://github.com/cmu-lib/dhweb_app.
The overall style was designed by Heidi Wiren Bartlett.
The server is hosted by CMU Libraries, with the generous help of Jonathan Kiritharian and Rashid Sidiqqui.
When setting out to do this project, we were faced with two conflicting desires:
Ideally, our data would come only from our original source materials: the published abstracts from conference proceedings or event programs. In short, our database needed to treat the individual works to be our core source of truth, with people's names and affiliations merely asserted by these works, rather than innate properties of that person.
To do this, we took a page from the event-based paradigm of the CIDOC-CRM, in which the attributes (like names, or affiliations) of most people or objects are not considered innate, unchangeable things, but instead the results of events or activities that took place at a certain time and place. Because this project is rooted in data from conference proceedings and event programs, we looked to the pattern of "context-specific assertions" that would let us encode that some individual
321 published as
Jane Doe from
UCLA in an abstract from the ADHO 2017 conference, and then published as
Jane C. Doe from
University of Maryland in an abstract from the ACH 2019 conference.
In this pattern, appellations and affiliations are connected to the authorship information for a given abstract, letting us point from each name for an author back to the source where we found it. (See, for example, the page for Scott B. Weingart.) We don't actually try to encode whether Jane C. Doe actually left UCLA in 2017 and joined UMD in 2019, just that they offered those affiliations and names in those particular public documents. Determining whether "J. Doe" is the same person as "Jane C. Doe" is a judgement call our editors make as they enter and clean data based on the context of the different abstracts and their knowledge of the field.