Design Patterns for DH Projects
May 14th, 2010 | hugh cayless
In my proposal, I listed 3 ideas:
1) I’ve recently started work on linked data for papyri.info, using data harvested from different parts of the collection and an RDF triplestore that provides entry points to the data from many angles. I’m very interested in tools and methods for doing this kind of work.
2) I’m co-PI on a project focused on linking text with digital images of text, and annotations, possible topics there include tools for working online with digital images, SVG, TEI facsimile, and/or transcription theory.
3) System design principles for digital humanities projects. I’d like to talk about ways of modeling information and delivery systems so that they are loosely coupled, RESTful, sustainable, easier to archive, and easier to integrate with other projects.
After a bit of reflection, I think these are all going in the same general direction, which is probably closest to #3. We’ve been doing online DH projects long enough that maybe we can start to distil useful patterns for representing types of data, handling the linkages between entities, formats and methods for encoding information, and interfaces for managing and interacting with information.
Can we come up with patterns for handling things like:
- Images
- Digitized manuscripts
- Geographic visualizations
- Data aggregation
- Search interfaces
- Browsing
I’m not so much talking about the implementation details here as the affordances. What ought you to be able to do with a project that exposes a lot of images, for example? I think the images should be available at high resolution, but should also be viewable in a browser in such a way that you can zoom, pan, and link to a particular view. That’s what I mean by a design pattern. Questions of “how” including toolsets, frameworks, and formats are interesting too, but those are often going to be dictated by the environment.
A few more patterns/practices off the top of my head:
- License with CC-BY
- Make data available as a single download
- Pay attention to URI design
What do you think?
May 14th, 2010 at 1:59 pm
I just have to say, hooray for how far RDF/Linked Data has come since the 1st THATCamp. It was mostly a curiosity then, and it’s really exciting to see you talking about it in terms of a specific project!
In other words, I’m totally there.
So far, I spend my time in javascript and PHP, so the tools I’m familiar with are ARC (arc.semsol.org) and manipulating the RDF/JSON format (haven’t used the jQuery plugin that’s supposed to do RDF).
The nice thing about ARC is the plugin that serializes graph to the SIMILE/Exhibit format, which lets you start up visualizations and navigations pretty quickly and easily, at least for smallish datasets per page.
Would love to talk about these ideas at THATCamp!
Patrick
May 15th, 2010 at 6:56 pm
I’m very interested in the Linked Data angle of this. I’m coming it at it from a less technical background, but have been working with Freebase, Internet Archive, and the Archives of Michigan to explore using Freebase as a semantic publishing platform and building apps around Civil War metadata from national and state sources. It’s been pretty slow going, but we’re confronting a lot of useful conceptual and technical questions as we go. More info here: wiki.freebase.com/w/index.php/CWD150.
May 17th, 2010 at 11:59 am
[…] version, to be deployed in case of a loss of technology. (Also of interest is Hugh Cayless’s session proposal). Some sites don’t have such easy answers, like older GIS sites that depend on a specific […]
May 17th, 2010 at 12:04 pm
This session idea ties into my (more loosely defined) session idea. In mine, I generally want to talk about what we can do to make projects sustainable in the long term, even if that means giving up some of the “cool” stuff (I’m thinking interactive maps and flash based things) in the long term but keeping the data, metadata, high res images, and databases available for download.
May 17th, 2010 at 1:48 pm
[…] to digital forms later on. In many ways, this points to individual application of themes that Hugh Cayless mentioned in his post. How should the lone scholar deal with archival resources that are not […]
May 17th, 2010 at 9:59 pm
Sounds like there’s interest in linked data and design patterns too. For what it’s worth, I wrote a fairly detailed post on the linked data stuff I did for papyri.info on my blog. A lot of the things you’d want to do in preparation for linked data work are also good design patterns, like treating entities (people, places, etc.) as resources and having separate (well designed) URIs for them.
May 22nd, 2010 at 1:25 am
[…] and standards remain a topic of interest to me: Hugh Cayless proposes that we take a serious look at how new design patterns could help digital humanists […]
May 22nd, 2010 at 3:31 pm
Here are my abbreviated notes for the session, ideas to help with project preservation:
* use virtualization for older platforms
* design site as static as data allows from the beginning
* limit the number of platforms, esp. ones with data export
* design sites that degrade gracefully in loss of technology
* create a video/screenshots, etc of a site as a form of documentation – it records the site’s functionality which can be ephemeral
* think of the site as layers – separate, for instance, map stuff (KML) from presentation
* make your site available for lots of copies
* available for spidering (e.g static HTML version)
* clonable repositories
* open licensing
* Documenting – white paper at the end of a project, talking about the idea behind the project. Should be a joint effort between scholars and coders
* Use standards
* Create content people care about 😉
Also, group discussed whether every project needs preservation & why there is no digital humanities version of Yelp for projects/tools and what would have to change in order for that to change (it comes down to solving the “social capital problem”).