Summit on Digital Tools for the Humanities

Introduction

Bernie Frischer and Anita Jones from the University of Virginia organized an invitational Summit on Digital Tools. This was sponsored by the Univesity of Virginia, the Institute for Advanced Technology in the Humanities and the National Science Foundation (USA). The web site is http://www.iath.virginia.edu/dtsummit/ . Participants were asked to submit a short concept paper which was used to select those to be invited. the summit was organized around discussion sessions not presentations of canned papers.

The Summit focused on the challenge of research using digital resources. How can we study digital resources? What sorts of questions can we ask of digital collections and how do the questions change? How can the computer help with the study of digital resources and what tools are needed to take assist in interpretation? Tools, while in the name of the Summit, are the technical end of a larger transformation in our research practices.

A report on the Summit will be put up, possibly at The report http://www.digitalhumanities.org.

Day 1

Keynote by Brian Cantwell-Smith of the University of Toronto

Descartess division of mind body was a tactical move to put aside the complex human problems and focus on the body and mechanics. This helped advance science but now haunts us.

Logic from Boole to Russell was an atempt to widen the success of the sciences to include what was left out. Computers are a continuation of that dream.

C-S has been asking what a computer is and has decided it is nothing. We have a sense that computers are special - there all sorts of arguments to that effect. But it isn't special. It doesn't fuse meaning and mechanism the way people do.

Computers are not what we wanted them to be - they have not overcome the mind/body split - they are not pure mechansism. They have dissappointed us in interesting ways. But there is still a belief that they can span mind/body - meaning/mechanism.

The ideas about whay computers are special is what matters! The history of the idea and cultural contexts of computing is what is interesting. Computers are a historic moment that started in 1850 that is not just getting past Descartes. Computing is only now getting past the Cartesian split to matter and what matters.

As humanists we have to be reflective - we have to take responsibility for the consequences of tools, their contexts, and the practices they legitimize.

Issue of generalization without abstraction. Myths can generalize, but they are not abstract. Blogs are that too. Could the humanities be moving to non-abstract generalization and their study.

Text Analysis

I moderated a work group on Text Analysis, Data Mining and Corpora.

Issues

• The importance of Google both as an example of something simple to use and as an example of something everyone uses. Likewise things like Flickr, wikis (and the wikipedia), deliciou.us and Amazon are genuinely useful and broadly available. Some like Google even have APIs on which more complex tools can be built.
• The need for a killer app that brings people in. We also discussed the issue of whether humanities computing should expect to produce killer apps or popular tools. Should scholarly computing expect broad application of itself?
• We discussed the problem of incompatible collections. Scholars in Classics can search most of what has been written in Ancient Greek with the TLG in one place which is why computer practices have penetrated the discipline. In other disciplines there is no coherent corpus or if the works are there, they are spread out over incompatible collections. On this note people commented on the need to revisit the TEI and how it has been used.
• We discussed the institutional obstacles. First there is a cluster of influential universities/departments that set the research agenda (Harvard, Yale, Princeton ...) where there is almost no recognition of the importance of digital practices. As long as disciplinary leaders ignore the digital we will be working at the edges of the disciplines. Second, the scholarly associations are focused primarily on print publication as evidence of progress. Thirdly, there are few rewards for those who do tool work or digital work.

Recommendations

• Create a lobby organization - this would push, shame and recognize digital scholarly practices. It could push the branding of a "scholar's web 1.0" that is distinguished from the Google-driven web. It would
  • set API standards (web services) so that collections can be accessed in a standard fashion by other tools
  • set data standards for scholar
  • set deposit standards for open access
  • set low level standards for projects to be classified as scholarly
• Need to understand the methods, practices
• Need to cross collections - we should be able to search across important stuff
• Need for utilities
• Need for training and documentation

Authoring and Teaching Tools

This section is from my notes. -- VikaZafrin - 04 Oct 2005

Wendell Piez reported on authoring and teaching tools.

The group talked not so much about authoring pedagogical tools as about the milieu in which these tools are used.

There may be no field as such; there's a mix of different things going on in different places.

Deeper Questions

• Role of technological standards, how and where that gets done;
• Problem of publication and communication awareness, how we draw people in;
• How to get credit for building tools when it's stigmatized in our communities (tool construction seems to fall on the teaching side of teaching & research);
• What "genres" need to emerge;
• What we're trying to author, epistemologically;
• How to make it relevant in humanistic terms;
• Relating this work to what humanists already know and do.

Discussion

• Looked at current tools & their limitations, including tools built on XML and HTML.
• Noted that we're doing pretty well with tools for people who can pay attention to angle brackets -- but not so well with regard to those who have no patience for code.
• Many tools are sufficient for the task but are ugly, not optimal and not going anywhere.
• Many are not customizable in the ways that we need.
• Non-text (graphics, sound etc.) supported insufficiently.
• Fields are dominated by one or two vendors and a bunch of student projects. (No conclusions from that, but interesting to contemplate. -WP)
• We need content management systems. (This is a huge problem in the private sector as well. -WP)
• Frustration with vendors: is it a general theme? Is it really open source vs closed?
  • Closed is often easier to manage and maintain.

Conclusions and Recommendations

• Good: more jobs are available, although they're not always what we need them to be (but at least our skills are in demand).
• Promising areas from which we could derive benefits:
  • social software;
  • peer-to-peer work modes;
  • students' attitudes (generally receptive);
  • platforms for collaborative work/learning;
  • students' competencies;
  • decentralized expertise and information sharing;
  • search engine building
• Would like to continue developing an environment of exchange, exposing the guts of the innovating systems and making underlying data visible and accessible.

Interface Design

Interface design raises instutional issues as well as engineering issues. Open scholarship, open data, open source would be required for this. We don't want an interface - a successful intervention will be invisible.

(GR) But what if we wanted interface to be taken seriously as a design issue. Then we don't want it to be transparent.

Education

Need to normalize tools in research and teaching. Avoid workshops and move to model where tools are easy to use, can be used in real projects, and are woven into curriculum.

We need to create centres of excellence parallel to the EPOCH program. We networks of centres where institutions focus on their strengths. The networks should span national boundaries.

Collaboration is important. We have to develop new collaborative practices like side-shadowing and wiki-like collaboration. We then need to account for this in peer review practices.

Ontologies, metadata and markup

In this group we discussed the challenges and needed tools for ontologies, metadata and markup. There were certain tensions:

• The tension between local and exchange ontologies. Many projects are developing idiosyncronic ontologies that can't easily be merged with others. Can translation up and down work?
• The tension between XML and all the investment in XML tools and the limitations of XML like its limitations for exploratory markup.
• The tension between content creators recording metadata and metadata being added at the time of deposit in repositories (with the help of experts.)

We came to the following recommendations:

• We should try to build on XML and build around its limitations.
• We should use the simple standards out there like Dublin Core
• We need good ontology management systems. These should be able to help us compare ontologies and create them.
• We need tools that automate the generation of metadata and ontologies
• We need tools to link parsers that identify things like people, places, and dates, to authority lists.
• Disciplines need to tackle the difficult issue of developing disciplinary ontologies
• We need simple exploratory markup models
• We need highlighter markup tools and tools that let us view texts through different markup views.
• We need to get editing tools into common and commercial tools like MS Word.

Research Methods

This group began by observing that there are different methods for different methods. Here is a list that was generated:

• Reading
• Finding, discovery, searching
• Comparison
• Interpretation
• Writing
• Synthesis - finding and reporting
• Publication
• Rich attribution
• Documentation of sources
• Collection - building corpora
• Selection and exclusion
• Filtering
• Authenticity analysis
• Analysis
• Corpus statistics
• Analysis
• Modeling and representation
• Classification interrelating
• Statistical analysis
• Geographic/spatial coordinating
• Visualization
• Temporal coordination
• Chaos
• Ordering and arranging
• Pattern recognition and representation
• Conjecture
• Serendipidous
• Criticism

The challenges and opportunities ahead are:

• We can develop new methods
• We can understand our methods through modeling them and automating them
• We automate methods
• We can define existing methods and teach them

Actionable recommendations included:

• Develop classification tools for developing and applying classifications
• Collaboration tools
• Simulation methods
• Standardize methods
• Teaching methodogy and reflect on methods - this issue of methodology is itself an issue
• Methods should inform tools

Geospatial Information Systems

Most of us are shoehorning ourselves into existing tools and fogetting to ask what it should do. We lack a development tradition in the humanities. This is political and structural not a software issue. We humanists expect that tools should be easy, but it isn't (won't) be true in GIS. Therefore we need "knowledge brokers". We need:

• A tool to manage space and time simultaneously
• A machine not a tool - a black box.
• Mapping tools that can work in 2D or 3D at any moment
• Authentication and documentation trails. Also collabortion trail
• Promote widespread adoption of GIS to young scholars. Also take these to leaders (distinguished scholars, chairs, deans and so on) and asking what would it take to get this to work.
• The who, what, when, how, and why tool

Collaborative Software Development

The issues were divided into issues around:

• Platforms,
• Tools
• Styles of programming
• Version control
• Cross-institutional collaboration
• Obligation and time management
• Tenure and reward
• User community
• Distance
• Legal issues, open source - need for a common license
• Economic sustainability - how can we engage commercial firms

Some technical issues were:

• Extensibility and modularity
• Shared libraries
• Iterative methods
• Existing project management tools

Some social-cultural issues:

• How to sell what we do
• How to inspire people
• How to manage
• Need for a portal - make digital tools the water cooler of the humanities
• We need to do science fiction - narrativize wish list

Questions

• Question of the tool. What is a tool? How do tools fit into the humanities?
• Who are tool builders serving? How do we engage the public? How do we address disciplinary powers?
• What are the cultures of practice? How would tools fit into them? How could computer assisted techniques change interpretative practices?
• How could we engage those not in the digital humanities? What sorts of educational and outreach activities do we need to engage in? Do we want to?
• What tools are out there? How would we review them?
• What tools are needed? How do we know what is needed?
• What do we know about who uses tools and how they weave them into practices? Who could help us study interpreters?

Day 2

In the morning of the second day we had breakout groups around four areas that emerged in the plenary of the previous day. The groups had to develop a sense of the goals in the area, case studies, and specs for an example tool. After lunch the groups reported back. The organizing group is looking to summarize the outcomes.

Resource Exploration

The Resource Exploration group, which I moderated, focused on the challenge of studying large and disparate resources across the web. In many fields there is a critical mass of digital resources, but those resources can't be easily merged for purposes of study. The structure of resources are incompatible and often behind license barriers. Content is out there, but it is hard to find, hard to aggregate, hard to compare, and hard to study across resources. The challenge was therefore to develop goals for research interoperability and cross project study.

The tool we imagined was a Scholar's Google Assistant. We recognized that it would be hard to replicate Google, which can currently search across the Web. The Scholar's Assistant would help overcome some of the limitations of Google. For example:

• It is hard to limit Google's search to sets of sites. The SGA would allow one to create aggegations of resources and restrict searching to those aggregations.
• Google doesn't take advantage of structure in resources, metadata, or ontologies. SGA would add intelligence to Google.
• Google doesn't let one save and share results (or aggregations.) A SGA service would keep track of ones search history and allow one to share search strategies and results.

Roy Rozensweig who was the reporter for this session has posted on his wiki a summary of this session that we worked on at http://echo.gmu.edu/toolcenter-wiki/index.php?title=Exploration_of_Resources.

Time, Space and Uncertainty

This group looked at the problem of how one indicates uncertainty when creating and exploring resources with spatial and temporal information. They proposed a tool that lets one navigate and annotate/link/mark 3D, 2D and temporal information. They proposed a certainty slider that lets one annotate any link or tag with multiple indications of certainty. Thus you could have certainty sliders for different criteria or different evaluators.

It is interesting the varieties of uncertainty we have. For example, I might know someone was born in December, but not which year - so I am certain of the month, but not the year. How does one represent that?

The point was also made that while many of the commercial systems we have are designed for certainty (exactly when you cheque was cashed), in the humanities we deal with uncertain events all the time. This could be something we can contribute out to the community.

Annotation and Interpretation

The workgroup proposed an Ignorance Management Tool - great name. They proposed a tool that was essentially a highlighter tool for reading and interpreting through annotation and markup. I found it interesting that reading is taken to be the typical proactice or site of interpretation. Is interpretation only an extension of reading? There is probably interpretation in writing - in the feedback loop of reading what you are writing. Are there other sites of interpretation?

Collaboration

The collaboration group proposed a wiki for the tool development community. To some extent we have that with this wiki and others.

-- GeoffreyRockwell - 29 Sep 2005