Main »

Draft Report

The workshop concluded Tuesday afternoon. We are working on this report so it is a document that will change significantly.

How the workshop was organized

Who participated?

The workshop brought together Humanists and HPC folk from SHARCNET universities, notably, Windsor, McMaster, Brock, Western and Guelph. Also invited were research administrators and deans at SHARCNET universities and granting council representatives. A fuller list of participants will be posted.

What did we do?

The workshop began with an invited talk by Stephen Downie of the University of Illinois, Urbana-Champaign. Downie has been using HPC through the NCSA for music information retrieval. He demonstrated a system that uses different techniques to categorize music in real time. The workshop then focused on identifying the challenges and opportunities at the intersection of humanities and HPC. We concentrated on what can be done by us at universities affiliated with SHARCNET.

Questions

Throughout the workshop questions were raised about the area of intersection, question like:

  1. Do humanists need HPC support? If SHARCNET supported humanities research, would there be any receptors?
  2. What are the questions in the humanities that need HPC support?
  3. How do we support humanities research within the SHARCNET remit? Could the pressing questions need other types of computing support?
  4. What can be done next? What can we try to see if there is a fit?

At the end we came up with three areas where we can do things: Documentation; Traning and Research Development; and Research First Steps.


Documentation

In answer to the questions above, one contribution we can make is to document the questions, challenges and success stories. We identified the following areas to document.

  1. Questions - What questions can humanists ask that HPC support could help answer?
  2. Challenges and Opportunities - What challenges and barriers do humanists face using SHARCNET? What opportunities are there?
  3. Success Stories - What examples can we point to of successful use of HPC in the humanities?
  4. Funding Pathways - Where can a humanist go for funding to support research that uses HPC? In particular, where can they get support to prepare datasets for HPC processing?
  5. Matchmaking - How can we connect interested researchers from the humanities to HPC folk?

Documentation can take various forms. Some forms are:

  1. To have success stories that act as examples on the SHARCNET site along with pathway information that can help humanists take advantage of SHARCNET support.
  2. To have pathway guides for research administrators that explain the options. These might also be useful resources for granting councils. One particular issue that came up was the need for a guide to explain to humanists how CFI can fund the preparation of text databases for research and HPC. See text database preparation model.
  3. To have a literature review and environmental scan of projects at the intersection that can help humanists learn more about this intersection. This could start on SHARCNET, but could also be submitted for publication. Likewise a theoretical paper that looks at what humanities questions can be asked with access to HPC would be useful to orient humanists and could go for publication.
  4. To have frank documentation of the challenges faced by humanists using SHARCNET in order to improve SHARCNET service. This could include a encounter audit that documents the process of using SHARCNET from the perspective of a humanist. Likewise to have documents that identify the accessible opportunities to encourage humanists.

Training and Research Development

Most humanists can't use HPC facilities without training and development support. We identified a number of ways that SHARCNET can work with the digital humanities community to develop appropriately trained personnel and interesting projects.

Graduate training

Few humanists understand how HPC can be used in the humanities. There is an opportunity to organize training for graduate students in the humanities in Southern Ontario that would build research capacity in the region while also providing a unique opportunity for graduate students. SHARCNET has the capacity to adapt existing training to provide truly innovative humanities research training which would, in turn, develop the community. We believe that in the humanities it is often through graduate students that transformations take place. SHARCNET will look to partner to develop a summer graduate training course.

Research assistantships

SHARCNET could offer Research Assistantships to graduate students in the humanities who are suitably prepared. Graduate students would be funded to work on humanities projects. This would probably work best where a graduate student is placed in SHARCNET so they can learn from HPC staff and explain the research practices of the humanities to staff. The RAship would facilitate projects while giving students innovative and in depth experience to follow on training courses.

Research fellowships

SHARCNET could set up a Fellowship program for researchers if there is the support. Researchers at SHARCNET institutions would apply to be fellows by proposing a project that could benefit from SHARCNET support. If accepted they would be given space and appropriate technical support for a year at a SHARCNET facility. Given that heavy teaching loads can be a problem in the humanities, we would also expect that SHARCNET Fellows would get teaching relief for the year. In turn Fellows would be expected to mentor colleagues and present their projects back to the community.

Charettes

One way to make rapid progress on models for humanities use of HPC would be to organize intensive prototyping sessions or "charettes". These would involve programmers, humanists, and computer scientists working together for a fixed number of days and presenting their prorotypes back after. This might be a way of modeling success stories that illustrate the opportunities. This would develop understanding on both sides.

Common tool building

We tentatively identified some areas where SHARCNET could develop some common tools. They include:

  • API for process running from another web server. A tool that would allow an authorized server, like a TAPoR server, to launch a secure process on SHARCNET and get back results.
  • A library of visualization tools for developing text visualizations and 4D visualizations.
  • Web drop-box tool for authorized running of code on SHARCNET. This could be used for competitions and exchanges.

SHARCNET could, where there are interested researchers, develop common tools in these areas.

Collaboration support

Humanists, like other researchers, work in teams across universities. SHARCNET has the capacity to offer easy-to-use conferencing for team research projects over the internet. The conferencing should be tested, documented and advertised as a service to humanities research teams.

---

Research First Steps

Some concrete projects that we imagined could move this intersection forward in the short emerged in our workshop. These are the opportunities.

Visualization and Agent Based

Visualization and Agent Based modeling are two areas where HPC techniques can be applied to research problems in the humanities. These might be areas to organize a week long charette that moves the collaboration forward.

On demand/responsive system work

Much digital humanities work is made available to researchers through the web whose practices involve using responsive web research systems. We hypothesize that digital humanities projects should probably run the web servers and develop the research publishing applications, but that there might be processes that a web server could launch on an HPC cluster. We need to find ways to connect projects like TAPoR to HPC clusters. See the section on building Common Tools above.

Large data storage and archiving

Humanities project increasingly need large scale data storage and archiving. While data preservation is the domain of libraries and archives, SHARCNET could provide data stores for services running on other servers. There is an opportunity to work with our libraries to develop a model that can support large-scale humanities projects.

Research Exchange Competition to support and credit tool building

A research exchange competition similar to MIREX (Music Information Retrieval Evaluation eXchange) could be run for Humanities HPC projects where a community would agree on challenges and have access to SHARCNET to advance methods that meet the challenges.

Navigate

PmWiki

edit SideBar

Page last modified on April 30, 2008, at 01:32 PM - Powered by PmWiki

^