Report for the STIM “Core Site”
The STIM “Core Site”
team consisted of Jim Coleman, Project Manager, Sigrid Mueller, Project
Editor, and David Soergel, Programmer. Lois Brooks was “on site” manager
for the team after Coleman left Stanford in November, 1998. Coleman continued
as the “off site” manager through the end of the project period, and was
responsible for the organization of the “intellectual content” of the August
conference. Rosemary Rogers, Assistant to Tim Lenoir, handled all of the
technical details with the conference, and artfully managed the flow of
information between all of the PI’s, conference participants, and Team
Evaluation of the Stanford Team
The report for Year
1 notes the startup difficulties for the STIM project. These included:
a lack of appropriate management experience by the original project managers
and their supervisors, a lack of sufficient programming resources to carry
out the technology required to meet the project’s goals, and the choice
of an inappropriate technology at the outset.
An early evaluation
of the staffing resources after Coleman became project manager in November
led to one of the original project manager being laid off, and those resources
being devoted to a programmer. A similar evaluation of the original technology
solution, MediaWeaver, led to its abandonment as well. A full discussion
of the technology employed in the Core Site is available as an Appendix,
Used in Science and Technology in the Making.
By early in the
second project year, the Stanford team was able to deploy all the major
technology features (threaded discussion lists, questionnaires, comment
sections, file upload). At this point the early frictions between the PI’s
and the Team had also been overcome, a good working relationship with all
of the PI’s had been established, and all parties were able to make good
While the Stanford
Team was successful overall in accomplishing the tasks it need to do in
support of the project and worked well with the PI’s, it is clear that
both the early problems and the early departure of Coleman hampered the
speed of the startup, and the conclusion technology pieces of the project,
cross-site searching and the creation of the originally envisioned CD-ROM
version of sites. Neither of these is, we think, crucial to the goals of
the project, but both would have added to the value and functionality of
the STIM site.
The STIM Team
was also felt both by the Foundation and by some of the PI’s to bear some
responsibility in managing, encouraging and coaching the PI’s on how best
to make contact with their audiences. Some of this activity included discussions
on site design to achieve more “eyeballs”, some included setting up meetings
and conference calls between the PI’s and their contacts, some included
explorations of what methodologies, both traditional and newfangled, might
result in the creation of more content or interactivity.
There were also
problems with the perceived underfunding of the Berkeley site. This led
to additional negotiations and requests to the Foundation for additional
funds to carry out the work. While the funding issue was resolved to some
extent, the Berkeley efforts were never up to the level of the other sites.
These were tasks
that the Project Manager and Team took on because it was essential to the
goals of the project, but we were also occasionally uncomfortable assuming
a responsibility for the success of the project in areas that was not entirely
ours to control. In future federated projects, the Foundation should consider
placing the management of finances under the complete control of the project
director, rather than dispersing the funds directly to the PI’s, so that
s/he has sufficient moral and economic influence on the entire enterprise.
Evaluation of Technologies Employed
STIM used five technology
approaches to try to drive up interactivity with its target audience:
Of these, I would
deem questionnaires and email to be successful it all of the places they
were used; threaded discussion lists were successful in the MouseSite and
EVOnline, but not elsewhere; and comments and electronic submissions failed
to be used much at all. This
suggests that the more “traditional” forms of gathering information, questionnaires
and direct contact, should not be overlooked, even when attempting to used
more advanced technologies. The experience with discussion lists suggests
that they need to be manned and seeded, that is, pertinent individuals
need to be enlisted to participate, if they are to work. Facilities to
include comments, as raw, unstructured areas of opinion collecting, probably
need to be present for the sake of inclusiveness, but we should not expect
much from them. As to electronic submissions, the jury is out completely
on how such facilities could be made effective.
of Digital Materials
In Conclusion: Epigrams from the STIM Experience
It is difficult
to transform communities that have a life outside of the Web into communities
that work on the Web unless they believe they are doing real work
To be successful,
Sloan projects should attempt to integrate the work of the community they
wish to address.
To be “sticky”,
sites will need to be self-sustaining and provide an archival presence.
By self-sustaining, we mean having the ability to continue to generate
interesting and useful content; by archival, we mean having the
ability to manage that content for the benefit of its users.
used in such projects should be proven. The support of a research project
should not be a research project in itself.
The cost of the
infrastructure required is not small; nor is the skill set of the technical
team. Invest as much as you can in both.
Technology Used in
Science and Technology in the Making
Science and Technology
in the Making (STIM) consists of five individual projects and associated
Web sites. The main goal of STIM is to use the interactive capabilities
of the Internet to gather survey information and personal histories, encourage
dialogue among the makers of history, and provide an (inter-)active archive
in which focus groups are invited to contribute material. Their contributions
in the form of stories, artifacts, and written accounts are part of the
history presented on the Web sites. This leads historians to multiply the
perspectives of contemporary history and engage the community who made
the technology in a collaboration to write their own history. Thus, the
traditionally silent subjects of a historical study become personally involved
in the writing of their community’s history. Organizationally, the technical
development for the five STIM Web sites is carried out by the “core” STIM
team, located at Stanford University. Four of the five sites are hosted
at the main STIM Web site, sloan.stanford.edu, while Berkeley’s
project is hosted at sunsite.berkeley.edu except for the
“interactive” portions of the site which are carried out at the Stanford
site. The content of the site, site design, and implementation is the responsibility
of the individual PI’s. The Core Team has been available through the project
to assist the PI’s in digitization, reformatting, conversions, HTML, and
as consultants on site design and user interface issues. In the course
of the project, a set of basic functionality was developed to support the
collection of data pertinent to the site and user interaction/interactivity
with the site. Some of this functionality was defined formally, some of
it grew organically out of the project, some of it was developed for the
purposes of site management, and some in support of larger digital library
activities. These basic functions are:
Threaded User Forums
to a Site
Of Digital Documents
Graphs Of User Data
Tables Of User Data
Tools For PI’s
is gathered in a variety of places. For the EVOnline site, EV users/owners
are encouraged to log into the site as they enter it, and to answer a questionnaire;
the logon enables the system to keep track of what sections of the questionnaire
a user has completed. The user name is chosen by the user, and is not associated
with an actual name, although other demographic information, including
e-mail account, is solicited.
For other sites,
the user login/identification facility is part of other functions, i.e.,
part of submission of digital documents, or participation in a Web site.
In all of the sites, threaded user forums require entering a name, although
submissions may be made anonymous to the site.
There was much
discussion about requiring users to log into each site as they entered,
and to develop a system to track user movement/navigation through the site.
After much discussion, most PI’s decided that such a login requirement
would be a disincentive to participation, and a more sophisticated version
of user login/tracking was unnecessary. The only user tracking occurs for
the EVOnline questionnaire, as explained above.
Two sites use questionnaires
to gather data: EVOnline and the Blackout Site. The questionnaires are
delivered as Web pages. In the case of the EVOnline survey, the pages are
divided into four segments, due to the length of the survey. User logon
is used to keep track of which segments a given user has completed.
THREADED USER FORUMS
Among the principal
components supporting user interaction are threaded user forums. Each site
has at least one forum, most have several devoted to several, occasionally
changing topics. EVOnline permits users to suggest topic threads. Two of
the sites give the site administrator the ability to “publish” forum entries.
That is, the individual entry is not posted to the site immediately, but
goes through a Web form-based “publication” procedure where the site administrator
can alter or edit the entry if appropriate. In these cases, the original
entry is preserved.
gathered from forum to forum differs somewhat, as does the appearance of
the messages submitted. In each case, however, users can see all of the
contributions in an “outline” form, can then navigate to individual entries,
respond to those entries, or start a new thread on the topic. If an entry
receives a response, the original author is informed of the response via
email (if an email address has been supplied). The amount of required information
varies from site to site. Most forums require users to submit names, all
permit the forum entries to appear as anonymous entries.
USER SUBMISSION OF DIGITAL DOCUMENTS
Since one of the
purposes of the project was to encourage the submission of digital information
from the user communities, facilities have been developed to permit the
submission of digital materials directly through a Web form. Users submitting
files are asked for information pertaining to author, title, source, file
and resource type, and keywords. Site administrators are then notified
of the file submission, can examine the file and then take appropriate
EVOnline and the
Blackout site also have employed a general “comment” facility that allows
users to submit comments about the site on almost every page. The Blackout
site has extended this facility to include tracking information. That is,
the system keeps track of where a user was when the “comment” button was
pushed. The user can then specify the type of comment: a comment about
this page, a general comment, and so on. The “comments” themselves are
posted within the site for the Blackout Site, but not for EVOnline.
GRAPHS OF USER DATA
A Java-based graphing
applet (PopChart) is used in EVOnline to supply graphs of data extracted
from the surveys.
TABLES OF USER DATA
HTML tables of user
data extracted from surveys have been developed for the Blackout Site and
EVOnline. The interface permits the user to select the fields he/she wants
to view. The list of fields can be restricted by the site administrator
so that personal or sensitive information is not disclosed.
Each resource in
the site-archival files, comment pages, digital documents, tables, graphs,
user forums-can be cataloged by a site administrator using a metadata scheme
corresponding to the Dublin Core. This resource-level metadata cataloging
is then used to perform “smart” site searches.
In the case
of the Blackout site, the metadata scheme has been greatly extended to
permit the site administrator, or any other so-enabled user, to establish
a series of complex relationships between any two objects or sets of objects.
In theory, this scheme would permit any user to construct his own “site”
or view of the site out of portions of the site. For the PI, this facility
permits the ad hoc construction of interlinking pages through changes to
the metadata structure.
SITE MAINTENANCE TOOLS FOR PI’S
A series of Web-based
tools have been developed to permit the PI’s to change their sites dynamically.
Web form to change/set
Web form to archive
Web form to publish
Web form to set
fields that can be shown in a public table view
“private” view of
“private” view of
all submitted data
objects have been
submitted by users
new forum topics
have been suggested
new forum entries
are available for editing
comments have been
WEB/DATABASE SERVER SOFTWARE
STIM uses the Netscape
FastTrack server, version 3.1, to provide Web services. The server is a
Sun Ultra 1, running Solaris 2.6. RDMBS services are provided by Oracle,
COLD FUSION FOR APPLICATION DEVELOPMENT
With the exception
of the Blackout Site, Cold Fusion, version 3.1, is used as the main application
development environment for questionnaires, user forums, comments, file
submissions, table views of user data, and the site maintenance tools.
WEBOBJECTS FOR APPLICATION DEVELOPMENT
WebObjects is used
for all of the user interactivity functions within the Blackout site. It
is also used for the metadata cataloging modules. Given the complexity
of the overall metadata model, and the specific complexity of Blackout
Site, WebObjects proved to be a better development environment, even given
its significant learning curve.
RDBMS USE IN THE APPLICATION DEVELOPMENT ENVIRONMENT
Virtually all of
the user interaction/interactivity and site management functionality in
STIM is derived from tables set up in Oracle.
single table tracks user logon and session information while in EVOnline.
in general, each questionnaire is a single Oracle table. User responses
are mapped into table fields, with the application being responsible for
error trapping where this is not excluded through the design of the form.
In EVOnline, each questionnaire “page” is a separate table.
a series of four tables manage information gathered by users, legitimate
forum names, current forum topics, previous forum topics, and threading.
The generation of all of the forum pages is dependent on specific application
calls to the database, and each page in these sections is generated on
the fly out of data residing in the appropriate tables. Since one of the
required functions is the ability to change user submitted data, a separate
“archive” table gathers this information. Within the main forum table,
entries are flagged if they are have been changed as part of an editing
process. All of the forum entries are part of the same table to facilitate
cross-forum searching from the main STIM site.
a series of tables likewise manages all of the comments submitted, along
with user information.
of digital documents , tables manage the metadata associated with the
object. Current the object is stored in Unix file space, not a table, until
the PI determines the dispensation of the object.
For the graphing
, the data is extracted from the database via a SQL query, and
passed as text strings into the applet. The applet architecture of our
version did not support direct calls to Oracle.
For the table
views of data,
a table containing user selectable fields is used to
control what information is capable of being passed onto the user. This
is more flexible, but for the current applications, hard-coded views would
have been as effective, and perhaps more efficient.
cataloging, a series of tables, one for each field, manages the data.
[N.B.: it should be noted that the relational model is not ideal for metadata
cataloging using structures like Dublin Core or RDF. Here the benefits
of an object oriented model might be more fully realized.]
Other tools used
in the creation of the STIM site included Web authoring programs, such
as NetObjects Fusion, Adobe Capture for the creation of PDF’s from page
images, Photoshop for general Web graphics work, PowerPoint for embedded
slides, and Oracle Web tools for data capture/extraction.