Cognitive Artifacts in Complex Work - Ambient Intelligence for Scientific Discovery - page 169

formation objects represent the information objective, its use. An information
object may be defined as the “required or desired objective of use from an in-
formation artifact.” The information objects consist of components drawn from
artifactssuchasarticlesandabstracts,and meet the unique requirements of a re-
search project. Bishop [4] refers to these components as “document surrogates,”
but here we find the objects inherently containing the objective of information
seeking. For the information object, the document is often the “surrogate.” These
surrogate objects are “building blocks” of research, but are not complete artifacts
or sources of knowledge. Objects are often drawn from the defined components
of an artifact (e.g., part of a method from the Methods section of an article),
but they are not the text of the component.
Individuals are also sought within this distributed system as information
and memory resources for specific knowledge. Lab members often first ask a
colleague for an answer, if known as a resource for that topic. This is supported
by behaviors such as the public act of distributing printed research articles to
other lab members; this displays personal interest in a particular line of research
and one's capacity to discuss or address issues represented by those papers.

4.5 Information Behavior and Search Patterns

While scientists in the study had a wide variety of information resources from
which to draw, they limited their research to only two services (PubMed and
Google) and were found to use few other resources available to them for pub-
lished research. The data show (across both field locations) over 90% used the
PubMed search engine for published scientific research. Most used Google for
publicly available information such as other scientists' web sites and chemical
suppliers, with 4-5 alternative search engines used on only an individual basis.
While they may spend hours locating and scanning papers online, and hours
finding referenced papers unavailable online, they spend very limited amounts
of time and cognitive resources in the task of searching .
Scientists preferred using the simplified search interfaces of the PubMed ser-
vice and Google search engine over the more sophisticated or specialized search
tools and interfaces available in their information ecology. They paid little atten-
tion to user interface features and showed no trend toward skill development by
using advanced search forms or features such as saved searches (e.g., PubMed's
Cubby ). Over time, search tasks become routinized operations.
Few scientists in the study sought new resources to supplement those in
regular use. Instead we observed that, once a set of resources and pattern of
use was established, it tended to remain in use over time, even as the available
resources in the ecology changed. We found scientists predominantly using the
research tools they had learned in graduate school. The PIs preferred MEDLINE
to PubMed, having learned its interface as graduate students before PubMed was
available. The current graduate students used PubMed almost exclusively.
We make a distinction between (cognitive) information artifact and infor-
mation object. The artifact is the physical printed document or digital file. An
information object is the component “objective” extracted from an artifact. It is