Skinner, Karen (NIH/NIDA) [E] wrote:
Many thanks to all for this lively discussion, the helpful references,
and your generosity with your knowledge!
I could not access the video presentation yesterday but was finally
successful late tonight. It indeed was very interesting. The paper:
"Understanding user goals in web search" also appears really relevant,
but it is not readily accessible through the NIH online publications.
I look forward to going through the other references, and your comments
about them have been helpful already.
Just a comment about "negative knowledge." I did not know it has any
sort of formal meaning. I coined the phrase for my own purpose in
reference to a situation where some information might exist, but a
potential user might not be aware of it. For example, a consumer could
go to a local store and compare prices for refrigerators. But if the
consumer visited more stores, she could learn even more about prices and
models. If she visited only one store, all other information about
prices would be "negative" knowledge to her, because it does not exist
for her -- i.e., she does not know about it. Certainly, at some point,
the cost of expenditure of effort to "know" exceeds the benefit, whether
that cost is determined by an hourly wage equivalent, or some subjective
measure of the value of her time.
We may also call it "inaccessible knowledge". To lower the barrier to
knowledge accessibility, standard knowledge representation formats are
needed so it will become easier and quicker for search engine to find
and access relevant knowledge. Of course, smarter and more powerful
search engines will also need to be developed to exploit fully such
standard knowledge format ...
In science, such an analysis quickly becomes very complex. In some
cases, an investigator may not care if a certain study has been
conducted because they only trust the reagents or data they themselves
generate, and the existence of data and resources is irrelevant to that
investigator.
I think part of it is me-Science
(http://www.gridtoday.com/grid/963514.html). Not only does the
technological structure need to be changed, but the
social/cultural/legal infrastructure will need to be changed to
transform me-science into we-science. Even some scientists are willing
to share their data, whether others will trust the data is another
matter. How can we increase the trustability of data is an interesting
question? Would provenance help, for example?
On the other hand, suppose that "database X" did not exist, but the
existence of information that would have been found in it can be
identified and obtained only through locating and reading thousands of
individual papers. At what point does the cost of locating and reading
the papers by "y" number of users exceed the cost of the database? It
would seem that most of the cost would derive from the expense of
determining IF the knowledge existed. How many papers would the
scientist have to read before being certain the knowledge or data did
not exist?
In the paper "will a biological database different from a biological
journal"
http://compbiol.plosjournals.org/perlserv/?request=get-document&doi=10.1371%2Fjournal.pcbi.0010034
The author envisioned that the gap between literature and databases will
disappear. Scientists will be able to go back and forth seamlessly
between papers and databases. Of course, this will require changing the
way the papers were published (e.g., more structured annotation of the
paper will need to be provided by not only the publishers but also by
the authors) and the way databases link to the literature (e.g.,
database entries will automatically be linked to relevant papers or
relevant parts of the papers).
Hope it helps,
-Kei
