background image
18 DigiCULT
tive of what we say and those that we in this room
need to request from the developers.'
Talk of better text indexing or image understand-
ing technology by the Rome Eleven would make lit-
tle difference because `so many people are already
clamouring for them', he said. Needs specific to dig-
ital cultural heritage had to do with veracity and `all
those things that make museums, archives and librar-
ies more interesting than bookshops'.
He stressed: `We urgently need a technology for
better understanding of similarities between docu-
ments in terms of usage, for example. Maybe if we
say this loudly enough here it will actually make a
difference because that is not currently a big issue in
the research field.'
The Google search process - `typing 1.87 words
into a window, looking at 60,000 documents in a
Google window, then at 12 of them and then you
read one of them' - was one way, he said, but the
process did not look into layers of information to
cover what Pia Borlund had earlier called `the per-
ceived difference between what I know and what I
need to know to do some task'.
The Forum set about building its own model
of the required process, a system that, as Traugott
Koch described, `invites you to start at some point in
a hierarchy, trickling down, exploring up and down
and wandering through the hierarchies or networks'.
The experts knew that users were often unsure
what they wanted, but knew what it was when they
saw it. Search engines had to be able to notice how
users acted during searches and try to detect what
type of search was being undertaken: specific infor-
mation, a home page, entertainment, or just ideas.
The engines, therefore, had to be aware that objects
were similar in many different ways in different infor-
mation spaces.
Sometimes information was just stumbled upon.
Sometimes it stood out because of the way it had
been used previously. Connecting these options could
create a powerful way of modelling information paths
without taking them beforehand.
Then put `attitude' into the mix, a connection
between objects of critical importance, and add
`trust'. Dr Karlgren commented wistfully: `Allowing
this sort of thing to emerge dynamically is impor-
tant ... to make a system that can extract this sort of
information and make it usable later. I think that is
the technological support I would be most happy to
build.'
In fact, some systems were already making analy-
ses on some of these different layers, he thought, but
nonetheless `focusing on the particular needs with-
in the cultural heritage sector might prove beneficial'.
The Moderator brought them back to practicalities
again, asking: `So what kind of technology allows us
to build this? Are there any technologies that allow us
these layers of information, and enrich as we interact
with them?'
Glamorgan University's Douglas Tudhope had good
news: `In the NKOS [Networked Knowledge Organ-
ization Systems] network,
5
we are currently look-
ing at standards for protocols for using knowledge
organisation systems, classification thesauri for differ-
ent kinds of use which could be search or browsing. I
believe in the next few years there will be some sort
of standard protocols that are going to be the build-
ing blocks to enable these different approaches.'
Seamus Ross could see value in this. `Participation
in the use of an information resource increases its
value and usefulness over time,' he contended.
Lund digital librarian Traugott Koch thought
museums could learn something from the library
Web services described by Lorcan Dempsey,
6
Vice-
President and Chief Strategist of the Online Com-
puter Library Center
7
(OCLC) in Dublin, Ohio, US.
He had devised a library service made up of thou-
sands of Web services providing `all the smart atom-
ic bits and pieces and they can be built into learning
object repositories or whatever scenario you might
want to use'.
Traugott Koch said: `I would ask museums to be
active in the creation of such services themselves
with things they have to do anyhow, to decompose
some of the things they do in order to show how
they could be reused by others in the first place.'
P
EER
-
TO
-
PEER
OR
NOT
D
r Jose asked about peer-to-peer retrieval. It sent
queries to relevant peers. It was more demo-
cratic. Usage information could be added to it to cre-
ate the kind of retrievals the experts were talking
about.
The others were dubious. `A fascinating technol-
ogy but it requires all peers to share an understand-
ing of the goal and representation of the data', said
one, adding: `You need to make sure that you are not
just equal but are basically identical.' Another thought:
`These ideas have been around for a long while and
they have not been solved.'
Jussi Karlgren wondered what he would tell a small
cultural heritage institution if it telephoned to ask
about digitising its collection. He mused: `Should I
start discussing the benefits of client-server versus
peer-to-peer solutions? I think not. I should probably
5
Networked Knowledge
Organisation Systems,
http://nkos.slis.kent.edu
6
Lorcan Dempsey, http://
www.oclc.org/research/
staff/dempsey.htm
7
Online Computer
Library Centre (OCLC),
http://www.oclc.org