|
Finding
Multimedia
Online:
Moving Beyond The Hunter/Gatherer
According
to monica schraefel, a CITO-supported researcher at
the University
of Toronto, we are still at the hunter/gatherer stage
of Internet
use. If a user wants explore a topic, it's up to them
to travel
the Web, hunting for relevant information and
gathering it
into some kind of cohesive whole.
The advent
of broadband and the explosion in online multimedia
only makes
this hunter/gatherer approach even more difficult and
less
effective.
Dr. schraefel's
research involves developing a better way to connect
users
to the content they want. She is currently working on
two
CITO-supported research projects; one will create the
foundation
for a seamless, intuitive tool to help users of
varying experience
find the music they want through a wide scope of
information
appliances. A second, related project explores how to
serve
and manage multiple streams of data for immediate,
interactive
and real-time response via the Internet, all while
allowing
the user to explore these multiple channels through a
highly
intuitive interface.
In
a recent
interview with Trends in Research, she outlined the
nature
of her research and the impact such technologies might
have
on both users and industry.
Trends
What are some of the research challenges in terms of
creating
ways to navigate multimedia content?
monica
Online, you get lists to search. That's only
useful
if you know what you want. So, we want to give users
that
data, the music, first. You're interested in finding
new music?
Ok. Come to this space, and you'll explore the music
by listening
to the music itself rather than scrolling through
lists about
it.
Several
challenges are: how to model this 'data first'
approach -
what's to the left or right of the current selection
and why?
Also, how do we let the user tag something to come
back to,
and if we do that, what does that tag look like so the
user
can find it again?
Trends
How does the technology you are suggesting work?
monica
That's part of the research question we are
trying
to solve. There are four main layers to the
development of
the technology: the interface and underlying structure
that
represents and serves the data, the Input/Output
devices that
let users explore that data, and the file structures,
or databases
that store the data and the network that responds to
requests
for that data. Our project creates interesting
problems that
need to be solved in each of those domains. My grad
students,
Gonzalo Ramos and Paulo Pacheo and i are working on
the interface/interaction
domain; Steve Mann and his students are collaborating
with
us on the Input/Output; Renee Miller and her PhD
student Periklis
Andristo are starting to work with us on the
multidimensional
database questions; Michael Murphy and Mark Banbury
from Ryerson
are collaborating with us and AT&T Research Lab's Jim
Snyder
to deal with the network/streaming issues; Sony Music
Canada
is contributing their content and work flow expertise.
Our prototype
represents classical music. We picked classical music
because
it has its own formal system for identification. But
that's
not particularly useful for someone who wants to
investigate
classical music and doesn't know a fugue from a
paratita.
So our approach is to let users hear the music first -
like
turning on a radio - and then change the dial to tune
in more
of what suits them.
One of
our research challenges is to discover what makes
sense for
users to be to the left or to the right (or up or down
or
back and forward) from that current position on the
dial.
Should "slow" be to the left of where they are and
number
of instruments up from there, so if the user moves up
they
get a symphony or down and to the left they get a slow
piece
with one instrument, and if they pull back they get a
slow
piece with a guitar instead of a piano? Part of our
preliminary
work is to get a better idea of how users who don't
know the
formal music terms describe music so that we can build
a better
model of what's left of Beethoven, for
instance.
Trends
Physically, how do users navigate through the data?
monica
I've been working with Steve Mann, who has
developed
miniature radar that allows you to use your hand in a
high-dimensions
space to manipulate the data. You move your hand up if
you
want to explore the area in depth and pull back a bit
for
less information. You could move your hand from right
to left
to change the speed or number of violins in the piece.
You
manipulate the space until you found something you
liked and
from there you could select another category to get
more information
about the artist or to sample other music like this.
This
is the advantage of mass rather than shelf space.
We've also
thought about using a rolling shift stick, which might
be
safer in a car; it depends on where and how it will be
used.
Trends
How will technology such as this influence how users
interact
with and search for information?
monica
Users will approach large amounts of data and
explore
it first, in the same way that you might pick a book.
You
go through the bookstore, look at the displays, look
at the
covers, read the descriptions. You can look at this,
music,
the same way. It's not a query-based approach; it's a
data
first approach. As you find something of interest, the
information
about the data will be there for support, not for
access.
That is, if you find a piece you like, you might say
"who
wrote that?" and that information will be available.
Indeed,
you would be able to click through on that information
to
find out more about that composer, should you want to
do so.
Or you might want to find out who the performer is, or
what
was happening in history. The goal of the system is to
give
you support in accessing the data, the music, first,
and then,
if you want, you can investigate information about
classification.
Imagine
looking through all the posters at the art gallery: if
you
don't know anything about art, you look through to
find something
you like. You either take it home and just enjoy it,
or you
find out who the artist is. Then maybe you look for
more work
by that artist, and then perhaps as you get into it,
if you
get into it, you learn about the Impressionist
approach to
art and so on. If you go on the Web today, you pretty
much
have to know you want to find Impressionists first or
a specific
artist. We plan to provide a more human-based approach
to
the information, and then use the power of the
computer to
enhance that approach.
Trends
How will this sort of technology affect the music
industry
as a whole?
monica
Sony is excited about this because it's an opportunity
to
expose people to music they otherwise may not have
known about
because it was too hard to access. I'm excited about
its potential
for foregrounding independent artists. No independent
artist
has broken through on the Internet.
Few people
have the time or motivation to download an artist
they've
never heard of. The most downloaded material on
Napster are
the top 40 hits - stuff where users already know the
name
and the song of the artist they're
downloading.
Our approach
will allow people to explore music without a high cost
of
time, travel or purchase to find what they want. Sony
is keen
again because this system can make it easy for a
person to
say "hey, I like that; where can I get that?" - which
is,
we've discovered, one of the top five questions users
have
about music they don't recognize - and have it
available to
them then and there, with rich supporting information
about
the recording, too, should they wish.
This is
our test space; if we can make this work for music as
we imagine
it, then we can translate it to multiple domains. This
lets
designers from libraries to industry make their
catalogues
available in meaningful ways - not as lists of names,
but
as versions of the things themselves. A colleague was
recently
pointing out how, coupled with 3D, this approach could
be
used to look for building materials, special
mouldings, for
renovations. Why not?
«Back
to top
|
