►
Description
In this presentation from Day 3 An interesting take on Swarm use came from Ameer Ahmed in his talk “Ontologies for structured data in Swarm” where he presented an approach to categorising structured data such as museum catalogues.
A
Hi
everybody,
so
this
is
a
two-part
presentation.
The
first
part
is
gonna
talk
about
essentially
my
experiences
in
taking
data
from
a
relational
database
systems
to
to
graph
to
specifically
to
linked
open
data.
So
so
the
my
part
will
talk
about
the
experience
on
how
we
did
it
for
a
particular
industry,
specifically
cultural
heritage
sector,
and
the
second
part
will
be
by
today.
We
will
talk
about
how
the
application
of
ontologies
can
be
used
in
in
swarm
and
data
form.
A
So
just
a
quick
background,
so
I've
been
in
this
space
for
about
14
years,
specifically
with
artstor
ethica
in
the
mellon
Foundation
I
started
off
with
CDW
a
light
schema.
This
is
this
is
a
schema
which
describes
how
to
describe
a
work
apart.
So
we,
the
problem,
was
that
they
have,
we
have
museums
all
around
the
world
and
they
all
have
heterogeneous
data
sources
and
different
systems.
A
So
the
goal
was
to
create,
with
the
help
of
Getty
research,
define
one
schema,
one
ring
to
rule
them
all
and
using
their
money
go
around
the
world
and
install
software
and
harvesting
software
and
harvest
all
that
data
into
one
global
repository.
So
we
started
off
with
with
the
Metropolitan
Museum.
We
converted
them
to
this
standard,
and
that
was
our
initial
trial
after
that
I
think
some
people
from
Germany
museum
that
specifically
they
liked
our
model
and
they
would
do
they
ought
they
took
the
I?
A
Guess
the
leadership
after
that
and
decided
to
change
the
project
and
it's
called
it
was
called
Lido
which
then
again
continue
to
work
with
getty
to
go
around
the
world
implementing
standards
for
museums.
So
you
have
universal
access,
so
the
the
problem
back
in
those
days
were
there
were
there
were
no
systems
to
catalog
art
history.
A
So
the
the
idea
of
shared
shelf,
which
is
a
concept
of
allowing
to
catalog
images
and
media
for
academic
institutions,
came
into
play.
So
we
took
that
proposal
and
we
gave
it
to
the
mellon
Foundation
and
they
accepted
it
and
we
ended
up
creating
shared
shelf
shared
shelf.
This
is
still
all
background
for
the
talk
I'm
gonna
talk
about
so
shared
shelf
is
a
it's
a
platform
which
allows
you
to
catalogue
objects
of
art
using
not
only
standard
text,
but
you
have
vocabularies
specifically
vocabularies,
which
are
provided
by
by
Getty.
A
A
Think
it's
180
yeah
about
a
hundred
80
institutions
across
the
world,
now
utilize
ship
shells.
So
just
to
give
you
an
example,
this
is
a
record
of
Abraham
Lincoln
in
shared
shelf.
You
can
see
you
have
titles
very
entitled
different
languages.
You
have
the
Creator
who
who
actually
you
made
the
sculpture
again.
This
is
your
authority
you're
linked
Authority.
The
linked
Authority
in
this
sense
means
that
the
data
is
curated.
A
So
while
we
were
doing
that,
we
were
working
with
Columbia
University
and
they
had
another
project
in
mind
which
was
built
works
registry
now
built
works.
The
this
they
claimed
to
be
the
biggest
curated
data
source
for
built
environment,
so
they
comprised
of
about
42
different
source
collections
around
the
world.
The
picker
I
think
strategic
partners
from
India
to
Italy
to
America,
so
we
had
a
very
diverse
set
specifically
relating
to
built
environment.
A
So
using
that
project
we
thought.
Okay,
we
will
use
this
project
as
a
guinea
pig
to
expose
the
data
in
the
linked
open
data
environment.
So
a
quick
review
on
how
we
we
started
this
project
with
our
partners,
part
institutions.
We
converted
the
data
into
a
format
ingested
it
into
shared
shelf,
the
the
BW
are
staff
links
it
to
various
authorities
and
then
finally,
they
push
it
out
to
the
public.
Who
can
then
utilize
that
record
again?
A
A
They
uploaded
to
the
community
to
the
local
layer,
the
editorial
board,
reviews
it
and
finally
approves
and
publishes
the
data
in
the
community
space
which
then
can
be
utilized
by
people
like
Liz,
who
was
a
high
school
student,
doing
research
on
Mayan
temples,
so
this
is
this
is
just
a
workflow
of
an
individual
versus
an
institution
on
how
data
is
contributed
to
this
method.
So,
yes,
all
of
this
is
just
a
setup
for
linked
open
data.
A
What
is
this
well,
this
is
it's
a
cluster
of
data
which
the
goal
here
is
is
to
make
data
openly
available
in
a
particular
format
and
to
provide
links
among
various
data,
so
it
becomes
machine
readable.
As
of
a
few
weeks
ago,
there
were
about
1200
data
sets.
It
started
off
in
2008,
with
like
a
few
hundred
when
we
were
doing
it.
It
was
like
around
500
to
750
and
now
it's
about
1200.
A
A
A
Well,
going
back
to
what
I
showed
you
before
shared
shelf
was
our
main
platform,
so
we
created
an
ontology
on
shared
shelf
and
then
we
map
that
on
top
ontology
to
our
data
schema
and
then
we
expose
our
data
through
a
semantic
portal,
essentially
endpoints
to
access
that
data.
And
then
you
register
the
data
set.
You
register
your
endpoint
URL
and
then
you
also
provide
your
entire
data
dump
to
the
data
hub
IO
linking
well.
We
have
to
obviously
link
this
to
make
it
work
right,
so
we
have
to
identify.
A
Who
do
we
link
to
get
IVA
cavalry?
Is
here
obviously
the
number
one
choice
because
they're
not
only
funding
the
project,
but
but
it's
a
extensive
vocabulary
used
in
art,
history
for
for
naming
for
Geographic
places
and
other
links
like
geo
names
and
dbpedia,
and
how
do
we
match
them?
Well,
this
also
happened
in
phases.
The
first
part
was
quite
simple
because
we
had
the
ID
of
the
record,
so
it's
creating
a
very
simple
triple
with
taking
that
ID
and
saying
is
the
same
as
this
ID.
A
This
is
a
quick
review
of
what
what
the
XS
D
is
so
over
over
here
in
the
middle.
We
have
a
work
record
again.
This
is
how
the
catalog
objects
of
art.
So
this
is
your
container
model,
and
this
is
your
display
record
the
example
here
being.
If
you
have
like
a
shot
to
Cathedral
or
Eiffel
Tower,
there
will
be
your
work
and
then
the
various
views
of
that
building
or
object,
we'll
be
your
display
rendered.
This
kind
of
this
is
fixed.
A
A
How
do
we
convert
it
to
go
into
LOD?
Well,
we
there's
a
transformation
process
between
this
XS
d.
That
I
showed
you
into
an
ontology
right,
which
is
a
shade
shelf
ontology,
which
we
are
going
to
use.
Once
we
have
that
conversion
we
can
generate
instances
based
on
data.
This
is
just
some
examples
on
how
we
actually
converted
the
X
is
D
to
its
counterpart
in
the
ontology
I'm.
Just
gonna
fly
through
that
and
that's
an
example
ontology
the
same
XS
d
that
I
showed
you
before
right
here.
A
So
it's
a
different
view
of
it,
and
this
is
your
display
record.
So
again,
the
concept
of
work
record
and
the
spray
record
is
is,
is
key
to
understanding
these
components
so
an
actual
live
example.
So
this
is
your
shade
shell
system,
but
display
record
kind
of
very
flat.
As
you
can
see,
you
publish
this
there's
an
ID
right
there,
you
publish
through
the
website
website
is,
is
using
an
Omega
plug-in.
Essentially
so
it's
all
automated
and
then
it
publishes
into
into
this
website.
You
can
obviously
visit.
A
This
is
all
open
and
once
you
click
on
the
semantic
view,
that's
where
you
see
that
data
in
its
RTF
format
and
from
here
we
can
actually
visualize
it.
So
you
can
see
the
project
is
built
works
through
registry.
It
contains
Wall
Street
building,
which
is
linked
to
New
York
record
and
then
from
New
York.
You
can
see
other
records
which
are
linked
to
New
York
and
then
going
over
to
to
the
Getty
vocabulary.
You
can
then
actually
navigate
that
asaurus
using
broader
terms
narrower
terms.
A
A
This
is
just
a
quick
overview
of
again
all
the
various
components
happening
with
ch
itself,
so
I
mean
it
was
a
it's
a
fairly
large
project
about
100
people
worked
on
it
over
the
course
of
10
years
and
their
various
tools
which
came
into
play,
and
this
is
the
one
section
that
we
are
talking
about
today.
So
this
allows
you
to
export
data
into
a
linked
open
data
environment.
A
So,
over
the
course
of
a
year
we
we
did
about
8
releases.
We
try
to
do
one
release
a
month,
but
it
didn't
really
work
out
that
way,
but
it
was
essentially
taking
baby
steps.
You
know
taking
something
starting
something
very
small,
starting
with
with
a
simple
flat
dataset,
which
is
a
display
record
and
pointing
to
a
local
instance
of
a
vocabulary
and
then
ending
the
year
with
with
full-blown
HL
support.
What
does
that
mean?
That
means
now
any
type
of
recovery
that
you
connect
to.
A
When
you
publish
your
data,
your
data
actually
goes
into
the
linked
open
data
space
with
links
to
all
the
different
cloud
sets,
and
one
final
point
I
want
to
make
here
is:
we
also
allow
the
project
itself
to
become
a
vocabulary,
so
that
means
you
could
catalog
and
then
you
wanted
to.
You
have
I,
don't
know
a
list
of
temples
that
temple
could
be
the
temple
vocabulary,
which
could
then
be
opened
and
used
by
other
people.
A
I'm
gonna
talk
about
briefly
about
data
enrichment,
just
how
that
process
works.
So
imagine
you
have
very
simple
life
of
taro
text
as
the
title
location
is
Paris.
You
bring
this
into
your
taking
the
export
of
this
and
import
it
into
hello.
Do
you
find
an
order?
You're
fine
is
a
it's
a
project.
It
allows
you
to
work
with
messy
data
and
also
has
links
to
various
vocabularies
in
the
LOD
space.
So
we
take
that
data
and
we
define
a
particular
data
source
in
this
case
with
dbpedia
and
the
thesaurus
for
Geographic
Names.
A
A
And
this
is
just
one
last
diagram
which
talks
about
how
we
actually
make
our
project
as
a
controlled
vocabulary,
but
essentially
what
it
comes
down
to
is
every
single
project
can
be
deployed
in
this
environment
and
ends
up
with
a
with
an
endpoint.
So
then,
you
can
share
this
endpoint
with
anybody
who
wants
to
reference
your
project
and
and
that's
it.