From Enriched Museum Collections to Social Web and TV: Seminar at BBC
1. Integra(ng
Social
Web
&
TV
with
help
of
seman(cs
Lora
Aroyo
&
Guus
Schreiber
Computer
Science
VU
University
Amsterdam
2. Seman(c
Web
@
VU
Amsterdam
• 40
people,
two
groups:
Web
&
Media
(Schreiber),
Knowledge
Representa(on
&
Reasoning
(van
Harmelen)
• A
few
ongoing
projects:
– europeana.eu:
EU
culture
portal
– NL
projects
on
access
to
cultural
heritage:
CHIP,
Agora
– EU
NoTube:
Web
&
TV
seman(c
integra(on
– PrestoPrime:
user-‐generated
annota(ons
and
content
for
TV
archives
– EU
LarKC:
plaRorm
for
massive
distributed
reasoning
3. The
Linked
Data
Web:
typed
resources
and
links
Painting Dublin Core ULAN
“Woman with hat
SFMOMA creator Henri Matisse
Web link
URL URL
7. The
power
of
simple
alignments
“Tokugawa”
AAT style/period SVCN
Edo (Japanese period) period
Tokugawa Edo
AAT is Getty’s SVCN is local in-house
Art & Architecture Thesaurus ethnology thesaurus
20. NoTube
slogan:
Pu#ng
the
user
back
in
the
driving
seat
Observa(ons:
• Personalized
services
are
now
common
• But:
user
data
is
s(ll
under
control
of
separate
applica(ons
• Result:
user
is
faced
with
mul(tude
of
distributed
personal
data,
hidden
in
tons
of
inaccessible
cookies
21.
22. NoTube
building
blocks
(1)
1.
TV
metadata
services
EPG
metadata
grabbers
–
from
170+
channels
(issue:
channel
URLs)
–
metadata
format:
TV
Any(me
–
real-‐(me
service
2.
Metadata
enrichment
– Add
links
to
external
Web
vocabularies
and
repositories:
Lupedia
service
23. NoTube
building
blocks
(2)
3.
Linked
Open
Data
for
TV
– Access
services
to
major
vocabularies
– Alignment
services
between
major
vocabularies,
where
needed
(e.g.
genre
typologies)
4.
User
acKvity
streams
– Standard
for
ac(vity
stream
representa(on,
i.e.
Atom
Ac(vity
Stream
– Access
services
to
ac(vity
streams,
e.g.
YouTube,
TwiVer,
..
– Trusted
access
to
“friend”
informa(on,
e.g.
implementa(on
of
standard
like
OAuth
2.0
24. NoTube
building
blocks
(3)
5.
User
profiling
Services
for
genera(ng
user
preferences
–
“Beancounter”
–
abstrac(ons
from
ac(vity
stream
User-‐model
representa(on
based
on
FOAF,
i.e.
weighted
interests
and
considering
context
6.
Recommender
services
Collabora(ve
recommenders,
e.g.
preferences
of
friends
Content-‐based
recommenders,
e.g.
program
about
Alma
Mahler
program
about
Walter
Gropius
Experiment
with
mix
of
these
recommenders
for
single
users
and
small
groups
of
users,
e.g.
families,
friends