1. Linked
data
at
the
BBC
Sofia
Angeletou
Data
Architect
Linked
Data
Pla3orm
Core
Engineering
2. overview
• what
is
the
BBC
doing
with
Linked
Data
• what
have
we
learnt
• what
are
our
main
challenges
3. Linked
Data
Pla3orm
• “To
organise
the
BBC's
online
content
around
the
things
that
ma5er
to
our
audience”
• Link,
reuse
content
and
expose
it
in
novel
and
interesAng
ways
• Available
to
many
online
products
4. what
does
that
mean?
• A
lot
of
data
• Defining
the
scope
and
the
data
model
very
well
and
clearly
• Enabling
tools
and
services,
self
provisioning
6. data
• creaAve
works
(content)
• things
(reference
data)
• ontologies
– content
– things
– management
and
ownership
7. services
• live
sites
– /sport
– /educaAon
–
news
pilot
• coming
soon
– music
news
– radio
8. lessons
learnt
Adding
complexity
while
not
directly
adding
value
to
the
proposiAon
is
not
a
good
idea
• “Nice
to
have”,
“cool
thing
to
do”
are
not
compelling
reasons
when
dealing
with
a
producAon,
24/7
live
pla3orm
9. lessons
learnt
• Trying
to
be
a
good
LOD
ciAzen
should
be
done
carefully
• Examples
– more
than
1M
unused
FOAF
triples
– Geonames
ontology
very
large
and
difficult
to
delete.
• Very
hard
to
manage
unused
models
when
serving
many
clients.
10. lessons
learnt
• CauAously
use
restricAons
– blank
nodes
– funcAonal
properAes
– disjointness
– domains
and
ranges
(cause
inference,
they
are
not
for
type
checking)
11. main
challenges
• decommissioning
legacy
• shared
ownership
of
common
topics
• culture
– tagging
with
a
purpose
– what
is
your
proposiAon
(why
you
want
to
do
it)
12. main
principles
• what
is
the
audience
proposiAon?
– only
build
what
delivers
the
funcAonality
and
extend
iteraAvely
– only
add
data
that
add
business
value
• clear
ownership
– content,
concepts,
ontologies
– management
and
permissions
• bo^om
up
approach
– first
test,
then
generalise
– controlled
pilot
builds