2. TV on the Web: growing trend
2 New trends in television: social and semantic
3. TV on the Web: channel explosion
3 New trends in television: social and semantic
4. Source: Nielson Three Screen Report, March 2010
4 New trends in television: social and semantic
5. From „ Mobile Shopping Framework: The role of mobile devices in the
shopping process” by Yahoo! and the Nielson company, January 2011
http://advertising.yahoo.com/industry-knowledge/mobile-shopping-insight.html
5 New trends in television: social and semantic
6. Including the Web in your TV
Yahoo! launches ConnectedTV platform for Web-
based widgets on TV (e.g. Flickr, YouTube,
facebook, twitter) – Jan 2009
6 New trends in television: social and semantic
7. Augmenting TV with the Web
Blinkx BBTV makes
video information
and its textual
transcript clickable,
and links to Web
sources such as
IMDB and Wikipedia
www.blinkxbbtv.com
Also Mozilla has a
project on showing
content around
videos using HTML5
www.drumbeat.org
7 New trends in television: social and semantic
8. Some Web-TV solutions today
Stand alone boxes such as
• TiVo – original DVR, added on-demand video,
YouTube, music and photos from the Web
• Boxee – STB offering its own store of apps
• AppleTV – relaunched as $99 product tied to
iTunes content, and iPhone/iPad integration
+ Hybrid boxes tied to specific IPTV providers
+ Games consoles (Sony, Microsoft, Nintendo)
also adding Internet and video services to TV!
8 New trends in television: social and semantic
9. Some Web-TV solutions today
First TVs with
integrated Web
and individual
app platforms
in 2011.
Future TVs will
be „connected“
as standard. LG SmartTV, pic courtesy
http://www.wired.com/gadgetlab/2011/01/lg-smart-tv/
9 New trends in television: social and semantic
10. State of the art in TV
• TV content shifting to the Web as delivery
platform
– An explosion in available content at any time
• Web content shifting to the TV as
augmentation of the TV experience
– An explosion in additional content at any time
10 New trends in television: social and semantic
11. Limitations of today‘s TV
• Too much content in one place
– How to find what you want to watch? Sort
between live TV, TV on demand, archives, video
portals and P2P-TV
• Too much functionality at any one time
– The whole Internet while you watch TV. But what
do viewers really want to be able to do
additionally (parallel) to watching TV?
11 New trends in television: social and semantic
12. Social TV
• Integrate the TV experience with the so-called
Social Web
– Who are my friends and what do they watch?
– What do my friends like -> maybe I‘d like it too
– Where are my friends now -> connect via the
shared TV experience
• Key goal for social TV
– Enhance my TV experience through my friends‘ TV
experience
12 New trends in television: social and semantic
13. Semantic TV
• Add formal semantic descriptions for
– TV programmes
– TV schedules (EPGs)
• Link those descriptions to other semantic data
on the Web, cf. Linked Data
• Two key use cases for semantic TV:
– Filtering of TV content -> personalisation,
recommendation
– Augmentation of TV content with Web data
13 New trends in television: social and semantic
14. NoTube project
• Integrating TV & Web with help of semantics
– Open and interlink TV content in a Web fashion
with Linked Open Data
• Putting the user back in the driving seat
– Connect multitude of distributed personal data
with explicit semantics
• TV is not bound to the device
– Computer as TV & vice versa
– Mobile device as remote control
14 New trends in television: social and semantic
16. Bridging Web and TV cultures
16 New trends in television: social and semantic
17. Rest of this slideset
• Technological background (Semantic Web,
Linked Data)
• Semantic annotations for TV data (semantic
TV)
• Extracting knowledge from my activities and
social graph (social TV)
• TV content recommendation (personalized TV)
• The further future: finally … interactive TV
17 New trends in television: social and semantic
18. (1) Semantic Web, Linked Data
“If computers can understand the
meaning behind the information
they can
learn what we are interested in
and
better help us find what we want.”*
* Source: http://www.slideshare.net/HatemMahmoud/web-30-the-semantic-web
18 New trends in television: social and semantic
19. The Semantic Web
The vision of what was termed the “Semantic Web“ first came to public
attention through an article in Scientific American in May 2001.
“The Semantic Web is not a separate
Web but an extension of the current one,
in which information is given well-defined
meaning, better enabling computers and
people to work in cooperation.”*
* Source: T. Berners-Lee, J. Hendler, O. Lassila; “The SemanticWeb”, Scientific American, 284(5):34–43, May 2001.
19 New trends in television: social and semantic
20. HTML
HTML was too limited for Web documents – it is purely a presentation
format. The tags in HTML have no meaning outside how content should be
rendered in the browser, and so the meaning of the content must be
interpreted by a human, hence excluding any possibility of machine
processing.
<u>James Bond</u> James Bond MI5
<b>MI5</b><br> Her Majesty's
Her Majesty's Secret Secret
Service<br> Service
Secret HQ<br> Secret HQ
<i>007 England</i><br> 007 England
20 New trends in television: social and semantic
21. XML
<name>James Bond</name>
<company> James Bond MI5
<shortname>MI5</shortname> Her Majesty's
Secret
<fullname>Her Majesty's Secret Service
Service</fullname> Secret HQ
<address><street>Secret HQ</street> 007 England
<postcode>007</postcode>
<country>England</country>
</address>
</company>
The core idea of XML – Extensible Markup Language – is to provide for
definitions of markup which allows self-describing tags, i.e. tags which
describe the meaning of the content they mark up rather than its
presentation
21 New trends in television: social and semantic
22. RDF
RDF provides a graph structure for making statements about things.
Individual things, and not just files, are given an URI identifier.
This is where the Semantic Web begins.
<flight>Flight AI288 http://my.org/flightAI288
<from>Vienna</from>- :from http://my.org/Vienna
<to>Innsbruck</to> :to http://my.org/Innsbruck
dep <dep>1.1. 1200</dep> :dep 01-01-2009T12:00
arr <arr>1.1. 1255</arr> :arr 01-01-2009T12:55
price <price>88€</price> :price „88“
</flight> :currency http://my.org/euro
from is a child element of flight from is a property of the resource
(syntactic structure) http://my.org/flightAI288
22 New trends in television: social and semantic
23. RDFS
RDF Schema begins to formalise the meaning of things spoken about in
RDF on the basis of computational logic. RDFS permits simple ontologies
(models about concepts and their properties) to be defined, which can be
used to conclude new knowledge.
http://my.org/Vienna
is a http://my.org/City http://my.org/flightAI288
:from http://my.org/Vienna
http://my.org/City :to http://my.org/Innsbruck
subClass of http://my.org/PopulatedPlace :dep 01-01-2009T12:00
:arr 01-01-2009T12:55
http://my.org/Vienna :price „88“
is a http://my.org/PopulatedPlace :currency http://my.org/euro
23 New trends in television: social and semantic
24. OWL
OWL broadens the possible expressivity of the ontology. This makes
richer models of knowledge about things possible, but at the cost of those
models being more complex for a computer to process.
http://my.org/Vienna
isPlaceIn http://my.org/Austria http://my.org/flightAI288
:from http://my.org/Vienna
http://my.org/Austria :to http://my.org/Innsbruck
isPlaceIn http://my.org/Europe :dep 01-01-2009T12:00
:arr 01-01-2009T12:55
isPlaceIn is a transitive property :price „88“
:currency http://my.org/euro
http://my.org/Vienna
isPlaceIn http://my.org/Europe
24 New trends in television: social and semantic
25. SPARQL
The final block of the Semantic Web that we will cover in this introduction is
SPARQL, the query language for semantic data using the RDF data model
(which includes OWL).
Is there a flight from Vienna to
somewhere in Austria for a price
under 100 euros?
http://my.org/flightAI288
SELECT ?flight :from http://my.org/Vienna
WHERE :to http://my.org/Innsbruck
?flight :from http://my.org/Vienna :dep 01-01-2009T12:00
?flight :to ?place :arr 01-01-2009T12:55
?place :isPlaceIn :price „88“
http://my.org/Austria :currency http://my.org/euro
?flight :price ?price
?flight :currency http://my.org/euro
FILTER
(?price < 100)
25 New trends in television: social and semantic
26. Semantic Web principles
• Every concept can be identified with URIs
• Resources and relationships are typed semantically
• Partial information is acceptable
• Absolute truth is not necessary
• Evolution as a development principle
26 New trends in television: social and semantic
27. Linked Data principles
• Use URIs as names of things
• Use HTTP URIs so that people can look up those names
• When someone looks up an URI, provide useful information
• Include links to other URIs, so that they can discover more things
27 New trends in television: social and semantic
28. Semantic Web vs Linked Data
“In contrast to the full-fledged Semantic Web vision, linked
data is mainly about publishing structured data in RDF using
URIs rather than focusing on the ontological level or
inference. This simplification - just as the Web simplified the
established academic approaches of Hypertext systems -
lowers the entry barrier for data providers, hence fosters a
widespread adoption.”
- Reference
vs
28 New trends in television: social and semantic
32. DBPedia Mobile
Pictures from revyu.com
Try yourself:
http://wiki.dbpedia.org/
DBpediaMobile
32 New trends in television: social and semantic
33. Resources and representations
non-information resource
http://dbpedia.org/resource/Berlin
HTML representation RDF representation
.../page/Berlin .../data/Berlin
33 New trends in television: social and semantic
34. Linking things, not documents
http://dbpedia.org/resource/ABBA sameAs
http://www.bbc.co.uk/music/artists/d87e52c5-
bb8d-4da8-b941-9f4928627dc8#artist
34 New trends in television: social and semantic
35. Browsing things, not documents
http://dbpedia.org/resource/ABBA
themeMusicComposer
http://dbpedia.org/resource/Knowing_Me%2C_
Knowing_You..._with_Alan_Partridge
35 New trends in television: social and semantic
36. Asking for things, not documents
Which music
artists have
composed the
theme music
for a BBC
comedy
program?
36 New trends in television: social and semantic
37. (2) Semantic annotation for TV
• What can we annotate in TV?
– The program schedule
– The TV program
– TV program segments
• How can we annotate TV?
– Feature description (low level, analysis based)
– Metadata (date, creator, legal notice)
– Content description (title, summary, genre,
concepts)
37 New trends in television: social and semantic
38. Why have metadata?
Archives from where
content has to be
found and retrieved
have been the place
where the need for
accurate
documentation first
arose.
38 New trends in television: social and semantic
39. Broadcast metadata
• Data about data
– All digital resources (A/V, scripts, contracts, reports, pictures, etc.) are
data
– Metadata is created at all stages in broadcasting from commissioning
to playout
• Three main categories
– Administrative metadata
• Replacing project and asset management paperwork
– Technical metadata
• Format, processing, identification, location, database, network
– Descriptive metadata
• All asset related information, human readable
39 New trends in television: social and semantic
40. Need for common standards
Broadcaster Broadcaster Broadcaster TV Content NoTube
1 2 3 Creator 1
Exchange of
information
hampered by lots
of proprietary
interfaces
TV Content TV Content TV Archive TV Archive
n+1
Creator 2 Creator 3 1 2
40 New trends in television: social and semantic
41. EPGs
Screenshot http://www.ifanzy.nl
41 New trends in television: social and semantic
42. EPG data
• An EPG is composed of two parts: content descriptions and
broadcast description
• Content descriptions contain static data about television
programmes such as a brand name (e.g. EastEnders),
description or plot summary, type of programme, (e.g. series,
movie, news), genre(s) (e.g. drama) actors, directors,
recording data, etc.
• Broadcast description is expressed by variable data, such as
channel (e.g. BBC ONE), format (e.g. 16:9) and broadcast
media (e.g. digital television)
42 New trends in television: social and semantic
43. TVAnytime (1/2)
• Unique document structure
– Program description
– Program location
– Program segmentation
– User description & personalisation
– System aspects
– Content rights
43 New trends in television: social and semantic
44. TVAnytime (2/2)
• Advantages of TV-Anytime
– It is network and middleware independent
– Supports related material, segmentation, locators, group
information etc.
• Applications of TV-Anytime
– ARIB
– DVB (MHP, DVB GBS, DVB IPI, DVB CBMS)
– Asian User Groups, Korea
– US’ Consumer Electronic Association
– HbbTV
44 New trends in television: social and semantic
46. Other models in use
• egtaMETA - a unique metadata exchange schema dedicated
for the exchange of ads between ads agencies and
broadcasters. NoTube was an early tester of the schema in its
personalised advertisements use case.
• BMF – an abstract semantic model designed for metadata
exchange in the professional media production domain. ARD
in Germany is starting to use BMF.
• Presto Space – format generated by the project of the same
name to provide for digital preservation of audiovisual
collections. Used by NoTube partner RAI.
46 New trends in television: social and semantic
47. Metadata interoperability via
NoTube
http://notube.tv/tv-metadata-interoperability/ for more information
47 New trends in television: social and semantic
48. BBC /programmes
The BBC have made their EPG data machine-
readable and published it on the Web
48 New trends in television: social and semantic
49. BBC /programmes: add .rdf
http://www.bbc.co.uk/program http://www.bbc.co.uk/program
mes/b00rl5y1 mes/b00rl5y1.rdf
49 New trends in television: social and semantic
50. BBC /programmes ontology
This may the first TV content
ontology, but certainly not the
last!
Key organisations in the TV
standards domain are exploring
the publication of metadata in
RDF or SKOS:
• EBU (Core)
• TV-Anytime
• IPTC (NewsML)
The final step must be a
common shared ontology
integrating the different
schemas (cf.W3C Media
Ontology and API)
From http://purl.org/ontology/po/
50 New trends in television: social and semantic
51. Channel identifiers
• Collected resolvable channel identifiers
together with relevant metadata in RDF, e.g.
1700+ channel identifiers of Freebase
http://www.cs.vu.nl/~ronny/notube/tv-channels.rdf
51 New trends in television: social and semantic
52. Genre taxonomies
• BBC, TV Anytime, YouTube, IMDB, tvgids.nl …
• Convert them into RDF concepts and define SKOS
relations between them, e.g. EBU has done this for
the TV Anytime Classification scheme
52 New trends in television: social and semantic
53. Concept extraction
• NLP tools identify named entities in text and attach
an unique identifier to them
e.g. OpenCalais, Zemanta
• Focus on key classes of entity such as person, place
or organisation
• Use of Linked Data for common concept identifiers
• Ontotext developed specifically for TV metadata the
tool LUPedia
53 New trends in television: social and semantic
54. LUPedia
(http://lupedia.ontotext.com)
54 New trends in television: social and semantic
56. Linking TV content to Web content
David Dickinson
starring
Tim Wonnacott
birthplace
Barnstaple
56 New trends in television: social and semantic
57. Pause
57 New trends in television: social and semantic
58. (3) Extracting knowledge about the
user
Idea: generating user profiles from data the user
creates on the Social Web, and in this way
facilitating a personalised TV experience
without an intrusive user profiling process.
58 New trends in television: social and semantic
60. Activity Streams
• RSS/Atom feeds include a title, description,
link and some other metadata;
• Activity Streams extend this with a verb and
an object type
– to allow expression of intent and meaning
– to provide a means to syndicate user activities
• Supported by Facebook, MySpace, Windows
Live, Google Buzz and…
60 New trends in television: social and semantic
61. 61 New trends in television: social and semantic
62. Getting TV into the Social Network
„ BBC iPlayer adds
Twitter and
Facebook to
socialise TV”
– Share what you are
watching on iPlayer
– Sync viewing with
friends
– Real time chat
Techcrunch Europe, May 26 2010
62 New trends in television: social and semantic
63. TV viewer actions
• Recorded
• Consumed
• Loved
• Bookmarked
• …
63 New trends in television: social and semantic
65. Bringing it all together
65 New trends in television: social and semantic
66. Eurovision example
• Analyse tweets with the #eurovision tag over
a set time period (during the program)
• Extract country and positive/negative remark
66 New trends in television: social and semantic
67. Getting the user‘s interests
67 New trends in television: social and semantic
69. FOAF
• RDF based format http://xmlns.com/foaf/spec/
– Defines properties for describing a person and
their relations to other people and objects
69 New trends in television: social and semantic
70. Weighted Interests
• Add weighting to the foaf:interest property
See http://xmlns.notu.be/wi/
70 New trends in television: social and semantic
71. FOAF as common vocabulary
71 New trends in television: social and semantic
76. (4) TV content recommendation
• Recommender strategy
– Collaborative recommendation
• You share interests with your friends
• Statistical analysis: what content is liked/watched
quantitively more by others with similar interests/history
– Content-based recommendation
• An interest in X means a potential interest in Y
• Pattern-based analysis: what content has related concepts
to the content liked/watched by you
– Hybrid recommendation
• Best of both!
76 New trends in television: social and semantic
78. Recommendation lifecycle
Graphic by Libby Miller, BBC
78 New trends in television: social and semantic
79. Linked Data recommendations
• The content-based approach:
– Identify weighted sets (patterns) of DBPedia
resources from user activity objects
– Compute distance between DBPedia concepts in
the user profile and in the program schedule
through its SKOS-based categorisation scheme
– Choose the matches above a certain threshold for
TV programme recommendation
79 New trends in television: social and semantic