4. Social Semantic Web
• Interconnect the islands of the social web with
semantic technologies
• Enhance the semantic web applications with
the wealth of knowledge from user generated
content
7. What do people do on the social web
• Create or state relationships
8. What do people do on the social web
• Exchange messages
• Search
• Buy
9. Social web
• Limitations
– Social platforms are isolated from one another
– There is no standard for exchanging data between
them
• Nobody wants to have a common standard for
exchanging data :)
– Database hugging (see TBL’s Ted talk from first
course) <=> User base hugging
10.
11. Social Web - content types
• Videos (youtube, vimeo, google video, …)
• Bookmarks (delicious, stumbleupon, digg)
• Blog articles (wordpress, blogspot, technorati)
• Microblog posts (twitter, facebook statuses)
• Images (flickr, picasa)
12. Social Web - content types
• Characteristics
– Very different types
– Most of them do not contain text
– Bookmarks are references to the other types of
resources
– Important who creates or bookmarks the web
resource
– The only way to describe all these types of content
=> tags
13. Social web - tagging - Folksonomies
• A folksonomy is “tagging that works”
• Another definition: the result of free tagging
of net objects, identified by URL’s. The set of
tags and resources annotated by users
constitutes the folksonomy.
14. Folksonomy characteristics
• No hierarchy, no is-a relations
• Vocabulary reflects the ideas and vision of the
users
• Terms in the folksonomy are related only
through co-occurrence
• Much more dynamic than ontologies to
represent the way users categorize content
15. Types of folksonomies
• Broad folksonomies
– Many people tag the same object. More people
can use the same tag on a resource
– Ex: delicious.com
16. Types of folksonomies
• Narrow folksonomies
– A resource is tagged by a limited number of
people (usually the author)
– Used more in content sharing sites (youtube,
slideshare, flickr) for describing multimedia
content
– Less information than the broad folksonomies but
also relevant
17. Folksonomies - problems
• Different lexicalizations of words
(plural, singular, synonyms)
• Words with multiple meanings
• There’s no hierarchy - problems for
classification of resources
18. Folksonomies - representation
• Hypergraph representation
• Hypergraph=a generalization of a graph where
an edge can connect any number of nodes.
• a hypergraph H=(V,E) where V is a set of nodes
and E is a set of non-empty sub-sets of V. E is a
subset of the power set of V.
19. Folksonomies - representation
• The nodes of a folksonomy hypergraph are:
– The users that are tagging
– The resources that are being tagged
– The tags that are being used
• The edges of the hypergraph are triples of type
(user, resource, tag) which represent the tagging
action performed by a user using a tag on a given
resource.
• This means that the edges in the hypergraph
belong to a subset of the U X R X T product
20. Folksonomy - representation
• => a folksonomy F=(U,R,T,Y)
– where Y U X R X T.
• Folksonomies can also be represented using
ontologies
• The hyperedge are represented through a
node of type tagging and the node is linked
through properties to the user, resource and
tag
21. Vocabularies for the Social Web
• Ontologies that describe the social web
– Content (SIOC, DC)
– Social relations (FOAF, Relationship, XFN)
– Tagging (Tagging, SCOT, MOAT)
24. Social Web Vocabularies
• Relationship vocabulary -
http://vocab.org/relationship/.html
• Defines different kind of properties that
describe the relations between two persons
• Ex: acquaintance, ancestor, apprentice, child,
close friend, collaborator, knows in passing,
knows by reputation
• Would be very useful in social networks where
you have hundreds of “friends”
25. Social Web Vocabularies
• XFN - XHTML Friends Network
• Simple microformat for describing
relationships
26. Social Web Vocabularies
• Neuman’s tagging ontology -
http://www.holygoat.co.uk/projects/tags/
• Tagging - the hyperedge n-ary relation
between a tag, a resource, a tagger and a
date.
• Tag - class to define the tags
• Doesn’t express tag meanings or
lexicalisations
27. Social Web Vocabularies
• SCOT ontology
– Extension of SIOC and Tagging (Neuman’s)
– + concepts and properties
• TagCloud (set of tags used in a context)
• CoOccurrence
• Source - the namespace where the tagging was
performed
28. Social Web Vocabularies
• MOAT - meaning of a tag
• it introduces the concept of TagMeaning,
linking a tag concept to a related concept in a
domain ontology.
• moat-project.org/ontology
30. Social Search
• 3 types of social search
– Collective social search
– Friend filtered social search
– Collaborative search (question-answering)
31. Social search
• Collective social search
– Similar with wisdom of the crowds
– Search could be augmented with “hot” topics
– Problems with trust - does the user really trust the
results from the masses
– Need to understand the user’s search process - is
the user exploring or narrowing the domain
32. Social search
• Friend-filtered social search
– Provide data “validated” by your friends or your
peers
– Add it next to traditional search results
– Advantage of TRUST
– See who added or recommended the piece of
content returned by the search
33. Social search
• Friend filtered social search - issues
– Do your friends have relevant and available
content for most of your searches
– Problems with the different types of social content
- most of them only characterized by tags
– Difficult to understand the context around a
resource (the tagging context for example)
34. Social search
• Collaborative Search - question answering
– A number of users work together to find the
answer to a problem
– Example: yahoo answers (asynchronous)
– Example: Aardvark - IM based - find the user from
your network who might know the answer and
dispatches the request
35. Social Buying
Amazon sets the example as always
Most read part of the product description
-people reviews
Just imagine having reviews from someone
you already know and trust
37. Semantic social web - our vision
• What we try to do
– Provide feedback for learning using social web
data
– Provide friend-filtered social search
– Provide relevant learning recommendations
– Help the learner establish valuable network
connections
– Understand the learner’s profile
– Our context: European research project -
Language Technologies for Lifelong Learning
38. Semantic social web - our vision
• Friend filtered social search - the “friends” are
the tutors and learning peers
• The tutor is (should be) well connected, the
learner might get introduced to valuable persons
• The learner should TRUST the results because
they come from tutors
• If the tutor invests time in adding resources to
the social web - he gains time afterward because
he has to answer less questions
39. Semantic social web - our vision
• Data sources
– Delicious.com - bookmarks
– Slideshare.net - presentations
– Flickr.com - images
– youtube.com - videos
• To be added:
– Twitter
– Facebook
– Blogs