Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
Bridging Folksonomies and Ontologies
For Knowledge Sharing
on the Social Web
Freddy Limpens1, Fabien Gandon1, Michel Buffa2
1 : INRIA , 2 : I3S (CNRS - UNSA), Sophia-Antipolis
SoSEA 2008 First International Workshop on Social Software Engineering and Applications,
L’Aquila, Italy, September 15, 2008
2
Social Software
=
Interaction
Exchange
Sharing
3
4
5
6
7
How to organize
this wealth?
8
Social Tagging ?
9
Or mess
of tags ?
10
Objectives
1. Overcoming the limits of folksonomies
11
Objectives
1. Overcoming the limits of folksonomies
2. Semantic Web : interoperability accross Social Platforms
12
- 1 -
Organizing / Structuring
Folksonomies
13
From folksonomies towards ontologies
(Mika, 2007)
14
15
16
From folksonomies towards ontologies(Mika, 2007)
17
Co-User Graph :Overlapping of
community of interest{Uj}
Ti
{Tj}
From folksonomies towards ontologies(Mika, 2007)
18
Co-Utilization GraphCo-occurrence on Resources
{Ti}
{Tj}Ri
From folksonomies towards ontologies(Mika, 2007)
19
Co-UserGraph
From folksonomies towards ontologies(Mika, 2007)
20
Broader/Narrower relationsvia sub-communities of users
From folksonomies towards ontologies(Mika, 2007)
21
Exploiting extra-information :Mapping with Ontologies
(Specia & Motta, 2007)
22
1. Clusters of co-occuring tags
2. Selecting ontologies
3. Mapping between tags and ontologies
4. Infering semantic links
Exploiting extra-information : Mapping with Ontologies(Specia & Motta, 2007)
23
Example of a cluster (del.icio.us) :
college
commerce
corporate
course
education
high
instructing
learn
learning
lms
school
student
Exploiting extra-information : Mapping with Ontologies(Specia & Motta, 2007)
24
Exploiting extra-information : Mapping with Ontologies(Specia & Motta, 2007)
25
- 2 -
Enriching Folksonomies
26
Limits of tags
Variabilty of spelling and language
new_york = newyorkmusic = musique
Guiding Tagging with Ontologies(Passant, 2007)
27
Limits of tags
Ambiguity :
toread ?RDF =? Ressource Description Framework
RDF =? Rwanda Defense Forces
Guiding Tagging with Ontologies(Passant, 2007)
28
Limits of tags
« Flat » organization:
paris ? France ? Europe ? country
Guiding Tagging with Ontologies(Passant, 2007)
29
Guiding Tagging with Ontologies(Passant, 2007)
Raising ambiguity thanks to ontologies
30
Raising ambiguity thanks to ontologies
1 tag
Guiding Tagging with Ontologies(Passant, 2007)
31
1 tag
2 concepts
Guiding Tagging with Ontologies(Passant, 2007)
Raising ambiguity thanks to ontologies
32
Keeping track
of Tagging Info
33
Keeping track of Tagging Info
1 tag = 1 object (Gruber, 2005)
1. Tagged object2. String of character3. User4. Domain of use5. Time
34
Semantic Wikis
Sweet Wiki (Buffa et al, 2008)
35
Sweet Wiki (Buffa et al, 2008)
Editing and Tagging wiki pages
36
Sweet Wiki (Buffa et al, 2008)
1 new tag = 1 new concept
37
Sweet Wiki (Buffa et al, 2008)
1 new tag = 1 new concept
Conceptual scheme iscollaboratively editable
38
- 3 -
Ontologies to
Interlink Social Softwares
39
SIOC (Breslin et al, 2005)
Semantically InterlinkedOnline Communities :
40
SIOC (Breslin et al, 2005)
Semantically InterlinkedOnline Communities :
* Keeping track of social interactions
41
SIOC (Breslin et al, 2005)
Semantically InterlinkedOnline Communities :
* Keeping track of social interactions
* Automatically generate semantic metadata
42
43
Good practices :
« Web of Linked Data »(Berners Lee, 2006)
44
Good practices :
« Web of Linked Data »(Berners Lee, 2006)
1. Http URIs to name
3. Provide extra info on each URIs
4. Linking URIs to other URIs
45
46
47
Revyu.com : sharing reviews on ... anything
(Heath & Motta, 2007)
48
Anything that has a name can be reviewed
49
50
Revyu.com (Heath & Motta, 2007)
Deriving semantics with extra source of info
ex : tag 'book'
=> Check ISBN (Amazon.com ou autre)
=> add rdf:type #Book
51
Conclusions
-
Perspectives
52
Objectives
1. Overcoming the limits of folksonomies
53
Organizing folksonomies
A. Analyzing Co-occurrence patterns(Mika, 2005)
>> How to augment precision of the links?
54
B. Exploiting extra-information(Specia & Motta, 2007)
>> How to select relevant sources ?
Organizing folksonomies
55
Objectives
1. Overcoming the limits of folksonomies
2. Semantic Web : interoperability accross Social Platforms
56
Guiding tagging with ontologies (Passant 2007)
>> How to ''naturally'' integrate structured knowledge ?
>> Incentives to contribute ?
57
Tags as ''shareable'' objects(Gruber 2005, Passant 2008)
>> 1 tag = 1 URI ?
>> 1 tag = 1 wiki page ?
>> 1 tag = several share-ablemeanings ?
58
Towards Inter-operability(SIOC, Web of Linked Data)
>> Developping ''viral'' plugins (SIOC)
>> Combining Social applications ?Bookmark + WikiPedia ?
59
Thank you !
60
Perspectives
>> Methods to select precisely the social data used for inference
>> Getting richer info about tagging
>> Exploiting acquaintance networks
>> Tools of monitoring/tunning of automatic treatments
61
Keeping track of Tagging Info
1 tag = several meaning (Passant, 2008)
MOAT : Meaning Of A Tag
62
63
64
65
66
C. Getting Users to contribute
>> Games with a Purpose ?espgame.org
ontogame.org
Organizing folksonomies
67
Tagging the Tags
(Tanasescu & Streibel, 2007)
68
Tagging the Tags (Tanasescu & Streibel, 2007)
69
Tagging the Tags (Tanasescu & Streibel, 2007)