View
25
Download
0
Category
Tags:
Preview:
DESCRIPTION
Collaborative tagging for GO. Domenico Gendarmi Department of Informatics University of Bari. Outline. Semantic Web Web 2.0 Collaborative Tagging An hybrid approach Current case study: digital libraries Potential case study: GO. The Semantic Web three-layers architecture. - PowerPoint PPT Presentation
Citation preview
Collaborative tagging for GO
Domenico GendarmiDepartment of Informatics
University of Bari
Outline
Semantic Web Web 2.0 Collaborative Tagging An hybrid approach Current case study: digital libraries Potential case study: GO
The Semantic Web three-layers architecture Sharing a common
understanding is a key reason for using ontologies
Creating and maintaining knowledge is a human-intensive activity
Community LayerCommunity Layer
Semantic LayerSemantic Layer
Content LayerContent Layer
Ontology issues for large-scale knowledge-sharing Lack of consensus
Formal representations of a specific domain imposed by an authority rather than based on shared understanding among users
Low dynamicity Knowledge drift asks for reactive changes to
ontologies High entry barriers
Ontology maintenance requires technical skills in knowledge engineering
Web 2.0 principles
Openess User generated metadata
Interaction Rich and interactive user interfaces
Community/Collaboration Social networks
The Web as “the global platform” Sharing of services & data
Collaborative Tagging systems Tags as user-generated
metadata Also known as
folksonomies = folk + taxonomies
The creation of metadata is shifted from an individual professional activity to a collective endeavor
Tag
ResourceUser
What’s new? Collaboration
You can tag items owned by others Instant feedback
All items with the same tag All tags for the same item
Communication through shared metadata Tight feedback loop Negotiation about the meaning of the terms
You could adapt your tags to the group norm Never forced
A formal model of collaborative tagging systems Tripartite 3-uniform hypergraph
N U T R E {(u,t,r) | uU, tT, rR)}
F (N,E)
U1 T1 R1
U2 T2 R2
U3 T3 R3
U1
T2
R3
U2 R2
T3
T1 R1
U3
<?xml-stylesheet type="text/xml" href="http://www.w3.org/2004/03/trix/all.xsl"?> <TriX xmlns="http://www.w3.org/2004/03/trix/trix1/" xmlns:u="http://example.com/userentity/"> xmlns:t="http://example.com/tagentity/"> xmlns:r="http://example.com/resourceentity/"> <graph> <uri>http://example.org/folksonomy</uri> <triple> <qname>u:U1</qname> <qname>t:T2</qname> <qname>r:R3</qname> </triple> <triple> <qname>u:U2</qname> <qname>t:T3</qname> <qname>r:R2</qname> </triple> <triple> <qname>u:U3</qname> <qname>t:T1</qname> <qname>r:R1</qname> </triple> </graph>
<triple> <qname>u:U1</qname> <qname>t:T2</qname> <qname>r:R3</qname> </triple> <triple> <qname>u:U2</qname> <qname>t:T3</qname> <qname>r:R2</qname> </triple>
<triple> <qname>u:U3</qname> <qname>t:T1</qname> <qname>r:R1</qname> </triple>
Collaborative Tagging applications
Social Bookmarking Del.icio.us, Fuzzzy, Simpy
Social Media sharing Flickr, YouTube, Last.fm
Social reference management CiteULike, Bibsonomy, Connotea
Other… Anobii, Library Thing, 43 things, …
del.icio.us: popular bookmarks and tags
del.icio.us: bookmark details
del.icio.us: saving a bookmark
Collaborative tagging trade-off
Benefits Reflects user vocabulary Sensitive to knowledge
drift Creates a strong sense
of community Emerging consensus
Limits Synonymy Polysemy Basic level variation Low precision & recall
Our vision
A community of users which collaborate for collectively evolving an initial knowledge structure (lightweight ontology) Help users in the organization of personal information
spaces Bring together different contributions to reflect the
community common ground
Proposed approach: 3-step iteration Users select
information they are interested into
Users organize their personal personal information spacesinformation spaces
Individual contributions are grouped to create shared information shared information spacesspaces
Step 1: Selection
Step 2: Organization Choose binder
name Browse space of
metadata Select metadata Update personal
taxonomy
B1
c1
c4
c3
Bncx
cz
cy… …
Personal Information Space
Personal Taxonomy
User Profile
Topic a
…
…
Topic k
Step 3: Sharing
Share personal binders
Browse shared information spaces
Express preferences on shared taxonomies
Gene Ontology Context
GO can be used for the annotations of a large amount of gene products
Two relationship types is-a part-of
Roles Curators Annotators
Three-step iteration applied to GO
Step 1: Selection Using existing tools for browsing GO (i.e. AmiGO) scientists
could select genes/gene products they are interested into
Step 2: Organization Scientists could create and organize their own private working
space where to annotate the selected genes with GO terms (existing or new ones)
Step 3: Sharing Sharing personal information about gene products among
people or groups with similar research interests could evolve the knowledge about selected genes by many individuals
Claims of verify
Personal information spaces could help scientists in laboratories to organize their own knowledge on gene products using their favourite terms, descriptions and annotations
Knowledge sharing among scientists with similar interests could create a feedback loop like in folksonomies
The GO could significantly benefit from this combination of ‘quasi uncontrolled’ knowledge spaces of scientists in the laboratories and a central organized knowledge structure
References F. Abbattista, F. Calefato, D. Gendarmi and F. Lanubile,
Shaping personal information spaces from collaborative tagging systems, KES 2007/ WIRN 2007, Part III, LNAI 4694, pp. 728–735, 2007.
D. Gendarmi, F. Abbattista and F. Lanubile, Fostering knowledge evolution through community-based participation, Proc. of the Workshop on Social and Collaborative Construction of Structured Knowledge (CKC 2007), at the 16th International World Wide Web Conference (WWW 2007).
D. Gendarmi and F. Lanubile, Community-Driven Ontology Evolution Based on Folksonomies, OTM Workshops 2006, LNCS 4277, pp. 181–188, 2006.
Acknowledgments
Thank you to: Prof. Filippo Lanubile Dr. Andreas Gisel
Contact: Domenico Gendarmi
University of Bari, Dipartimento di InformaticaCollaborative Development Group http://cdg.di.uniba.it/
Recommended