Date post: | 27-Jan-2015 |
Category: |
Education |
Upload: | aravind-sesagiri-raamkumar |
View: | 129 times |
Download: | 3 times |
Taxonomies and Metadata
Agenda
• Introduction to Metadata and Taxonomy• Folksonomies• Ontologies• Metadata and Taxonomy combined• Taxonomy Development• Software and Tools• Current Challenges
What is Metadata?Metadata is structured information that describes, explains, locates, or otherwise makes it easier to retrieve, use, or manage an information resource[NISO]
Title
Author(s)Year of publication
Metadata Types
• Descriptive -> for resource discovery and identification
• Structural -> defines the physical/logical structure of resources
• Administrative -> for managing resources
“Metadata is simply data about data”
Purposes of Metadata
Additionally…• Facilitate interoperability between systems• For Archiving and Preservation
Retrieval
Resource
Discovery &
IdentificationManagement Classification
Connect with
other resources
Authorship &
Access Rights
Evolution of global metadata standards…Metadata Scheme – set of metadata elements designed for a particular purpose
Metadata Specification – when metadata scheme is adopted by many other organizations
Metadata Standards – when metadata specification is accepted by a ‘standards’ body such as ISO
“Metadata Standards are required at a global level mainly
for enforcing Interoperability between systems”
Popular Metadata Standards
What are Taxonomies• In KM perspective, taxonomy is a hierarchical topic structure where
information items are assigned through the dual processes of classification (filing to a location) and categorization (tagging with corresponding metadata) [centralized taxo]
• Taxonomies facilitate discovery (browsing & searching), retrieval and content re-use of resources within a system
“Taxonomies are hierarchical classification systems”
Where are they?
Most commonly used taxonomy
Taxonomy and Knowledge Organisation Systems (KOS)
• In the Information Science domain, Taxonomies are a type of Knowledge Organisation Systems (KOS) which are meant to model the underlying semantic structure of a domain [Hodge]
• Among KOS types, taxonomies are somewhere in the “middle” in terms of creation/maintenance complexity and expressive power
http://www.slideshare.net/TriviumRLG/from-taxonomies-
to-ontologies
Structured KOS and their applicability
Type Directionality Description Applicability
Taxonomy Groups resources into categories
For creating simple classification
schemes
ThesaurusCaptures different
names of resources and finds close relationships
For creating classification
schemes along with associative
relationships
Ontology
Captures multi-dimensional
relationships b/w both within and
between groups of resources
For maintaining a network of resources
with multiple relationships and
properties
Folksonomies – Web 2.0 based alternative to Taxonomies
• A new breed of web 2.0 resource sharing systems allow users to add their own keywords(or tags) to resources
• Tags used for both resource description & classification and for later retrieval• Outcome of tagging activity in a systems => Folksonomy• Folksonomies are the most dynamic KOS system• Two types :
– Broad folksonomies: Anyone can add any resource and tag any resource– Narrow folksonomies: The author adds the resources and adds the tags while other users are restricted in adding tags
• Popular systems: Flickr (Image sharing system), Delicious (Social bookmarking system)
VS
Taxonomy created with Experts Folksonomy developed through users
Professional touch
Highly compliant with historical resources
Rigid
Dependent on experts
People power
Highly compliant with current resources
Volatile
Takes time for vocabulary convergence
Spelling mistakes
Spams
Why Folksonomies ?
Leveraging both Taxonomies and Folksonomies
1. Start with a controlled vocabulary created by experts2. Create the taxonomies based on the controlled vocabulary3. Provide the users with the feature to add tags to the resources in the
system4. Monitor tagging activity and tag convergence for resources5. Modify the controlled vocabulary to include the popular tags thereby
modifying the taxonomy too
Expert touch + People choice = Relevant Taxonomies (Controlled (Tags) Vocabs)
Ontologies – most advanced KOS type• What are Ontologies?
– A networked collection of concepts and their corresponding properties and relationships in a particular knowledge domain
• Support for all different properties– Transitive– Symmetrical– Functional & Inverse Functional
• The biggest benefit of ontology is its inferencing abilityCan Taxonomies and Ontologies co-exist?
• Both ontologies and taxonomies can be built from each other
• The relationship between components in a taxonomy is implicitly understood by users
• The relationship between components in a ontology is explicitly specified and can be understood by
semantic systems
• In reality, ontology subsumes taxonomy and therefore taxonomy can be built from ontology without
any loss of data
More on Ontology…• Ontology is the central binding component of the proposed “Semantic Web”
architecture• Semantic Web represents the next generation web of data where systems understand data• Semantic Web technologies such as RDF, OWL and SPARQL are already used in many websites• Anyone can design an ontology using the Web Ontology Language (OWL) or
Resource Description Framework (RDF) and publish in the web• Simple Knowledge Organisation System (SKOS) is an vocabulary that can be
used by organisations to express their taxonomies, thesauri and other classification schemes
More on SKOS and KM…
• Use SKOS type ontologies in your company if you are interested in using semantic technologies
• Semantic technologies aid the “Linked Data” vision where the aim is to connect data in one organisation to data from
other organisations to facilitate re-use and better understanding
• Caveat: These technologies have not reached mainstream adoption yet
SKOS is able to express both taxonomy relationships (broader/narrow)
and thesaurus relationships (preferred label)
Taxonomy and other KOS systems – a summary• Taxonomies are not just a set of folders
• They are an entry point to the pool of resources (documents)
• They are built on top of controlled vocabularies
• Taxonomies can be built through expert analysis
• Folksonomies make use of the public vocabulary for providing continual updates to taxonomies
• Ontologies help in re-using concepts and applying semantics to the concepts
• Web based ontologies help in inter-operability across other systems
Why
Metadata + Taxonomy ?
Complimentary relationship of Metadata and Taxonomy
• Metadata describes a resource well and is very much part of the resource• Metadata doesn’t capture relationships between resources sufficiently -> this is
where taxonomies come in• Taxonomies are external to the resource and are good for modelling relationships
between resources• Taxonomies are road-maps and serve dual purposes of describing current
resources and also predicting where the future resources will be placed
Metadata
Taxonomy
Data about items
Classification
&
Labeling
Finding resources
Helping in decision making by providing a pool of
resources with their corresponding information
Visualizing the integrated working mechanism of metadata and taxonomies
Document, Content
& Records
Management
Thesauri
Ontologies
Filing & Storage
Resource Metadata
&
Tagging
Search
Engine
Visualisation
Resource
Navigation
Intranet / Portal
User Interface
Back End
Components
Front End
Components
Taxonomies
Knowledge Organisation
Systems
[Centralized taxonomy]
What are the indications of a good taxonomy?
• Taxonomy vocabularies need to be understandable and meaningful to common users
• The users should be able to get an overall idea of the structure of the domain by looking at the taxonomy
• The resources are to be easily located in taxonomies with smaller paths• The users should also be able to anticipate where new resources would be
placed• Most importantly, taxonomies should be easy to navigate
Taxonomy Development• Taxonomies are essentially “living organisms” with dynamic
nature -> continually evolving over a period of time• One-time development followed by periodic updates is the
norm with taxonomy management
Whittaker’s seven steps of taxonomy development
Determine
RequirementsIdentify Concepts
Develop draft
taxonomy
Review with Users
and SMEsRefine taxonomy
Apply taxonomy to
content
Manage and
maintain
taxonomy
Other Approaches to Taxonomy Development
Ovum’s approach• Start with a knowledge/information audit
– Study of the requirements• Build on top of existing taxonomies and categorisation models
– Use internal draft taxonomies or adopt from other companies• Develop a draft taxonomy
– By making use of categorisation tools• Refining the taxonomy
– To ensure navigability and logical correctness• Testing
– Piloting with few users to iron out the defects• Applying the classification model
– Bring in the documents• Monitoring
Challenges related to Taxonomy Development and Management
• There is not just ‘one’ correct taxonomy for the entire organization
• Development from scratch vs. Adapting someone else’s• Taxonomy creation at start or end of information lifecycle• User-oriented or content-oriented taxonomies• Document-centric or people-centric taxonomies• Taxonomy integration
Popular Software
Software and Tools• Synaptica – Commercial taxonomy building software• Poolparty – Thesaurus management software with SKOS editor• MultiTes Pro – Thesaurus building software• Protégé – Free ontology building software• TopBraid Composer – Ontology editing software• Microsoft Sharepoint – Most popular content and document management
platform with enterprise search
ReferencesAcademic ReferencesWhittaker, M., & Breininger, K. (2008, August). Taxonomy development for knowledge management. In 74th General Conference and Council of the World Library and Information, Quebec, Canada.Woods, E. (2004). Building a corporate taxonomy: Benefits and challenges.Ovum expert advice.General Web ReferenceHodge, G. (2013, June 18). Taxonomies and ontologies: definitions, differences and use. Retrieved from http://info.nfais.org/info/Hodge_Post.pdfLei Zeng, M. (2004). Metadata standards. Retrieved from http://marciazeng.slis.kent.edu/metadatabasics/standards.htmNISO. ANSI, (2004). Understanding metadata. Retrieved from website: www.niso.org/standards/resources/UnderstandingMetadata.pdfTen taxonomy myths. (2002, November). Retrieved from http://www.montague.com/review/myths.htmlSlideshare ReferencesBarbosa, D. (2008, September 29). Centralized taxonomy management for enterprise information systems. Retrieved from http://www.slideshare.net/danielabarbosa/centralized-taxonomy-management-for-enterprise-information-systems-presentationChampeau, D. (2009, November 24). Taxonomy and metadata. Retrieved from http://www.slideshare.net/dchampeau/taxonomy-and-metadataConnors, C. (2010, January 21). From taxonomies to ontologies. Retrieved from http://www.slideshare.net/TriviumRLG/from-taxonomies-to-ontologiesCooksey, D. (2008, April 08). Taxonomy is user experience. Retrieved from http://www.slideshare.net/saturdave/taxonomy-is-user-experienceMetaschool Project. (2006, December 16). Retrieved from http://www.slideshare.net/metaschool/module-37-2731159White, L. (2012, May 22). Taxonomy: Do i need one. Retrieved from http://www.slideshare.net/ElemSrc/taxonomy-do-i-need-one