Domain Ontologies

The e-LICO team is developing two domain ontologies corresponding to the two pilot application areas of the project:

Kidney and Urinary Pathway Ontology (KUPO)

The Kidney and Urinary Pathway Ontology (KUPO) is a joint collaborative effort of the e-LICO project and COST Action EuroKUP. Participants in EuroKUP contribute domain knowledge while e-LICO members provide the ontology engineering support.

 

Digital Multimedia Repositories Ontology (DMRO)

The Digital Multimedia Repositories Ontology (DMRO) focus is to be on the applications in recommender systems, personalization, and adaptive faceted browsers of digital resources. Being oriented towards the abovementioned applications, the ontology being built will be an application domain ontology. Thus, we envisage re-using terms from relevant domain ontologies, and vocabularies (such as Dublin Core, FOAF) that cover the domain of DMR ontology .

Kidney and Urinary Pathway Ontology (KUPO)

The latest version of KUPO is available at http://www.e-lico.eu/public/kupo/kupo.owl. The KUPO is being used to annotate data in the Kidney and Urinary Pathway Knowledge Base (KUPKB), a demo of the ontology in use is available at  http://www.e-lico.eu/kupkb. For more information about the KUPO see:

 

 

 

 

 

When describing biological samples for investigations and the findings from those investigations, biologists working in kidney and urinary pathways need to be able to describe a broad portion of biology. The initial version of the KUPO reflects this position. We are using ontologies or portions of ontologies that describe:

We will also wish to describe pathologies and aspects of the investigations on those experiments such as those in transcriptomics and proteomics.

To this end we have chosen to use the following ontologies:

We will also be taking a look at the disease ontology and  the ontology of biomedical investigations (and related efforts).

While taking a portion of the mouse anatomy makes sense -- we want kidney and the rest of the urinary system -- it makes less sense for GO and some of the others. Whilst we want only mammalian pertinent aspects, it is more difficult to limit GO in the same way as we might limit or scope an anatomy. To some extent,  what we want is a mammalian KUP Slim. We also need to extend portions, such as adding cells pertinent to the KUP domain. Such extensions will be submitted to the appropriate ontology efforts. Other extensions will be the insertion of extensions that will promote querying.

Below is a screenshot of the KUP Ontology  in the Protégé-OWL editor developed by the University of Manchester

 

 

A tiny subtree of the 835-concept initial version of the KUP ontology:

 

KUPDB

IFrame: 
KUPDB

Digital Multimedia Repositories Ontology (DMRO)

 

DMRO download page

 

1 Purpose.

The aim is to make the ontology in a way to be useful for internal e-LICO platform experiments, and in the further step also for the use in tasks of general interest, such as the task of the second e-LICO open challenge. In this way, our goal is to exploit the Digital Multimedia Repositories Ontology (DMRO) for providing semantic annotations of the test dataset(s) (in particular Videolectures.net dataset to be provided for the second e-LICO open challenge) for the tasks of:

a) design of recommendation, and personalization solutions for digital multimedia repositories,

b) meta-learning/meta-mining on the DMER (Data Mining Experiments Repository),

c) testing semantic data mining algorithms.

2 Scope.

The ontology focus is to be on the applications in recommender systems, personalization, and adaptive faceted browsers of digital resources. Being oriented towards the above applications, the ontology being built is an application domain ontology. Thus, we envisage re-using terms from relevant domain ontologies that cover the domain of DMR ontology. The specific coverage is specified by the competency questions and terms identified. Level of ontology development will be assessed according to the nature of the use-cases to be presented as the proof of concept. The domain ontology (DMRO) must cover all the most relevant concepts of the selected domain and concepts related to e-LICO tasks in more detail.

3 Implementation language.

The ontology is to be developed in Web Ontology Language (OWL).

4 Intended End-Users.

We have identified the following intended end-users of DMRO:

User 1. Web browser user searching for multimedia content

User 2. Multimedia providers

User 3. Web site administrator

User 4. Data-miners

5 Intended Use Cases.

Use 1. Video lecture recommendation.

Use 2. Dynamic adaptation of a faceted browser interface.

Use 3. Personalizing faceted browser interface.

Use 4. Testing semantic frequent pattern discovery algorithms in the context of recommendation and personalization application scenarios (e.g. semantically described frequent clickstreams, semantic features of videos frequently watched together).

Use 5. Testing semantic subgroup discovery algorithms.

 

6. Reusing Ontology Statements

From the analysis of the competency questions on ontologies relevant for our domain and application it follows that:

Dublin Core - can be reused for modeling general metadata properties of Resources,

FOAF - for modeling the users, authors, participants, their personal data, and social aspects,

RDF Review - vocabulary for the annotations, reviews and marks,

SWRC ontology - for Events,

OBO Relation Ontology - for standard relations.

 

DMRO download page

 

Digital Multimedia Repositories Ontology is constituted by a set of modules listed below.

DMRO.owl - the main file importing all other modules.

DMRO-Resource.owl - ontology module created using DOLCE Ultralite (DUL) as upper ontology. We have used 'IOLite', extension of DUL, as an inspiration for modelling DigitalResources. We have used 'Topic' ontology design pattern (http://ontologydesignpatterns.org/wiki/Submissions:Topic) for modelling the topics. We have reused statements from DCTYPE, ORE and RO. We have also used ontology CSnCs 'Computer Science for non-Computer Scientists from Project LT4eL (http://www.lt4el.eu/)' as inspiration.

DMRO-AgentParticipantRole.owl -  ontology module created based on the ParticipantRole ontology design pattern (http://ontologydesignpatterns.org/wiki/Submissions:ParticipantRole). We have used DOLCE Ultralite as upper ontology. We have reused statements from FOAF and RO.

DMRO-Event.owl - ontology module created using DOLCE Ultralite as upper ontology. We have reused statements from SWRC, RO and DC.

DMRO-Place.owl - ontology module created based on the Place ontology design pattern (http://ontologydesignpatterns.org/wiki/Submissions:Place). We have used DOLCE Ultralite as upper ontology. We have reused statements from RO.

DMRO-Review.owl - ontology module created by reusing the RDF Review Vocabulary (http://vocab.org/review/terms.html) and using DOLCE Ultralite (DUL) as upper ontology. We reused statement from FOAF, DC and myExperiment Ontology.

DMRO-Annotation.owl - ontology module created by reusing the Viewings & Downloads module  of the myExperiment Ontology (http://rdf.myexperiment.org/ontologies/) and using DOLCE Ultralite (DUL) as upper ontology. We have reused statements from the Annotations module of the myExperiment ontology, and some concepts from the Tagging ontology design pattern (http://ontologydesignpatterns.org/wiki/Submissions:Tagging) and the RO ontology.

 

There are also available DMR knowledge resources constituting a knowledge base on Videolectures.net domain:

  • portal categories
  • automatically generated video lectures topic ontology, and
  • an RDF version of the videolectures.net dataset (released for the ECML-PKDD 2011 Discovery Challenge), based on manually created mappings from DMRO to the VL.net dataset.

 

VLNet-Categories.owl - structure of categories of Videolectures.net portal.

VLNet-OntoGen-Topics.rdf - topic ontology generated automatically by the OntoGen tool based on the corpus from Videolectures.net.

DMROKB.owl - an RDF version of the VL.net dataset based on DMRO (the main file importing a set of files with instance data). All instace files ziped (around 10.4 MB).