From Athenawiki

Jump to: navigation, search

Contents

Make your terminology interoperable

Error: must specify an image in the first line

After having conceived your terminology the closer to the ideal form you could, you have to make it interoperable. The conception step aimed at producing a thesaurus from the internal point of view of the museum in a human perspective. Now we recommend you to metaphorically go out of the museum, and to take into account the machine concern by SKOSifying your thesaurus.


Benefits from using SKOS

RDFS and OWL are the languages that have been formally defined for knowledge representation. SKOS is one language among this formal languages' family. The major difference is that SKOS has been designed to model every type of controlled vocabulary. It can be used to represent a thesaurus as well as a classification or a subject headings list. Then it is a good compromise for the institutions who are using these types of resources, and who are willing to be compliant with the Semantic Web technologies without developing sophisticated ontologies.

The SKOS data model is consistent with the formal ontology language OWL. Therefore the migration from a SKOS version of a terminology towards a formal ontology in OWL can be handled without major difficulties.

Since the SKOS model is very simple, but still complete enough, the implementation of a SKOS version has a low cost for migration. As we made the distinction in the D4.2 introduction, SKOS is not a formal knowledge representation. But for an institution managing simple list of terms, or classifications and thesauri in the best case, it would be extremely costly and time consuming to develop a formal ontology perfectly compliant with Semantic Web technologies (using OWL for example). Therefore SKOS provides a structure based on classes and properties which give a powerful data model for migrating and porting these terminologies towards Semantic Web technologies.

Institutions must keep in mind that the adoption of the SKOS model is not a total replacement of the data model in use in the institution but a format for publishing and reusing their terminology and for ensuring the portability of this terminology for a semantic interoperability. Indeed usually knowledge organization systems (KOS), e.g. controlled vocabularies and thesauri, are used for indexing, and then porting these KOS into SKOS would enable the use of these indexing KOS for retrieval as well.

However SKOS may not be the appropriate language for every type of controlled vocabulary. For instance, authority files which usually provide a list of persons cannot be migrated to a SKOS version properly since the scope of this type of terminology is real persons and not concepts. Another point is that the SKOS semantic relations properties cannot really apply to authority files since a person cannot be related to another one with hierarchical (narrower/broader) or associative (related) links.

If we would like to sum up all the reasons for you to use SKOS as a format for expressing your descriptions, we would remind:

  • First SKOS is particularly well adapted to multilingual terminologies.
  • Then SKOSification is an economical way to get to the conceptual level without employing an ontology. You can benefit from SKOS by migrating your thesaurus to a simili-ontology with a minimum of time and financial costs.
  • Finally it is important to consider that SKOS is evolving and it will be easier and easier to customize it thanks to new SKOS classes you will be able to define by your own or the ones that will be proposed with the evolution of this standard.

Anyway, if you are considering to SKOSify your thesaurus, you have to change a bit your perspective. So far you were dealing with terms.Now you have to manage concepts since the Semantic Web in a multilingualism perspective requires concepts than terms to exploit. You also need to be distanced from your professional framework and make explicit a knowledge that you can easily infer considering your professional background.


Methodology

In fact when you SKOSify your thesaurus, you are technically applying the connection you have defined previously at step A3: Define your connection with the datamodel. SKOS is the format we advice you to use among those which fit today with Europeana requirements for the ingestion and the exploitation of your digital resources. As a format it enables the mapping of terminology models. And in our case the two terminology models are:

  • Your own terminology model (that you have defined in step A5: Organise your terms into a thesaurus structure)
  • The official terminology model designed in your datamodel (in LIDO if you have chosen it)

So this second big step is about the conversion of your thesaurus into a SKOS version. To complete the implementation, a set of actions can be done . Most of them concern the mapping of terminology models. As it has been done for the first step, we present them as a linear process for reasons of readability. But of course, these tasks are iterative as well.

Image:WP4_D4.3_ensemble_inter_format_termi_models.jpg



The different tasks we are going to detail are:

  • B1: Evaluate how far SKOS is compliant with your terminology features
  • B2: Roughly SKOSify your terminology
  • B3: Define with precision the labels expressing concepts
  • B4: Identify your concepts and validate the structure
  • B5: Ensure the documentation of concepts
  • B6: Map your concepts
  • B7: Map your (multilingual) terms
  • B8: Validate your SKOSification

Navigation

We invite you to pursue the step by step process by going to the next step: B1: evaluate how far SKOS is compliant with your terminology features.

Error: must specify an image in the first line

You can also navigate through the recommendations by using the synoptic map below. This map will be available on each page of the recommendations process. In order to know the name of a step in particular, just rollover and stay a bit on the very box so that the name appears.

Error: must specify an image in the first line

This page was last modified on 4 June 2011, at 10:15.This page has been accessed 6,721 times.