Data vocabularies: why are they important?

Each individual, organization or group uses a number of common words in their daily communication, which would be their personal vocabularies. The greater the number of words we use, the greater our ability to communicate, but it can also be a barrier, increasing the difficulty of understanding other people who are not familiar with the same terms as we are. The same is true in the world of data.

To avoid misunderstandings, we must use a controlled vocabulary, which is nothing more than a defined list of terms to systematically organize, categorize or label information.

What is a data vocabulary?

For a successful implementation of a data governance initiative, whether private or public, we must provide the process with a business Glosary or reference vocabulary. A reference vocabulary is a means of sharing information and developing and documenting standard data definitions to reduce ambiguity and improve communication. These definitions should be clear, rigorous in their wording and explain any exceptions, synonyms or variants. A clear example is EuroVoc, which covers the activities of the European Union and, in particular, the European Parliament.  Another example is ICD10, which is the Coding Manual for Diagnoses and Procedures in Health Care.

The objectives of a controlled vocabulary are as follows:

• To enable common understanding of key concepts and terminology, in a precise manner.
• To reduce the risk of data being misused due to inconsistent understanding of concepts.
• Maximize searchability, facilitating access to documented knowledge.
• Drive interoperability and data reuse, which is critical in the open data world.

Vocabularies vary in the complexity of their development, from simple lists or selection lists, to synonym rings, taxonomies or the most complex, thesauri and ontologies.

How is a vocabulary created?

When creating a vocabulary, it does not usually start from scratch, but is based on pre-existing ontologies and vocabularies, which favors communication between people, intelligent agents and systems. For example, Aragón Open Data has developed an ontology called Interoperable Information Structure of Aragón EI2A that homogenizes structures, vocabularies and characteristics, through the representation of entities, properties and relationships, to fight against the diversity and heterogeneity of existing data in the Aragonese Administration (data from local entities that do not always mean the same thing). For this purpose, it is based on proposals such as RDF Schema (a general vocabulary for modeling RDF schemas that is used in the creation of other Vocabularies), ISA Programme Person Core Vocabulary (aimed at describing persons) or OWL-Time (describing temporal concepts).

A vocabulary must be accompanied by a data dictionary, which is where the data is described in business terms and includes other information needed to use the data, such as details of its structure or security restrictions. Because vocabularies evolve over time, they require evolutionary maintenance. As an example, ANSI/NIZO Z39.19-2005 is a standard that provides guidelines for the construction, formatting and management of controlled vocabularies. We also find SKOS, a W3C initiative that provides a model for representing the basic structure and content of conceptual schemas in any type of controlled vocabulary.

Examples of Vocabularies in specific fields created in Spain

In the Spanish context, with a fragmented administrative structure, where each agency shares its open information individually, it is necessary to have common rules that allow us to homogenize the data, facilitating its interoperability and reuse. Fortunately, there are various proposals that help us in these tasks.

The following are examples of vocabularies created in our country for two fundamental sectors for the future of society, such as education and smart cities.

Smart cities

An example of the construction of domain-specific vocabularies can be found in ciudades-abiertas.es, which is an initiative of several city councils in Spain (A Coruña, Madrid, Santiago de Compostela and Zaragoza) and Red.es.

Among other actions, within the framework of the project, they are working on the development of a catalog of well-defined and documented vocabularies, with examples of use and available in several representation languages. Specifically, 11 vocabularies are being developed corresponding to a series of datasets selected by the municipalities that do not have a defined standard. An example of these vocabularies is the Municipal Agenda.

These vocabularies are generated using the OWL standard language, which is the acronym for Web Ontology Language, a markup language for publishing and sharing data using ontologies on the Web. The corresponding contexts for JSON-LD, N-triples, TTL and RDF/XML are also available. In this explanatory video we can see how these vocabularies are defined. The generated vocabularies are available in the Github repository.

Education

In the field of universities, on the other hand, we find the proposal for open data content for universities developed by the UniversiDATA community: Common Core. In version 1.0, 42 datasets have been identified that every university should publish, such as information related to Degrees, Enrolments or Tenders and contracts. At the moment there are 11 available, while the rest are in the process of elaboration.

For example, the UAM (Autónoma de Madrid), the URJC (Rey Juan Carlos) and the UCM (Complutense de Madrid) have published their degrees following the same vocabulary.

Although much progress has been made in the creation and application of data vocabularies in general, there is still room for progress in the field of research on controlled vocabularies for publishing and querying data on the Web, for example, in the construction of Business Glosaries linked to technical data dictionaries. The application of best practices and the creation of vocabularies for the representation of metadata describing the content, structure, provenance, quality and use of datasets will help to define more precisely the characteristics that should be incorporated into Web data publishing platforms.

Content elaborated by David Puig, Graduate in Information and Documentation and responsible for the Master Data and Reference Group at DAMA ESPAÑA.

Contents and points of view expressed in this publication are the exclusive responsibility of its author.