Vocabularies: Term Lists and Thesauri
Karen L. Coyle · Library Technology Reports · 2012
Abstract Chapter 4 covers controlled vocabularies, which in linked data are made public on the Web. This allows for machine-checking on validity and also allows communities to describe the meaning of their terms and harmonize them with those of other communities. This chapter describes controlled term lists, subject lists, and thesauri from library and nonlibrary communities. In the previous chapter we saw that there are data elements available for use covering many things you might want to describe in metadata. Reusing existing data elements is one way to assure that your data will find links on the linked data Web. Another way to increase the meaning of the links is to control the content by using controlled vocabularies as the content of your metadata statements. Controlled vocabularies in linked data are published--that is, they are made public on the Web. This allows for machine-checking on validity and also gives communities an opportunity to describe the meaning of their terms and harmonize their terms with those of other communities. Let's use the concept of color as an example. If two communities are using the same data element for color but do not share a controlled list of terms, they have no way to know if plum and are more or less the same color. If the plum people and the people want to share data, they can each create a controlled vocabulary for their terms, and within these vocabularies they can, using Semantic Web standards, say, Plum is a close match for aubergine and Aubergine is a close match for plum. If they have followed the recommended practice of linking to broader vocabularies as well, they could have each linked their color vocabulary to a vocabulary of common colors (red, purple, blue, yellow, green, orange)--Plum is a kind of and Aubergine is a kind of purple. In any situation where data is being shared, there is now sufficient information for applications to be written that can use plum, aabergine, and purple interchangeably, if desired. Apart from the advantages of sharing, controlled vocabularies allow verification at the time of input and receipt of metadata. If the data value must be a member of a list, the input program can verify that whatever is provided meets that rule. Drop-down lists for selection can also aid input so that the person doing input does not have to remember or look up valid values. Term lists are usually experienced by human users as words from natural language. Having lists that are actually made up of natural language terms, like red or blue, is not entirely useful on the linked data Web, where one may be sharing data globally. Wherever possible, linked data terms lists use identifiers for the terms. This is not terribly different from the use of codes in the fixed fields in MARC records, although the linked data codes are full URIs. With the members of the list identified in a language-neutral way, user displays can be developed for any desired language. As discussed in chapter 3 in the section on SKOS, for each language, one can define a primary display term and any number of alternate and hidden terms: ex:color rdf:type skos:Concept; skos:prefLabel red@en; skos:altLabel ruby@en; skos:hiddenLabel reddish@en; skos:prefLabel rouge@fr; skos:prefLabel rosso@it; skos:altLabel ciliegia@it. Controlled lists can be as simple as a single-level list of terms (red, green, blue), or they can be structured with broader and narrower concepts, like a taxonomy. ex:dogs rdf:type skos:Concept; skos:prefLabel dogs@en; skos:narrower ex:working_dogs; skos:broader ex:animals. The color example is, by the way, not at all farfetched. While the artistic and social perception of colors is hard to define, as this example shows, colors in the computer technology world are easily defined using the RGB coordinate method. …