Knowlege organization systems and linked data: meaning and identifiers

Ronald Siebes, DANS KNAW (Netherlands)
Andrea Scharnhorst, DANS KNAW (Netherlands)

Abstract | In this talk we discuss the characteristics of Linked Data-based resource discovery and its limitations in finding content by direct processing of information resources — in comparison to the solution provided through content metadata supported by knowledge organization systems (KOS) such as thesauri, classification and subject descriptors. KOS are traditional information discovery tools that determine the meaning and control ambiguities of language, hence they are often referred to as controlled vocabularies. They are used by libraries as well as by publishers and bookshops. However, most of KOSs are designed for and used in traditional information environments and are often not readily accessible by programs. Semantic technologies such as linked data offer solutions for expressing KOS in a more formalized and machine-understandable way. They provide a way of uniquely identifying and contextualising semantically meaningful units irrespective of their possible linguistic or symbolic representations. This “unique identification” (URI) is the key element of linked data technology: anything that can be identified can be linked. The publishing of KOSs as linked data has become the most important form of sharing and using controlled vocabularies in the Web environment. This is also a solution for accessing the meaning and knowledge stored in the collections indexed by KOS (both directly or indirectly). As more and more KOSs are being published as linked data and more and more collection metadata containing KOS concepts join the linked data cloud, some obstacles to linking collection metadata and KOSs have become more obvious. Human knowledge is in constant flux and KOSs develop over time to embrace new terminology and new fields of knowledge. These changes affect unique identifiers used in KOS and consequently all links between KOS and resource collections. In this talk we will draw attention to the importance of the use of identifiers in the form of URIs as placeholders for a shared conceptualization of meanings referring to KOS concepts. How to choose them? Should they be resolvable? Should they be future proof? Who is the authority behind the namespace? What is the relationships between changes in a KOS and a URI? How to bridge/map concept identifiers used by various KOS? And how to deal with updates in the KOS?

Bio | Ronald Siebes is a senior researcher at DANS and the TU-Delft where he works on everything related to Linked Data in the domain of Humanities. Ronald got his Ph.D. in  the field of artificial intelligence at the Vrije Universiteit Amsterdam and worked in a number of research projects such as OpenPHACTS, BigDataEurope, ArchiMedial and MultiMediaN. At DANS Ronald works issues related to knowledge organisation systems and their publishing in the Semantic Web domain.

Bio | Andrea Scharnhorst is Head of Research at Data Archiving and Networked Services (DANS), an institute of the Royal Netherlands Academy of Arts and Sciences (KNAW) and of the Netherlands Organization for Scientific Research (NWO). She has a background in Physics (Diploma in Statistical Physics) and in Philosophy of Science (PhD). She has published extensively in international peer reviewed journals about topics of knowledge dynamics. Andrea is editorial board of the journal Scientometrics and the Data Science journal. Since 2016 she has also served as the Chief Integration Officer for DARIAH.

Event Timeslots (1)

Day 2 | Friday, June 21
Ronald Siebes & Andrea Scharnhorst, DANS KNAW (Netherlands)

Leave a Reply

Your email address will not be published. Required fields are marked *