Irene Travis

Irene Travis, Editor

Irene L. Travis, Editor

Bulletin of the Association for Information Science and Technology

We are fortunate to have, in this issue, an in-depth discussion for practitioners of the current state of linked data, particularly in the world of library, archive and museum metadata. In “Linked Data and the Charm of Weak Semantics,” guest editors Thomas Baker, an organizer of the Dublin Core Metadata Initiative, and Stuart Sutton, its director, have assembled five articles that discuss not only the definition and origin of the linked data concept, but some of technical issues surrounding its adoption and use. Primary among them are the relative merits of weak versus strong semantic definition and the problem of data typing and validity using the two current primary linked data tools, the Resource Description Framework (RDF) and the Web Ontology Language (OWL).

In knowledge organization systems (KOS), to look at the first problem, there has been a great deal of analysis since at least the beginning of the last century about different types of semantic relationships that may exist between concepts. For example, in many KOS both part/whole relationships and true set-inclusion relationships are labeled Broader Term/Narrower Term, but the latter is transitive (what is true of the broader concept is true of the narrower one) and the former is not. Similarly, the relation between the concepts Archives and Archivists is not the same as that between Archives and Preservation, but both might be represented as “Related Terms.” Distinguishing such cases has advantages such as supporting more sophisticated inference, but its cost and difficulty make widespread adoption prohibitive, particularly in the absence of strong indications that it can improve search results. In an environment increasingly focused on data sharing and re-use, such distinctions may also inhibit vocabulary integration. These trade-offs reappear, magnified, in the linked data environment, which is designed for computer manipulation and originally encouraged stronger levels of specificity to support sophisticated machine reasoning. Sutton and Baker and other authors argue for less identification to make feasible a more populated environment that contains more links that may be processed in useful ways.

Another issue discussed is validating data in the linked data environment. For example, is it a mistake if there is no value for an attribute such as Title in a document description? Is “205” a valid value for someone’s age? What are the methods, advantages and disadvantages of adding validation to linked data?

The section is written from the perspective of practice and is essential reading for anyone contemplating putting their metadata on the web in linkable form.

In our RDAP Review, Edwin Henneken of the Smithsonian Astrophysical Observatory in Cambridge, Massachusetts, focuses on preservation of and access to astronomy datasets.

Finally, on the President’s Page current ASIS&T president Sandy Hirsh reports on the ASIS&T strategic planning process, including analysis of data gathered from focus groups and posters at last year’s Annual Meeting in Seattle. Sandy encourages member participation. The schedule calls for continuing data collection and analysis through April, drafts and comment through September and a final plan for presentation at the Annual Meeting in St. Louis in November.