High-quality metadata can help users discover data and other resources in many ways. Good titles, abstracts, and keywords, along with spatial and temporal extents, go a long way with many search strategies. Adding schema.org metadata to landing pages can also help.
A second discovery strategy, more like browsing, comes into play once someone finds something interesting. This is a classic approach – discovering papers, reports, data, or software referenced from an article or linked to from a web page. Crossref is dedicated to migrating the old approach (references in papers) to the new (web links) and currently manages a database of over 100,000,000 linked resources, many from journal articles, with that goal. DataCite makes connections between resources using relatedIdentifiers that support many kinds of relationships. The ISO metadata standards connect these approaches by supporting references to associated resources in two ways: citations and identifiers.
ISO citations provide all of the information expected in classic citations: title, responsible individuals and organizations in many roles, publication dates, etc. as well as any number of URLs for citations to resources on the web. The linked resources can serve many functions, including download, information, order, etc. Having all of this information included in the metadata makes it possible to provide that information to users to help them decide if they are interested in the cited resource. These ISO citations also include an optional identifier for the cited object which also includes a namespace and a citation to the authority for the code.
Letting users know how resources are related is another important role of connections made from metadata. The ISO standards combines two types of connections: association types and initiative types, both optional. Initiative types can be used to identify campaigns, experiments, investigations, missions, projects or programs that a resource is related to. Association types are used to identify cross references, larger work citations, and other types of associations.
I have been exploring how discovery using related is supported in several metadata dialects. I will be reporting on these connections in the future.