The Open Archives Initiative—Object Reuse and Exchange has released Compound Information Objects: An OAI-ORE Perspective by Carl Lagoze and Herbert Van de Sompel.
Here’s an excerpt from the document’s "Introduction and Motivation" section:
In summary, the web architecture expresses the notion of linked URI-identified resources. Information systems can leverage this architecture to publish the components of a compound object and thereby make them available to web clients and services. But due to the absence of commonly accepted standards, the notion of an identified compound object with a distinct boundary and typed relationships among its component resources is lost.
The absence of these standards affects the functionality of a number of existing and possible web services and applications. Crawler-based search engines might be more useful if the granularity of their result sets corresponded to compound objects (a book or chapter, in this example) rather than individual resources (single pages). The ranking algorithms of these search engines might improve if the links among the components of a compound object were treated differently than links to the object as a whole, or if the number of in-links to the various component resources was accumulated to the level of the compound object instead of counted separately. Citation analysis systems would also benefit from a mechanism for citing the compound object itself, rather than arbitrary parts of the object. Finally, a standard for representing compound objects might enable a new class of "whole object" services such as "preserve a compound object".