Unifying the concept of collection in digital libraries

The notion of collection plays a key role in Digital Libraries, where several kinds of collections are typically found. We claim that all these kinds can be unified into a single abstraction mechanism, endowed with an extension and an intension, similarly to predicates in logic. The extension of a collection is the set of documents that are members of the collection at a given point in time, while the intension is a description of the meaning of the collection, that is the peculiar property that the members of the collection possess and that distinguishes the collection from other collections. The problem then arises how to automatically derive the intension from a given extension, a problem that must be solved e.g. for the creation of a collection from a set of documents. It turns out that our notion of collection is very close to the notion of formal concept in Formal Concept Analysis, which provides a well-founded framework to formalize the problem and very useful tools to solve it. We exploit this framework to study the problem of automatically deriving a collection intension from a given extension.We then show how intensions can be exploited for carrying out basic tasks on collections, establishing a connection between Digital Library management and data integration.