Typing OpenDLib repository service: strengths of an information object type language

摘要

End-users perceive a Digital Library (DL) as a set of Functionality ( FDL ) operating over the digital objects of an Information Space ( IDL ). At this level of abstraction, objects represent entities of the end-users application domain; for example, conference proceedings, i.e. collections of article objects. The structure of an IDL is typically given in terms of high-level modeling primitives, e.g. a combination of classes of objects, each class describing domain-specific structure, namely properties, constraints, and behavior, of the objects it contains. DL designers main task is the definition of an appropriate structure for the IDL.End-users perceive a Digital Library (DL) as a set of Functionality operating over the digital objects of an Information Space. The structure of a DL information Space is typically given in terms of high-level modeling primitives, e.g. a combination of classes of objects, each class describing domain-specific structure, namely properties, constraints, and behavior, of the objects it contains. DL designers usually realize DL Functionality and Information Space as software components extending and/or customizing the functionality supported by a DL Repository Service (RS) that maintains a Repository Information Space. Intuitively, DL developers would implement functionality in terms of RS primitives and represent objects of the DL Information Space with persistent objects in the Repository Information Space. Unfortunately, unlike DL Information Spaces, Repository Information Spaces are generally not organized into sets of objects with the same user-defined structure. As a consequence, DL developers "emulate" DL Information Space abstractions outside the boundaries of the Repository Information Space, by "embedding" high-level modeling primitives into the engineering of components, and rely on the Repository Information Space only for storage issues. The problem of such scenario is that the Repository Information Space alone is "unaware" of its "real" content. This paper presents the experience of the OpenDLib project in experimenting the realization of a new RS based on a rich and flexible document model - T-DoMDL - as a step toward the realization of RS supporting the creation of domain-specific Repository Information Spaces, directly matching designersu27 intuition of DL Information Spaces.