Summary:
In the contemporary unified information and knowledge presentation field of the
World Wide Web, there's an emerging need to represent complex structures,
comprising individual semantic entities that can possibly act autonomously.
This work, introduces a generic and expressive model for information
representation, which comprise information objects and relationships as well as
properties they may have. This way we are capable of creating complex graph
structures that can back the definition of modern, semantically rich, content
hierarchies. Additionally, a dynamic type system has been developed for
defining classes of information objects and relationships that determine, a
common set of specifications to which, instances of that type must
automatically conform, in terms of the properties they can have, relationships
they are allowed to participate in and other uniform system-wide behaviour. The
use of types, further refines our model’s expressive power and promotes
reusability. Its capability to represent information is tested against
reference use-case systems and the models they support. We, furthermore,
discuss the application of the aforementioned information model into a notional
storage architecture, that will contribute to the homogenisation of diverse
storage infrastructures and will enable the augmentation of current systems'
expressive power, to match that of the model of this work. This functionality
is exposed through a proposed general yet powerful programming interface with
which the user can handle information access and retrieval as well as data
definition. Finally, a brokerage mechanism is proposed, lying inside the
architecture, which by exploiting the creation of a common storage resource
pool, can be used to, transparently, distribute the information to the
underlying infrastructure, based on user-defined factors, such as the cost, the
available free resources or the data replication policies.
Keywords:
Information Model, Type System, Information Object, Meta-data Management, Storage
