200 likes | 318 Views
HAROKOPIO UNIVERSITY. A Distributed Architecture for Building Federated Digital Libraries. Mara Nikolaidou Department of Informatics and Telematics, Harokopio University of Athens George Pyrounakis Department of Informatics and Telecommunications, University of Athens.
E N D
HAROKOPIO UNIVERSITY A Distributed Architecture for Building Federated Digital Libraries Mara NikolaidouDepartment of Informatics and Telematics, Harokopio University of AthensGeorge PyrounakisDepartment of Informatics and Telecommunications, University of Athens euroCRIS Meeting, Athens 11-13 May 2009
Current State in Federated Digital Libraries • A central server harvests metadata from the federation repositories • In most cases Open Archive Initiative Protocol for Metadata Harvesting (OAI-PMH) is used • Digital library services are usually dependent on the repository software • Difficulty to build advanced services in a federated environment • Functionality offered by federated digital libraries is not rich
Architecture Using OAI-PMH DL services DR OAI-PMH Harvester OAI-PMH Provider OAI-PMH Provider OAI-PMH Provider DR DR DR
DL Services not supported • Workflows for submitting, editing and publishing digital objects • Management of user access policies applied on digital objects and their components • Management of digital collections • Digital content handling (conversion, watermarks, compression etc.) • Preservation policies for digital content
Middleware and Distributed Architectures • Current architectural structures (Web Services, CORBA, etc) introduce the concept middleware in order to abstract digital objects. The abstraction separates the physical (stored) from the logical representation of objects • We adopt a similar approach to create a middleware that interprets the structure and semantics of digital objects stored in different repository software located in multiple servers • Using Harvesting digital objects are imported to a central server in a simplified form and the DL services are built on top of that. • A DL middleware preserves the role of the “interpreter” between the DL services and the digital objects that remain located in there local repositories.
DL Architecture Using a Middleware DL services DL services middleware API middleware connector connector connector DR DR DR
Proposed Concepts • A middleware is proposed for the creation of federated DLs called Digital Object Abstraction Layer (DOAL) • Scope • Build federated digital libraries • Create reusable digital library services • DOAL provides an API to DL services, which use an abstract but semantically-rich form of digital objects, independent from the repository that are stored in. • DOAL should work over heterogeneous repository software. • Digital Object notion within DOAL should be handled independently from its stored form (or forms).
Abstract Digital Objects • DOAL uses an abstract form of digital objects that retain their semantics (metadata sets, digital content, relations, behavior) and • Hides their repository-dependent characteristics including • the location of the stored digital object, • the repository software used for its storage and • the encoding used for its serialization. • Abstraction of digital objects is realized using the Digital Object Prototype (DOP) Mechanism
DOAL Benefits • Abstracts the serialized digital object in a form manageable by DL services independently of the repository software where is stored • Associates digital objects to their DOP, so they are handled in a common manner by the DL services according to the their type • Homogenizes the digital collections of different repositories that are part of a federated DL • Distributed Digital Objects (DDOs) are used which parts are stored in multiple heterogeneous repositories
Pergamos Digital Library • The concept of DOP has been implemented in Pergamos Digital Library by the Libraries Computer Center of the University of Athens • The DL is able to host heterogeneous digital collections using different metadata, digital content, hierarchies and views • A common web interface for all collections is used by cataloguers as long as end users • A type is assigned to each digital object using the notion of DOP
Pergamos Architecture (I) • Centralized 3-tier Architecture • Fedora Digital Repository version 2.0 is used as a core for the storage and retrieval of digital objects • DOP Mechanism is responsible for the abstraction of digital objects based on the assigned type • The Application Logic Layer uses the abstract digital objects, which are semantically rich
Pergamos Collections • Pergamos is hosting four digital collections in http://pergamos.lib.uoa.gr • Historical Archive of the University of Athens • Folklore Collection • Theatrical Collection • Psachos Music Library Collection • Eight more collections are ready to be published (papyri, medical images, mineralogy, published elections, etc.)
DDO Simple Example DOAL DDO 1 DR DR SDO x SDO y DC WEB THUMB HQ Node A Node B
DOAL Architecture Workflow Manager Browsing - Searching Personalization Object Editor DL Services DOAL API DDO Registry Abstraction Mechanism DDO DOAL DDO DOP Registry DDO Connectors Pool DDO Pool DR Registry Fedora Repository DSpace Repository Custom Repository Repositories
DOAL Modules • Registries keep information about DDOs, Digital Repositories and DOPs • Connectors are implemented for each repository software participating in the federated digital library • Abstraction Mechanism is responsible for the abstraction of digital objects according to their type (described in DOP) • DDO Pool holds the DDO instances which are available to be used by the DL services
Current State • Implementing connectors for Fedora and DSpace • Implementing DOAL Registry • DL Registry • DDO – DOP Registry • Implementing DDO Handlers • First version of DOAL expected by February 2010