Monday, September 8, 2008

A Framework for Building Open Digital Libraries

"Open" DLs build directly on concepts/philosophies of Open Archives Initiative

Most existing systems classified as DLs resulted from custom-built software dev projects - each involves intensive design, implementation & testing cycles.
- why repeat effort?
- some software toolkits: Dienst, Repository-in-a-box

Most programming environments adopt a component model

Oct 1999 - OAI launched.
- focus on high-level communication among systems & simplicity of protocol
- OAI protocol for Metadata Harvestingn (OAI-PMH): system of interconnected components
- OAI protocol can be thought of as glue that binds together components of a larger DL (or collaborative system of DLs)

DLs modeled as networks of extended OAs, with each OA being a source of data and/or provider of services.
- this approach closely resembles the way physical libraries work
- research & production DLs differ

ODLs guided by a set of design principles & operationalized with aid of OAI-PMH extensions. proven techniques from internet development
- simplicity of protocols, oppeness of standards, layering of semantics, independence of components, loose coupling of systems, purposeful othogonality, reuse

Formal Principles:
1) All DL svcs should be encapsulated w/in components that are extensions of OAs
2) All access to DL svcs should be through their extended OAI itnerfaces
3) Semantics of OAI Protocol should be extended or overloaded as allowed by OAI protocol, but w/out contradicting essential meaning
4) All DL svcs should get access to other data sources using extended OAI protocol
5) Dls should be constructed as networks of extended OAs

OAI Harvester obtains datastream which creates indices for searching.

Components in prototype systems:
- Union: combine metadata from mutl src
- Filter: reformat metadata from non-OAI data srcs
- Search: search-engine functionality
- Browse: bategory-driven browsing fx'lity
- Recent: sample of recently-added items

Tested on NDLTD system - good feedback

Designing by principles & implementing for real-world scenario

No comments: