CNI White Paper on Networked Information Discovery and Retrieval
Outline: Chapter 2
Architectures and Technologies to Support the NIDR Process
Part I: The Machinery of Discovery
- targets for indexing; objects and information spaces
- the composition and subsumption of information spaces and the role of gateways among spaces
- defining collections and bounding searches
- the need for a modular indexing architecture (to allow introduction of new indexing methods)
- "push" vs. "pull" models: archie, veronica, webcrawlers, harvest.
- the central role of the "gatherer"; interactions with privacy, intellectual property, economics
- redistribution and aggregation of gathered indexing information -- index brokers and related proposals.
- protocol issues; interoperability, quality assurance, introduction of specialized extensions and new versions within an open architecture.
- the user side of discovery: "user history" databases. integration of multiple indexing systems. the mechanics of ranking and duplicate detection.
Part II: The Machinery of Retrieval
- characteristics and limits of current retrieval protocols
- "fetch and use" vs. "use across the net" models
- the URC framework for invoking retrieval
- the need to extend current retrieval protocols
- format conversion
- integrity issues
- browsing and sampling (thumbnails, derived data, sample data)
- component/subobject extraction and retrieval
- caching and replication; retrieving from managed distributed spaces