Community:Collections and Metadata/ServiceProvider
From NSDLWiki
Guidelines / Best Practices for OAI Service Providers and Harvesters
- SP to SP communication
- Sharing commonly needed and desired features among SPs
- features vs. data
- registry of set descriptions?
- registry of deleted record implementation settings?
- Sharing ways to recover from DP errors
- DP OAI compliance error reporting (centralized)
- Test suite for OAI harvesters (Simeon)
- Aggregator issues (i.e. exposing aggregations as data providers)
- rights
- provenance
- branding
- Example aggregator instructions for harvesting
- DP issues (also see DataServiceCommunications)
- Deleted records
- Dependence on DP practices
- e.g. transient deleted records at DP imply SP must do what for its own deleted rec implementation?
- SetPractices
- Clustering analysis
- Share tools and strategies among SPs
- arXiv.org SVM approach
- NMF per Martin
- Use full-text
- NSDL-iVia autogen md approach
- Harvesting issues
- Transitive closure issues, i.e., harvesting loops by aggregators
- Can provenance solve this?
- How to be a well-behaved harvester?
- e.g., 503 response for novices, throttle on server (Simeon)
- When XML schemas are inaccessible (due to network glitch, server temporarily down, etc.)
- Record identifier changes (data providers are not required to keep the same identifier for the same record)
- Periodic "clean slate" reharvests are recommended for updating since <oai-identifier> elements cannot be relied upon from harvest to harvest
- Metadata issues (see also MetadataContent and [GeneralOAICommunity Strategies and Tools])
- Metadata Formats (see also MultipleMetadataFormats)
- Acquiring and processing metadata formats other than oai_dc
- crosswalks
- from Jeff Young
- NSDL safe transforms - Naomi
- Name authority issues
