The platform
Orchestra
Orchestra is an Open-Data Publishing Platform. Each stage of the process — Harvest, Analyze, Map fields, Transform, Enrich, Publish — is a service inside it.
Six stages, one continuous flow
Harvest
Metadata is pulled from wherever it lives — OAI-PMH endpoints (the standard protocol collection systems use to share records), SIP packages (source records bundled with their mapping), XML and CSV dumps, existing databases. Sources can be re-harvested on a schedule, so updates keep coming in instead of going stale after a one-off import.
Analyze
Before a single field is mapped, Orchestra profiles the source data — coverage, value distributions, quality, structure — so the mapping starts from what is actually in the data, not guesses about what should be.
Map fields
Source fields are mapped to your data model with a live preview, so each transformation is checked before it runs, not after a full run.
Transform
Mapping designs the transformation; transforming applies it — across the entire dataset, converting original source records into your target schema, ready to enrich and publish.
Enrich
Records become far more useful when they're linked. Enriching connects them to people, places, and concepts, adding the authorities and context that make search meaningful.
Publish
Publishing indexes your records into a live search index — the engine behind both your public search and your open data layer. Indexing is publishing — the moment records land, they're searchable.
DIW — the search your visitors actually use
DIW (Delving Instant Websearch) is the front end for the published index: fast, faceted, and installable on almost any web platform, WordPress included. It reads straight from the index, so what you publish is what people search.
An open data layer, not a black box
Everything you publish is available through an open search API, and on request as linked open data with RDF and a SPARQL endpoint — so you, or anyone you choose, can build on top. See the developer story →
Every collection, the same care
A local historical society and a national archive go through exactly the same steps: the source is checked before anything is mapped, fields are mapped with a live preview, records are linked to shared authorities, and the result lands in the same search index and open API. What scales with your collection is the price, not the care — and everything is published under your own name. That includes datasets the big aggregators don’t take in.
Beyond heritage
Nothing in the process assumes heritage data. Harvesting reads any structured source, mapping targets the data model you choose, and search and the API work on whatever you publish. If your organisation holds a large, structured dataset with the same problem — making it findable, and making it reusable — we’d like to hear about it.
Have a collection that deserves to be found?
Tell us about your data. We'll show you what Delving can do with it.
Get in touch