New IIIF Support in FromThePage
The latest version of the FromThePage crowdsourcing platform includes major enhancements to its IIIF support. The full documentation is available at FromThePage Support for the IIIF Presentation API and Web Annotations, but here are some highlights:
We've used the IIIF manifests produced by FromThePage as the main integration point for digital library systems. To that end, integration endpoints are exposed using linking properties on the IIIF manifests produced for any object made public on FromThePage.
Links to human-readable application pages to transcribe pages, translate pages, read documents or access document tables of contents are exposed via
related elements on canvases and manifests. Human-readable transcripts and translations of an entire document in verbatim plaintext, HTML, and TEI-XML appear as
rendering elements for a sequence. Machine-readable exports of text in various plaintext formats for full-text search, analyisis, or editing as well as CSV exports of tagged subjects are presented as
seeAlso elements for the manifest and corresponding canvas.
Exposing transcripts, corrected OCR, translations, and notes via properties on derivative manifests allow content providers (or anyone else) to access crowdsourced contributions, but how do they find out what contributions have been made or what materials of theirs have been imported into the platform? While discovery of enhancements made to IIIF content has been under long and productive discussion by the community, integration of crowdsourced contributions into content provders' digital library systems presents a pressing and particular use case. To that end, we have added a simple
contributions API to FromThePage.
The user contributions API allows a content provider to pass the domain name that appears in their manifests and optional start-date and end-date parameters to a FromThePage server and receive a IIIF collection containing links to derivative manifests for all projects created from the provider's manifests. For ease of integration, each manifest link contains a
metadata block with a single
dc:source element containing the orginal manifest IRI, as well as a contribution status service.
When is a transcription or translation "done"? We've found that the answer varies from institution to institution, from project to project, and sometimes from work to work. We've tried to satsify the needs of disparate projects by creating a fine-grained status service at the work/manifest level and page/canvas level. At the manifest level, the service exposes the percentage of the work which as been transcribed, OCR-corrected, translated, indexed, marked as needing review or marked as blank. (A
pctCompleted element rolls up the most common use cases into a single number.) At the canvas level, flags represent whether a page has a translation, a transcription, subject tags, or has been marked as blank or needing review.