Open Addaci opened 6 years ago
Input from @rsimon, Technical Director of Pelagios & Lead Developer of Recogito [April 3rd 2018, 15.16]
Rainer Simon replied to the topic IIIF, Transkribus & Recogito annotation approaches/ecosystems in the forum Recogito Users forum
"Hi,
<I would find it useful to have a high level description of the relationship between IIIF, Transkribus, and Recogito annotation approaches>
There is no existing high level description I'm afraid ;-) be interested in a bigger picture of how Recogito can fit with Transkribus, myself, e.g., because that's entirely unexplored territory.
Some thoughts with regard to IIIF though:
Recogito has been supporting for a while "import" of images served via the IIIF image API protocol, although it's been somewhat of a hidden feature. (Technically not an "import" of couse, since those images are served directly from their respective institutions' image servers.)
I have recently (last week) extended Recogito to provide support for manifests conforming to (what I think is) the Presentation API v2. That means you can drop in the URL to a manifest for a particular item (e.g. a manuscript) and will get the images listed in Recogito, properly labeled and sorted etc. We'll see what additonal metadata makes sense to pull out from the manifests later, but for the time being that's all we really need.
I've recently seen an example of a IIIF annotation list (which I think is an addition in Presentation API v3.0? But not sure. As a single developer, juggling a bunch of different user interests of which IIIF just one, it's hard to keep up with the huge & moving standard...) Anyways: adding "import" for such annotation lists looks pretty straightforward & is definitely on the list.
Likewise, it should be pretty easy to expose/export annotations created in Recogito as such an annotation list, as one additional format among the several we support already.
That being said: Recogito's annotation model does provide more structure and possiblities as what I've seen in IIIF examples so far. My impression is that annotations in IIIF are often just a snippet of HTML. With regard to importing, Recogito would probably jsut treat that as a single textual comment, with no author attribution information.
Annotations created in Recogito, on the other hand, can be a series of comments (and replies), tags, transcriptions, URIs to place or person authority files; all individually stamped with information about what user created them, and when). Exporting those to a IIIF annotation list would likely have to end up being a lossy process, where IIIF annotations would be simplified versions of our internal originals, for the sake of compatibility.
We can (and do, for the JSON-LD export) make our annotations compliant to the Web Annotation data model. But as far as I'm aware there are few implementations that actually do handle structured annotations, originating from other systems, so it's not possible to set up a real world test case. Input and ideas around this are appreciated!
P.S.: the way our annotations are structured in Recogito is the result of working with scholarly users over the past two years. By now, I think Recogito has a lot to offer that's highly relevant in every scenario where annotation needs to go beyond just the basic case of highlighting a region and attaching some commentary. AFAIK it's not something that UV and Mirador offer out of the box. I regret that there has been very little interaction between us and them so far. Primarily, this is really due to lack of time on my end to "dance on all weddings at once", as we say in German. (I guess in English it's simply "juggling too many balls at once" ;-) I'd certainly love to see more engagement with the IIIF community, but it's currently restricted by the realities of available resources. (That doesn't mean that members from the Pelagios community can't step in, though! ;-)
Pelagios/recogito2 issue #441, opened No 29, 2017 and now closed, provides an overview of issues to do with upload of IIIF URL's of an image's info,json into recogito2 upload box
https://github.com/pelagios/recogito2/issues/441
See also recogito2 Annotation UI Enhancements planned for end April 2018
Twitter message from MarineLivesorg account, April 4th, 2018: "We are doing a compare and contrast on IIIF, Transkribus & Recogito/Pelagios ecosystems generally, and especially annotation capabilities as input to one strand of our June 5th Stadsarchief workshop on #Signsofliteracy Who should we contact at @Transkribus https://github.com/Signsofliteracy/Signoff/issues/5"
<I would find it useful to have a high level description of the relationship between IIIF, Transkribus, and Recogito annotation approaches>
@rsimon Regarding your earlier comments in the Pelagios Community discussion area in reply to my query, and reposted above by me in this repository, it is useful that Glen Robson, IIIF technical coordinator, has been taking an interest in MarineLives. I first came into contact with Glen when he was systems head at the National Library of Wales about eighteen months back. Glen encouraged me to submit a paper to the IIIF Washington DC Conference, which has been accepted. He is very busy on the Conference organisation at the moment, but I am hoping that he will be able to Skype into the June 5th, 2018 workshop at the Stadsarchief Amsterdam. This will enable us to have a decent discussion of annotation capabilities and aspirations in IIIF and Recogito as well, I hope, with someone on the technical side from Transkribus concerned with semantic annotation capabilities in their tool suite, as well as other aspects of their tools of interest to us as we look at the potential for pattern recognition and machine learning to look at machine defined areas in manuscript images.
I would be interested in discussing IIIF with representatives of the Pelagios Community. With the next version of IIIF version 3, the plan is to move to Web Annotations so hopefully that will make the import/export from Pelagios and IIIF easier. IIIF currently uses Open Annotations which was a precursor to Web Annotations.
@rsimon I have been in touch with @glenrobson, IIIF technical coordinator, to see if I can get him to the June 5, 2018 workshop on Technology to explore Historical Literacy, which I am organising with the Alle Amsterdamser Akten team at the Stadsarchief Amsterdam, and which I know you are considering attending. He can't physically be at the workshop, but is checking his diary to see whether he can join the workshop by Skype
He has signalled two things
(1) He is interested in meeting with the Pelagios/Recogito team in the context of IIIF
(2) He is interested in discussing the potential for some form of IIIF/Pelagios annotation linkage in the context of IIIF Version 3, stating "With the next version of IIIF version 3, the plan is to move to Web Annotations so hopefully that will make the import/export from Pelagios and IIIF easier. IIIF currently uses Open Annotations which was a precursor to Web Annotations"
One possibility to get members of the Pelagios/Recogito core team together with Glen, would be for one of you to attend the EuropeanaTech pre-conference workshop which Glen is organising on IIIF on Monday, May 14th, 2018 (also in Rotterdam). Europeana Tech itself is on May 15th and 15th, also in Amsterdam, and will be heavily attended by GLAM institutions. Dr Mia Ridge, digital curator at the BL, will be speaking at EuropeanaTech on LibCrowds, the IIIF compliant crowdsourcng platform I am interested in for the Signs of Literacy community and Glen himself will be speaking on IIIF.
I am also checking with @voetnoot (member of the Alle Amsterdamser Akten project team at the Stadsarchief Amsterdam) and Marc Holtman (head of digitisation at the Stadsarchief Amsterdam), whether there will be someone from the Stadsarchief at Europeana Tech.
If one of the Pelagios/Recogito team were to go to Rotterdam for that workshop, I would also signup to the workshop and fly out to Rotterdam for the Monday. I am not sure I have time to go to the subsequent two day Europeana Tech, since I am flying to Washington DC on Saturday, May 19th for the IIIF 2018 Conference the following week, where I am speaking about technology & IIIF enablement of the Signs of Literacy community, but I would be interested in your reactions to IIIF at the workshop and the chance to learn more about IIIF annotations. Wikidata and Wikimedia are also running workshops on May 14th, and again it might be interesting to think about the potential availability of Pelagios annotation datasets and/or gazetteers and gazetteer data standards through Wikidata/Wikimedia.
Hi @glenrobson,
great! Incidentially, my colleague Bekka - Pelagios Community Manager - will be at EuropeanaTech in Rotterdam. I see that your workshop is fully booked already. Also, I'm not sure whether Bekka will be available in the morning of May 14 already. But we had a quick chat today, and she thinks she could drop by in the afternoon (unconference) session. Let's use this for a quick touch base/first contact meetup, and then take things further from there!
As far as tech is concerned: we currently have export that's (hopefully) Web Annotation compliant; and seem to be handle display of Image API URLs as well as Presentation API (v2) manifests just fine. I recently got a chance to look at an annotation list in (what I think was) Presentation API 3 format. Getting annotations from such a manifest into Recogito looks pretty straightforward (with some minor caveats, as written above). Vice versa, we could easily expose an identically formatted annotation list out of Recogito, to cover the full roundtrip, into and out of Recogito.
If IIIF moves to Web Annotations it's certainly helpful. But we'd also have an Open Annotation export available now already (RDF/XML, TTL). I have to admit that this one is slightly unfinished. But could be patched up to the full spec, if anyone asks for it. Either way, I'm not terribly worried about exact formats and syntaxes. We have been building crosswalks for different use cases for a while (from CSV to GeoJSON to RDF), and are pretty quick in adding additional export interfaces as the need arises.
P.S.: for a sample see
The JSON-LD button will export to Web Annotation format. This, I guess, could already serve as a basis for a potential future IIIF-compliant annotation list?
It would be useful to have a high level description of the relationship between IIIF, Transkribus, and Recogito annotation approaches. This will provide context for one strand of discussion at the Stadsarchief workshop on June 5th, 2018.