Will S core version be graph based?


I’'m looking into the application of Omeka for multiple different collections that will however be related in several ways, by (sub)topics, involved persons, locations, and context. (To make it a little more concrete, think historical conflict sites of the 20th century and ~10 related international collections of institutions and memorials.) From reading this blogpost (much more info is not easily found) I get the impression the S version might be better for our needs than the current 2.x (classic). We plan to use the platform to primarily interface to our own front-end applications first, so website publishing isn’t immediate priority, but flexible catalog authoring by multiple users definitely is, and a LOD/ RDF-based approach to relations and ontology will be also. We will have to develop and adjust on top of any way we choose, but of course we want to reduce development repetition.

Because of several unknowns and the expected heterogeneity of the datasets, we (=few people, still) can’t really oversee how best to gradually upscale from single ‘clients’, to interface to joined collections or separate client installs, nor what ontology will emerge. But since we’re at the start with 0 data to migrate still, we might choose the option that can best ‘grow along’ in agile fashion. In that long-term context, a Graph-based core approach seems better to me than a RDBMS one.

From the above, do you think Omeka S the better option to start, or would you still recommend to start with Omeka Classic (or perhaps Ozmeka) and take leverage from its larger set of stable add-ons for the time being, and just plan migration & development steps?

Thanks a lot for any advice!

The first thing to consider is that Omeka S is still in alpha, if you need something stable in the short term, that counts against it.

On the other hand, while Omeka S core isn’t strictly graph-based, it does get pretty close to it. The API uses JSON-LD, and the data input builds in many graph principles. You can also import RDF vocabularies and OWL ontologies, though some data from those is not represented in Omeka S.

While it will be possible to import data from Classic to S, that import will still miss what sounds like the graph features you are talking about, for example the item relations in Ozmeka. What add-ons will be recreated for Omeka S is a big question.

So, if you don’t need immediate stability, but creating and exposing data as RDF (in the form of JSON-LD) is a significant consideration, Omeka S sounds like the way to go.

Thanks @patrickmj for your swift answer. From it I understand that Omeka S might not give us immediately visible leverage in, say the coming six months (unless we help build elements ourselves, fair enough), while Omeka Classic could do so for some functionality while limiting us for what we need, eventually. Looks like we came at the right time :wink:! I guess we’ll try to run a small Classic case that could be moved, and another in S in parallel for a short while.

Thanks again.