Make OntoPortal work with multiple Triple stores (4store, Alegrograph, Virtuoso or GraphDB)

[syphax-bouazzouni]

Context

We have been using 4store as our triple store for the past 10 years, since its inception on December 6, 2012. However, we have encountered several issues with it in recent years. Despite its initial success, 4store has not been maintained since 2017, which has been 6 years ago.

This lack of maintenance has resulted in performance issues and frequent downtimes, especially when we have more than a hundred ontologies and a lot of metadata. Furthermore, we have no way to get support for it, and we do not have the resources to fix the issues ourselves. As a result, we have decided that we need to migrate to a different triple store.

The @ontoportal/bioportal team started working on this migration years ago, and they chose to migrate to Alegrograph. In November 2023, they will deploy it in production for Bioportal.

At Agroportal, we also plan to migrate out of 4store in the hope of better performance in our relatively small infrastructure. Going from the work done for Alegrograph by the Bioportal team, we did a first iteration on making Ontoportal also work for Virtuoso and GraphDB (most of the details can be found here: ontoportal-lirmm/goo#48). Since AllegroGraph is not mostly free, we will surely go for Virtuoso, and we plan to do so in the second quarter of 2024.

Roadmap

• Bioportal: Merge and optimize the AllegroGraph branch code ncbo/bioportal-project#122, planned to November 2023
• Agroportal : Update our triple store to a new one (Alegrograph, Virtuoso or Graph DB) agroportal/project-management#229, planed to March 2024

[syphax-bouazzouni]

For those who are using 4store and have no intentions of transitioning to a new triple store, I believe we will continue to offer support for 4store for some time to come.

However, at Agroportal, we have encountered certain constraints while using 4store, which have restrained our ability to execute some elaborate SPARQL queries (refer to ontoportal-lirmm/ontologies_api#25 and 4store/4store#164) in our newly developed features. Consequently, we are considering shifting our focus towards newer triple stores, such as Alegrograph, Virtuoso, or GraphDB, which offer enhanced support for more complex SPARQL queries. We plan to initiate this change starting in 2025 at Agroportal.

I welcome feedback from anyone with expertise in triple stores or work with experts in this field, as it would be really helpfull to know which one to choose and focus on.
Additionally, if you believe there are other triple stores we should consider integrating into Ontoportal, please feel free to share your suggestions.

FYI @mdorf, @alexskr and @naouelkaram

[jonquet]

FYi @agipap can put us in touch with Ontotext GraphDB to discuss possible strategies there.

[jonquet]

Rquirements from the OntoPortal Workshop 2023:

• Create a SPARQL query editor/viewer. See Offer a SPARQL endpoint query mechanism #41
• Create an option to enable the use of a separate public SPARQL triplestore

[turbomam]

I would love to interact with my local (or the public) `Portals through a modern triplestore. If you're going to build all of this into the Docker container, then I suppose the ease of installation doesn't really matter. But I will say that managing GraphDB is way easier than managing Virtuoso.

Have you also considered Jena/TDB/Fuseki?