neosemantics

neosemantics copied to clipboard

Hi, I would like to know if there is any plan to add more advanced reasoning capabilities (like classical reasoners such as Hermit, etc.) to neosemantics ? I heard about GraphScale (https://neo4j.com/blog/neo4j-rdf-graph-database-reasoning-engine/) but it does seem like the project was abandoned.

I have the same question. By looking over the docs here, it seems classical reasoning is not possible, or am I missing something?

Hi, Sorry, I have just discovered this thread. We have developed the GraphScale technology as a proof-of-concept to show how rule-based OWL reasoning can be effectively implemented on top of any graph store. It's not compatible with the way neosemantics works (yet). Can I ask about your objectives wrt. reasoning and the status of your project?

Hi.

Our use case is pretty simple : RDFify tabular data and load all the triples to a triplestore/graph DB that uses an ontology as its semantic backbone. When new triples are loaded to the graph, we would like the reasoner to derive additional triples, given the ontology axioms (and also additional handmade rules).

For the moment, we are using RDFox, which gives us satisfaction, but it is an in-memory database, which has some drawbacks. So I am looking for an alternative, e.g. Jena, BlazeGraph...

Hi, anything new about this?

Hi. There is nothing new wrt. GraphScale. We also use RDFox at customer projects with great success (disclaimer: we are partner of OST, the company behind RDFox). RDFox uses pretty optimized data structures that allow to deal with large volumes of data. We didn't hit a barrier in this respect. When switching to a secondary-memory engine you have to keep in mind that reasoning might cause to load large parts of your dataset into main memory again to evaluate the axioms/rules. Maybe you can split your data into packages that allow for sequential batch processing. Or do you have a streaming use-case?

I had some infrastructure constraints that prevented me from using RDFox, but these constraints are no more, and I could turn to this solution again. Actually, my lab is interested in using Neo4j to create and manage some graph databases (with no reasoning), and ideally I'd like to use Neo4j for my own use cases that include ontological reasoning, so that we do not multiply the number of different DB tools. But it seems that rule-based OWL reasoning is out of Neo4j scope. Thank you anyway.