Suggestions for RDF Alternatives to SQLite with Python API

Matthias S. · 2023-10-06T09:49:29.198Z

Hi, I am looking for something similar to SQLlite but for RDF. Any suggestions? (it is even better if there is a Python API that comes with it) I am looking at this project https://github.com/egerber/spaCy-entity-linker/. The authors are using an SQLite DB and I find the queries to retrieve Wikidata entities quite complex. I am wondering if we could not have the same approach with an RDF triple store. Any thoughts concerning the size limitations? For those interested, with a colleague (@Marion S.), we started implementing an Ontology Learning python library as part of our PhDs: https://gitlab.insa-rouen.fr/msesboue/ontology-learning/-/tree/olms2?ref_type=heads It is still at a very early stage but we are always happy to have feedbacks and discussions 🙂 Thanks.

The Knowledge Graph Conference

8 comments

· Sorted by Oldest

Pierre Monnin
·
Hi! Have you checked out Corese? (https://project.inria.fr/corese/) it can be used to set up a SPARQL endpoint using RAM or disk persistence.
Bob DuCharme
·
The free version of Ontotext's GraphDB triplestore include spaCy integration, which I wrote about in my personal blog at https://www.bobdc.com/blog/spacy/. (Full disclosure: I an an Ontotext employee.) I have often wished that there was an RDF equivalent of SQLite. If you want to use a Python API for something like that, the RDFlib python library may do nearly everything you need, including the storage of triples that you can then query with SPARQL.
Matthias S.
·
Hi Pierre Monnin and Bob DuCharme, Thank you so much for your answers, I will definitly check that out. Bob DuCharme I love Ontotext, i am using it though I did not know they have an integration with spaCy, that is awesome, thnaks 🙂 Pierre Monnin I know Corese and played a bit out with it. But I thouht it was a server like database. Does it work with an on-disk file in a serverless manner as well ?
Pierre Monnin
·
Matthias S. It is a server like database but can be used w/o disk persistence (only RAM, loading a turtle/rdf+xml/... file). So it could correspond to a lightweight database in the sense of SQLite. But that needs to be confirmed w.r.t. to your specific requirements 🙂
Matthias S.
·
It looks very interesting, I will go check this out. Are you developing Corese or is it Remi still working on it? (Nice to see some Wimmics folks there by the way 🙂 ) Thanks
Pierre Monnin
·
Remi is the developer. I am just advertising his work 🙂
Ryan B.
·
What about https://owlready2.readthedocs.io/en/latest/onto.html?
Owlready version 2 includes an optimized triplestore / quadstore, based on SQLite3. This quadstore is optimized both for performance and memory consumption. Contrary to version 1, Owlready2 can deal with big ontologies.
It provides a Python API around an RDF triplestore that uses sqllite3 as the in-memory db. It’s been a very handy tool.
👍1
Matthias S.
·
Oh! Nice catch !! thanks Ryan B. I will definitly check this out 🙂 thanks

Pierre Monnin
·
Hi! Have you checked out Corese? (https://project.inria.fr/corese/) it can be used to set up a SPARQL endpoint using RAM or disk persistence.
Bob DuCharme
·
The free version of Ontotext's GraphDB triplestore include spaCy integration, which I wrote about in my personal blog at https://www.bobdc.com/blog/spacy/. (Full disclosure: I an an Ontotext employee.) I have often wished that there was an RDF equivalent of SQLite. If you want to use a Python API for something like that, the RDFlib python library may do nearly everything you need, including the storage of triples that you can then query with SPARQL.
Matthias S.
·
Hi Pierre Monnin and Bob DuCharme, Thank you so much for your answers, I will definitly check that out. Bob DuCharme I love Ontotext, i am using it though I did not know they have an integration with spaCy, that is awesome, thnaks 🙂 Pierre Monnin I know Corese and played a bit out with it. But I thouht it was a server like database. Does it work with an on-disk file in a serverless manner as well ?
Pierre Monnin
·
Matthias S. It is a server like database but can be used w/o disk persistence (only RAM, loading a turtle/rdf+xml/... file). So it could correspond to a lightweight database in the sense of SQLite. But that needs to be confirmed w.r.t. to your specific requirements 🙂
Matthias S.
·
It looks very interesting, I will go check this out. Are you developing Corese or is it Remi still working on it? (Nice to see some Wimmics folks there by the way 🙂 ) Thanks
Pierre Monnin
·
Remi is the developer. I am just advertising his work 🙂
Ryan B.
·
What about https://owlready2.readthedocs.io/en/latest/onto.html?
Owlready version 2 includes an optimized triplestore / quadstore, based on SQLite3. This quadstore is optimized both for performance and memory consumption. Contrary to version 1, Owlready2 can deal with big ontologies.
It provides a Python API around an RDF triplestore that uses sqllite3 as the in-memory db. It’s been a very handy tool.
👍1
Matthias S.
·
Oh! Nice catch !! thanks Ryan B. I will definitly check this out 🙂 thanks