[EMAIL PROTECTED] a écrit : > I'm a noob to the semantic web though not to python, and am looking to > experiment with the dbpedia data (now that's a possible killer app :-), > which is sizable, like a few million triples it seems: http://dbpedia.org/ > > I'm trying to figure out how to get the dbpedia data loaded and usable > with rdflib, i.e. accessible via a SPARQL interface on my local machine, > and could really use some advice - from the API docs I don't > understand what needs to be done with store(), except that it's needed, > and I probably don't want to load several hundred megabytes into memory. > :-) Could anyone give some high-level advice on how I can load big > n-triples files (that would be the dbpedia .nt files) using rdflib, > then have them all accessible via SPARQL? Thanks in advance.
I guess you will need one a proper RDBMS-based backend and then find a way to load the data by chunks doing a database commit every N added triplets. This can be a really long process so having a ready to load low level-formatted store might help a lot. I did not know about the DBpedia.org project: this is a really cool project. What would be a really nice killer app would be a python-based natural language client to dbpedia (along with wordnet maybe). Using nltk-lite might help a lot there: http://nltk.sourceforge.net/index.php/Main_Page -- Olivier _______________________________________________ Dev mailing list Dev@rdflib.net http://rdflib.net/mailman/listinfo/dev