Possible storage option for data extracted from DBPedia - owl

Question

I'm developing an application that allows users tag product purchases (via a Web App).
I intend to use the tags to automatically query DBPedia (Possible other Open Data Sources such as FreeBase).
The top N results returned from DBPEdia will be displayed to users and they will select the one that most closely resembles the tag they entered. (I will only extract specific data).
For example:
User enters tag 'iPhone' and SparSQL query sent to DBPedia. Results are parsed and some data on each result shown to user who then selects the one that most closely resembles what they bought.
I want to extract some of the data from the users selected DBpedia result and store it for marketing purposes at a later stage. (Ideally via some call to an API)
I was thinking either Bigdata or Protege OWL but have no experience of using either
Can anybody suggest the best tool for this task and advantages/disadvantages/learning curve/etc...?
Thanks

Answer 1

It all depends on what you want to do with the data that you've extracted. The simplest option is just to store the reconciled entity URI along with your other data in a relational database or even a NoSQL database. This lets you easily query Freebase and DBpedia for that entity later on.
If you want to pull in "everything there is to know" about an entity from Freebase and DBpedia, then you're probably better off with a triple store. With this approach, you can query all the data locally; but now you have to worry about keeping it updated.

Answer 2

For the kind of thing you have in mind, I don't think you necessarily need a highly scalable triplestore solution. More important seems to me that you have a toolkit for easy execution of SPARQL queries, result processing, and quick local caching of RDF data.
With those things in mind, I'd recommend having a look at OpenRDF Sesame. It's a Java toolkit and API for working with RDF and SPARQL with support for multiple storage backends. It has a few built-in stores that perform well for what you need (scaling up to about 100 million facts in a single store), and if you do find you need a bigger/better storage solution, stores like BigData or OWLIM are pretty much just drop-in replacements for Sesame's own storage backends, so you get to switch without having to make large changes to your code.
Just to give you an idea: the following lines of code use Sesame to fire a SPARQL query against DBPedia and process the result:
SPARQLRepository dbpediaEndpoint = new SPARQLRepository("http://dbpedia.org/sparql");
dbpediaEndpoint.initialize();
RepositoryConnection conn = dbpediaEndpoint.getConnection();
try {
String queryString = " SELECT ?x WHERE { ?x a foaf:Person } LIMIT 10";
TupleQuery query = conn.prepareTupleQuery(Querylanguage.SPARQL, queryString);
TupleQueryResult result = query.evaluate();
while(result.hasNext()) {
// and so on and so forth, see sesame manual/javadocs
// for details and examples
}
}
finally {
conn.close();
}
(disclosure: I work on Sesame)