Freebase Issue about defacto HOT 6 CLOSED

The demo in the defacto-restful module, creates models in the defacto-core. If it is not working then I think the defacto-core is not supporting it.

The given resources need to have at least the property rdfs label as far as I remember.

In the log file is a Nullpointer:
java.lang.NullPointerException
at org.aksw.defacto.search.crawl.EvidenceCrawler.crawlEvidence(EvidenceCrawler.java:115)
Because the subject label in the model is not set.

The solution is, for sure, to set the label correct. Not sure where the bug is.

I copy pasted DeFacto/defacto-gui/src/main/java/org/aksw/defacto/util/DeFactoModelGenerator.java line 77 in DeFacto/defacto-restful/src/main/java/org/aksw/defacto/restful/core/RestModel.java line 168 there are DBpedia endpoints to search for the label and the given URI. I guess there it fails because of the endpoints?

from defacto.

And in case of [1] you already have the labels so you do not need to search for them. Instead you can build the model directly. But the demo in the defacto-restful module is not build for that case.

from defacto.

Yes, of course. But it turns out that I am implementing a new feature that must swap the resources. The main problem is that DeFacto has its model (DefactoModel) built up over a jena model and the current method for set a resource dynamically (subject or object, does not matter in this case) just change the resource itself (DefactoModel value object), but not the extra information already existing over the jena model (the "low level" layer)!! This leads to incorrect DeFacto scores. That´s why the method you created in the rest-module for dynamically generate the models would be useful (btw, we should move it later as a feature in the core layer...). Back to the Freebase issue, currently (as far as I see) defacto-core module works with FactBench examples, i.e., you already have all the information needed (as you have pointed out correctly). I can extend the function I´ve created to swap that dynamically also in the jena model, but I would like to avoid that once it becomes more complex and I could just look for that information based simply in the input triple (defacto-rest function). I glanced quickly at the run-time implementation to generate the model and the error seems to be exactly that, for DBPedia resources you get the information, but looks like the method protected Model generateModel(Set<String> uris, Set<String> langs)isn't covering Freebase resources. Do we have a FreeBase endpoint somewhere (still working)?

from defacto.

I see 2 options here: I can either extend your script (defacto-rest) to cover Freebase resources (dynamically) once we have the livepoint somewhere or extend mine at defacto-core level and update the jena model as well (the drawback is that, in the end, it would cover just FactBench examples...)

from defacto.

Here is an endpoint: http://factforge.net/sparql. But I don't know what is the best option to fix the issue.

from defacto.

Solved: eba8efb

from defacto.