search

SmartDataAnalytics / DL-Learner

A tool for supervised Machine Learning in OWL and Description Logics
http://dl-learner.org
GNU General Public License v3.0
152 stars 34 forks source link

Not being able to assign object of N-triples as a positive example #64

Closed Demirrr closed 6 years ago

Demirrr commented 6 years ago

Hello,

I would like to employ DL-Learner in order to learn class expression of my positive example. To this end, the subset of dbpedia 2016-10 is being used as one may very well see in the below.

screen shot 2018-07-13 at 15 58 22

I would like to know why it is not possible to assign predicate or object of N-triples as positive examples.

ks.fileName = "spring_KnowledgeBase.owl" reasoner.type = "closed world reasoner" reasoner.sources = {ks} lp.type = "posonlylp" lp.positiveExamples ={ some URIs} alg.type = "celoe" alg.maxExecutionTimeInSeconds = 1

Any suggestion appreciated :)

SimonBin commented 6 years ago

hi, this is a bug tracker. please join our mailing list https://www.freelists.org/list/dl-learner-discussion . You should be able to use any Named Individual as positive example, so this is the first thing I would check

LorenzBuehmann commented 6 years ago

As a comment, most algorithm of the DL-Learner learn descriptions of a set of individuals. As long as the objects of the triples are individuals, it will work. For literals indeed it doesn't make sense or better said, we don't have nay algorithm so far given that describing a set of literals is somewhat limited in OWL.

I saw that you used data from DBpedia, if so, don't forget to add the DBpedia ontology triples as well, otherwise the OWL API parser isn't able to determine the type of the properties and each property is treated as annotation property, thus, DL-Learner won't consider them as candidates to be used.

But let's continue on the mailing list...