Hi Dave, So in that case we dont need to add non-inference model to the generic reasoner because after all, our whole data is in the model.
On Sat, Nov 26, 2016 at 11:55 AM, Dave Reynolds <dave.e.reyno...@gmail.com> wrote: > On 25/11/16 13:54, tina sani wrote: > >> I have something like this after my Jena forward chain rules and SPARQL >> query: >> >> Reasoner reasoner1 = new GenericRuleReasoner(Rule.parseRules(rule)); >> InfModel inf = ModelFactory.createInfModel(reasoner1, model); >> >> Reasoner reasoner2 = ReasonerRegistry.getRDFSReasoner(); >> >> InfModel inf2 = ModelFactory.createRDFSModel(model); >> > > If you want your rules to see the results of the RDFS closure then you > need something more like: > > InfModel inf = ModelFactory.createRDFSModel(model); > Reasoner reasoner1 = new GenericRuleReasoner(Rule.parseRules(rule)); > InfModel inf2 = ModelFactory.createInfModel(reasoner1, inf); > > Dave > > > On Fri, Nov 25, 2016 at 4:46 PM, tina sani <tinamadri...@gmail.com> wrote: >> >> So all I should so is to create rdfs model: >>> >>> InfModel inf2 = ModelFactory.createRDFSModel(); >>> >>> What would be the arguments of the ModelFactory.createRDFSModel(). I >>> have two models here, one the simple non inference model and other the >>> inference model : InfModel inf = ModelFactory.createInfModel(reasoner, >>> model); >>> >>> >>> >>> >>> On Fri, Nov 25, 2016 at 3:59 PM, Lorenz B. <buehm...@informatik.uni- >>> leipzig.de> wrote: >>> >>> >>>> >>>> Yes I use GenericRuleReasoner, so using the following rules along with >>>>> >>>> my >>>> >>>>> own rules will solve the problem? >>>>> >>>>> ?x rdfs:subClassOf emp:ContractEmployee--> ?x rdfs:subClassOf >>>>> >>>> emp:Employee >>>> >>>>> ?x rdfs:subClassOf emp:PermanantEmployee--> ?x rdfs:subClassOf >>>>> >>>> emp:Employee >>>> That might be incomplete as rdfs:subClassOf is transitive. Simply use >>>> the RDFS reasoning. >>>> >>>>> >>>>> On Thu, Nov 24, 2016 at 11:39 AM, Lorenz B. < >>>>> buehm...@informatik.uni-leipzig.de> wrote: >>>>> >>>>> If I pass OntModelSpec.OWL_MEM_MICRO_RULE_INF” to the OntModel, will >>>>>>> >>>>>> it do >>>>>> >>>>>>> the RDFS reasoning? >>>>>>> >>>>>> See [1] in general and in particular [2] which shows how to setup an >>>>>> RDFS reasoner >>>>>> >>>>>> InfModel inf = ModelFactory.createRDFSModel(rdfsExample); >>>>>> >>>>>> or >>>>>> >>>>>> Reasoner reasoner = ReasonerRegistry.getRDFSReasoner(); >>>>>> InfModel inf = ModelFactory.createInfModel(reasoner, rdfsExample); >>>>>> >>>>>> [1] https://jena.apache.org/documentation/inference/ [2] >>>>>> https://jena.apache.org/documentation/inference/#generalExamples >>>>>> >>>>>>> b) adding the particular rules to your set of rules >>>>>>> I have already created the rules, mentioned in my first email. You >>>>>>> >>>>>> means >>>> >>>>> rules other than that? >>>>>>> >>>>>> I don't know how you setup your rules, but if you use the >>>>>> GenericRuleReasoner it does only apply your rules and you would have >>>>>> to >>>>>> add the rules that do the rdfs:subClass/rdf:type inference. >>>>>> >>>>>>> On Wed, Nov 23, 2016 at 11:50 AM, Lorenz B. < >>>>>>> buehm...@informatik.uni-leipzig.de> wrote: >>>>>>> >>>>>>> Dave meant that you need an additional layer of reasoning like RDFS >>>>>>>> >>>>>>> or >>>> >>>>> even less to get the inferences that you describe. >>>>>>>> This can be done by >>>>>>>> >>>>>>>> a) nesting two InfModels one that works on your specific rules and >>>>>>>> >>>>>>> one >>>> >>>>> that uses some kind of RDFS or >>>>>>>> b) adding the particular rules to your set of rules >>>>>>>> >>>>>>>> Ok sorry Dave, actually I did not completely understand your answer. >>>>>>>>> >>>>>>>>> "You could have an inference model with the appropriate >>>>>>>>> >>>>>>>> configuration >>>> >>>>> to >>>>>> >>>>>>> deduce membership of employee" >>>>>>>>> >>>>>>>>> Yes I have the inference model: >>>>>>>>> Reasoner reasoner = new GenericRuleReasoner(Rule.parse >>>>>>>>> >>>>>>>> Rules(rule)); >>>> >>>>> InfModel infer = ModelFactory.createInfModel(reasoner, >>>>>>>>> >>>>>>>> model); >>>> >>>>> >>>>>>>>> Query query = QueryFactory.create(queryString); >>>>>>>>> >>>>>>>>> QueryExecution qe = QueryExecutionFactory.create(query, >>>>>>>>> >>>>>>>> infer); >>>> >>>>> >>>>>>>>> What additionally I have to do apart from my Jena rule (Forward >>>>>>>>> >>>>>>>> Chain) >>>> >>>>> mentioned in previous email and the inference model. >>>>>>>>> >>>>>>>>> And of course, I will have a SPARQL query : >>>>>>>>> Select * >>>>>>>>> where{ ?x rdf:type emp:QualifiedEmployee } >>>>>>>>> >>>>>>>>> On Tue, Nov 22, 2016 at 6:03 PM, Dave Reynolds < >>>>>>>>> >>>>>>>> dave.e.reyno...@gmail.com> >>>>>>>> >>>>>>>>> wrote: >>>>>>>>> >>>>>>>>> On 22/11/16 14:56, tina sani wrote: >>>>>>>>>> >>>>>>>>>> Let me explain a bit. >>>>>>>>>>> >>>>>>>>>>> String rule = "[rule1:(?x http://www.w3.org/1999/02/22- >>>>>>>>>>> >>>>>>>>>> rdf-syntax-ns#type >>>>>>>> >>>>>>>>> http://www.semanticweb.org#Employee) " >>>>>>>>>>> + "( ?x http://www.semanticweb.org#Salary >>>>>>>>>>> >>>>>>>>>> ?salary >>>> >>>>> )" >>>>>> >>>>>>> + "greaterThan(?salary, 10,00) " >>>>>>>>>>> + " -> (?x http://www.w3.org/1999/02/22-r >>>>>>>>>>> df-syntax-ns#type >>>>>>>>>>> http://www.semanticweb.org#QualifiedEmployee. )]" >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> Classes in my ontology are >>>>>>>>>>> >>>>>>>>>>> Employee (Super class of all employees) >>>>>>>>>>> * ContractEmployee* (Subclass of Employee) >>>>>>>>>>> FinanceManager >>>>>>>>>>> ITManager (Subclasses of ContractEmployee) >>>>>>>>>>> * PermanantEmployee *(Subclass of Employee) >>>>>>>>>>> Analyst >>>>>>>>>>> Programmer >>>>>>>>>>> >>>>>>>>>>> In this case, will I use ?x rdf:type Employee or ?x rdf:type >>>>>>>>>>> ContractEmployee >>>>>>>>>>> All the instances are either from subclasses * FinanceManager , >>>>>>>>>>> >>>>>>>>>> ITManager >>>>>>>> >>>>>>>>> or Analyst, Programmer* >>>>>>>>>>> >>>>>>>>>>> *Is it must that if I use ?x rdf:type Employee, then >>>>>>>>>>> >>>>>>>>>> FinanceManager & >>>> >>>>> ITManager should also be sub classes of general super class >>>>>>>>>>> >>>>>>>>>> "Employee" >>>>>> >>>>>>> ?* >>>>>>>> >>>>>>>>> I don't think any of these details change my earlier answer, quoted >>>>>>>>>> >>>>>>>>> below. >>>>>>>> >>>>>>>>> Dave >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On Tue, Nov 22, 2016 at 5:30 PM, Dave Reynolds < >>>>>>>>>> >>>>>>>>> dave.e.reyno...@gmail.com> >>>>>>>> >>>>>>>>> wrote: >>>>>>>>>>> >>>>>>>>>>> On 22/11/16 10:12, tina sani wrote: >>>>>>>>>>> >>>>>>>>>>>> Inline image 1 >>>>>>>>>>>> The mail list doesn't support attachments so the image didn't >>>>>>>>>>>> >>>>>>>>>>> come >>>> >>>>> through. >>>>>>>>>>>> >>>>>>>>>>>> However, I'm guessing it included a class Employee with >>>>>>>>>>>> >>>>>>>>>>> sub-classes >>>> >>>>> ContractEmployee and PermanantEmployee. >>>>>>>>>>>> >>>>>>>>>>>> I have this ontology: Now if I want to use some rules like >>>>>>>>>>>> >>>>>>>>>>>> if x rdf:type Employee and ?x salary>Euro10,000. then ?x >>>>>>>>>>>>> QualifiedEmployee. >>>>>>>>>>>>> >>>>>>>>>>>>> My question here is should I use ?x rdf:type Employee or ?x >>>>>>>>>>>>> >>>>>>>>>>>> rdf:type >>>>>> >>>>>>> ContractEmployee or PermanantEmployee >>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> Depends on your set up. >>>>>>>>>>>> >>>>>>>>>>>> You could have an inference model with the appropriate >>>>>>>>>>>> >>>>>>>>>>> configuration >>>> >>>>> to >>>>>>>> >>>>>>>>> deduce membership of employee and then query that with SPARQL or >>>>>>>>>>>> >>>>>>>>>>> with >>>>>> >>>>>>> a >>>>>>>> >>>>>>>>> second inference model with your own rules in. >>>>>>>>>>>> >>>>>>>>>>>> If you want just one layer of rules and want to combine OWL/RDFS >>>>>>>>>>>> >>>>>>>>>>> with >>>>>> >>>>>>> your >>>>>>>>>>>> custom rules then that's possible (so long as you set all the >>>>>>>>>>>> >>>>>>>>>>> appropriate >>>>>>>> >>>>>>>>> flags, see documentation) but make sure that your own rules are >>>>>>>>>>>> >>>>>>>>>>> backward >>>>>>>> >>>>>>>>> not forward rules. [The default Jena rule sets for RDFS and OWL >>>>>>>>>>>> >>>>>>>>>>> are >>>> >>>>> hybrid >>>>>>>>>>>> rules so some of the inferences are only available to backward >>>>>>>>>>>> >>>>>>>>>>> rules >>>> >>>>> in >>>>>>>> >>>>>>>>> the >>>>>>>>>>>> same rule set.] >>>>>>>>>>>> >>>>>>>>>>>> Or if you don't want the cost of full inference then you can >>>>>>>>>>>> >>>>>>>>>>> indeed >>>> >>>>> rewrite the "natural" query to explicitly check for the base >>>>>>>>>>>> >>>>>>>>>>> memberships. >>>>>>>> >>>>>>>>> Dave >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> -- >>>>>>>> Lorenz Bühmann >>>>>>>> AKSW group, University of Leipzig >>>>>>>> Group: http://aksw.org - semantic web research center >>>>>>>> >>>>>>>> >>>>>>>> -- >>>>>> Lorenz Bühmann >>>>>> AKSW group, University of Leipzig >>>>>> Group: http://aksw.org - semantic web research center >>>>>> >>>>>> -- >>>> Lorenz Bühmann >>>> AKSW group, University of Leipzig >>>> Group: http://aksw.org - semantic web research center >>>> >>>> >>>> >>> >> >
