The pdf version of the manual can be downloaded at the end of this page in the download section.


BUNDLE performs probabilistic inference on ontologies that follow the semantics DISPONTE. BUNDLE finds explanations for the query and then builds a Binary Decision Diagram to represent the explanations and to calculate the probability. Explanations are found using a modified versione of the Pellet explain function.

The system can take as input an ontology defined in various OWL formats, including RDFXML. It can compute the probability of all the queries that are allowed by Pellet explain function.

In order to assign probabilities to axioms, BUNDLE takes advantage of the opportunity given by OWL 2 to declare annotations to the axioms. We have defined different annotations to specify the different types of probability used. Each of these annotations is then associated with a real number in the range [0,1] which specifies the probability.

The annotations are:
  • bundle:probability  for epistemic probability;

BUNDLE takes as input one or two RDFXML files. In the first case, the single input file contains the ontology and the annotations for the axioms too. In the second case, the first file contains the full ontology and the second one contains all the annotations. In this case, the annotations contained in the first file are not used.

An example of annotated axiom is shown below:

For modeling the fact that the class A is subclass of the class B, the ontology have to contain the following axiom:

<owl:Class rdf:about="#A">
<owl:Class rdf:about="#B" />

To assign one or more probability annotations, just add the following to the end of ontology file or in the second file:

<rdf:subject rdf:resource="#A"/>
<rdf:predicate rdf:resource="&rdfs;subClassOf"/>
<rdf:object rdf:resource="#B"/>

In this case, we associate to the axiom A subclassof B an epistemic probability of 0.8.


bundle.sh <arguments>

The program accepts the following arguments:
  • -gui, --gui, gui use the graphic interface;
  • -help, -h, help show the help with all the argument;
  • -v  print detailed exceptions and messages about the progress
  • -m, --max maximum number of generated explanations
  • -t, --time maximum time allowed for the inference, 0 for unlimited time. Format: [Xh][Ym][Zs][Kms]
  • --noProb disable the computation of the probability

In textual mode, the queries are specified as follows:

[bundle-args] --<query-type> <query-arguments> <files>

where <query-type> can be:
  • subclass
  • instance
  • property-value
  • unsat
  • all-unsat
  • inconsistent
  • hierarchy

<query-arguments> depends on <query-type>:
  1. for query subclass, <query-arguments> are two class contained in the ontology: class1,class2;
  2. for query instance, <query-arguments> are an individual and a class: individual,class;

  3. for query property-value, <query-arguments> are two individuals and an object property: individual1,property,individual2;

  4. for query unsat, <query-arguments> are a class class;

  5. for query all-unsat, inconsistent and hierarchy, <query-arguments> will be empty.

Finally, <files> indicates a single file or two files:
  • file:<ontology-file-with-annotation.owl> in case of a single file;

  • file:<ontology-file-without-annotation.owl> file:<annotations-file.owl> in case of two files.

Riccardo Zese,
Dec 16, 2013, 7:23 AM