\section{Semantics}
\label{semantics}

Finding the explanations for a query is important for probabilistic inference. In the following we briefly describe the DISPONTE semantics~\cite{RigBelLamZes15-SW-IJ}, which requires the set of all the justifications to compute the probability of the queries.

DISPONTE \cite{RigBelLamZes15-SW-IJ,Zese17-SSW-BK} applies the distribution semantics \cite{DBLP:conf/iclp/Sato95} to Probabilistic Description Logic KBs.
In DISPONTE, a \emph{probabilistic knowledge base} $\cK$ contains a set of \emph{probabilistic axioms} which take the form
\begin{align}
& p::E\label{pax} &
\end{align}
where $p$ is a real number in $[0,1]$ and $E$ is a DL axiom. 
The probability $p$ can be interpreted as the degree of our belief in the truth of axiom $E$. 
For example, a probabilistic concept membership axiom
$
p::a:C
$
means that we have degree of belief $p$ in $C(a)$.
A probabilistic concept inclusion axiom of the form
$
p::C\sqsubseteq D
$
represents the fact that  we believe in the truth of $C \sqsubseteq D$ with probability $p$. 

For more detail about probabilistic inference with the TRILL framework, we refer the interested reader to Appendix~\ref{app:disponte} and to~\cite{ZesBelCot18-TPLP-IJ}.

The following example illustrates inference under the DISPONTE semantics.
\begin{example}
	
	\begin{align*}
	&& (E_1)\ 0.5 ::\ & \exists hasAnimal.Pet \sqsubseteq PetOwner\\
	&& 				& \fluffy: Cat \\
	&& 				& tom: Cat \\
	&& (E_2)\ 0.6 ::\ & Cat\sqsubseteq Pet\\
	&& 				& (kevin,\fluffy):hasAnimal \\
	&& 				& (kevin,tom):hasAnimal
	\end{align*}
	It indicates that the individuals that own an animal which is a pet are pet owners with a 50\% probability and that $kevin$ owns the animals $\fluffy$ and $tom$, which are cats.
	Moreover, cats are pets with a 60\% probability.

	The query axiom $Q=kevin:PetOwner$ is true with probability $P(Q)=0.5\cdot 0.6=0.3$.
\end{example}

the translation of this KB into the TRILL syntax is:
\begin{verbatim}
subClassOf(someValuesFrom(hasAnimal, pet), petOwner).
annotationAssertion(disponte:probability,
                    subClassOf(someValuesFrom(hasAnimal, pet), petOwner),
                    literal('0.5'))
classAssertion(cat, fluffy).
classAssertion(cat, tom).
subClassOf(cat, pet).
annotationAssertion(disponte:probability, subClassOf(cat, pet), literal('0.6'))
propertyAssertion(hasAnimal, kevin, fluffy).
propertyAssertion(hasAnimal, kevin, tom).
\end{verbatim}
Optionally, the KB can also contain the following axioms
\begin{verbatim}
namedIndividual(fluffy).
namedIndividual(kevin).
namedIndividual(tom).
objectProperty(hasAnimal).
annotationProperty('http://ml.unife.it/disponte#probability').
class(petOwner).
class(pet).
\end{verbatim}