Page - 176 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

,whereUP /∈FP∪FM. Itmayseemoddthat thedispatcherreturnsanunknownresultUP. Incausalmodeling,weexpresswithanunknownvariableUP thatwedonotexplainhow it is caused. OutgoingmessageM′. Previously, thedispatcherPhas selected thecorresponding lo- cal operation according to the operation specifiedwithin themessageM. The outgoing messageM′ contains the result of the executionof theoperationop,whichwedefineas the followingequation. M′ := fM′(P,M,UM′) := { R , ifP=op∧op∈M UM′ , otherwise ,whereR=op(M) is the result of the operation op taking the arguments from the in- comingmessageM. If opwas not known to the dispatcher, the result is the unknown variableUM′whichsays thatwedonotdetermine itsvalue fromthecausalmodel. Discriminator D. Finally, the discriminator identifies whether the outgoing mes- sageM′ is theoperation’s result or anunknowndispatcher’s result. D := fD(M′,UD) := { 1 , ifM′=R 0 ,otherwise The discriminator returns 1, ifM′ results from the operation execution. Therewith, the discriminatorestablishes thedependencebetweenincomingandoutgoingmessage.Note thatwedonotneedtoknowtheconcrete resultof theoperationexecution. In thismodel, theoperation resultRonly identifies itsorigin. Graphicalmodel. Finally, everySCMis associatedwith a graphical causalmodel, or simplyagraph.Thegraph’snodes represent thevariablesV andU from table1and the edges represent the structural equation inE. The figure 3 depicts the graphical causal modelof themessagepassingbehaviorof interoperableprocesses. Figure3. Graphical causalmodelof themessagepassingbehaviorof interoperableprocesses S. Kotstein and C. Decker /AnApproach for Measuring IoT Interoperability176