This paper proposes a technique for dynamic slicing of UML architectural models. The presence of related information in diverse model parts (or fragments) makes dynamic slicing of Unified Modeling Language (UML) models a complex problem. We first extract all relevant information from a UML model specifying a software architecture into an intermediate representation, which we call a Model Dependency Graph (MDG). For a given slicing criterion, our slicing algorithm traverses the constructed MDG to identify the relevant model parts that are directly or indirectly affected during the execution of a specified scenario. One novelty of our approach is computation of dynamic slice based on the structural and behavioral (interactions only) UML models as against independently processing separate UML models, and determining the implicit interdependencies among different model elements distributed across model views. We also briefly discuss a prototype tool named Archlice, which we have developed to implement our algorithm.

This paper proposes a novel dynamic slicing technique for object oriented concurrent programs. We introduce the notion of object oriented concurrent program dependence graph (OOCPDG). Our dynamic slicing technique uses OOCPDG as the intermediate representation and is based on marking and unmarking the dependence edges as and when the dependences arise and cease at runtime. Our approach eliminates the use of trace files and is more efficient than existing algorithms. Besides, it encompasses different aspects of object oriented programming paradigm viz. inheritance, polymorphism from the slicing arena. It can handle dynamically created object based processes. It can also handle process interactions through shared memory and message passing. The updating to the intermediate representation is truly concurrent. Multiple processors execute different object based processes concurrently and require special handling. We also report a dynamic slicing tool called CDSOOCP (concurrent dynamic sheer for object oriented concurrent programs) which implements our dynamic slicing technique.