Using explanations to improve the validation of executable models

Abstract

Executable modeling languages are now found in several commercial and experimental CASE environments. The ability to execute conceptual models has been considered vital for the validation of their dynamic properties. However, during model execution only observable aspects can be validated, and it can be very hard to check whether the internal computations are modeled correctly. Traces provide some of the informations necessary to understand the model's behavior in depth, but they are often just as difficult to read as the conceptual models themselves.In this paper we show how explanation generation technology can be combined with the execution of conceptual models. A component that can explain the internal reasoning of the model as well as its user interaction is presented, and it is also indicated how the component is being implemented in the PPP CASE environment. Using this integrated validation approach, we are able to expose both the external and internal reasoning of executable conceptual models.

A general explanation component for conceptual modeling in CASE environments

“…As the elements of the conceptual model are treated as instances of metamodel elements, they inherit their reference predicates from their counterparts in the metamodel. For the representation of elements of the Conceptual Modeling in CASE Environments • execution trace, the reader is referred to Gulla and Willumsen [1993] for a more thorough discussion.…”

“…Typically, it would say that Account_balance is retrieved from the Accounts database, whereas Withdrawal_rejection is received from Customer. A more thorough discussion of model execution and their explanations is given in Gulla and Willumsen [1993].…”

Testing and test-driven development of conceptual schemas

“…Explanation generation is another research direction to contribute to the validation of conceptual schemas. (Gulla et al 1993) describes that "explanation generation components can form a question-answer facility, where explanations are user tailored and may include graphical model views in addition to the textual descriptions".…”

“…As we have seen in the state of the art on conceptual schema validation presented in Section 2.2, several existing techniques are used to achieve correct and complete conceptual schemas in different degrees. These include paraphrasing in natural language (Rolland et al 1992, Dalianis 1992, Métais 2002, Frederiks et al 2006, generation of abstractions and abstracts (Jesus et al 1992), explanation generation (Gulla et al 1993, Olivé et al 1996, Gulla 1996, constraint acquisition (Hartmann et al 2009) or simulation and animation (Van Lamsweerde 2009, Lindland et al 1994, Dignum et al 1987, Gogolla et al 1995, Gogolla et al 2007.…”