Scenario Networks and Formalization for Scenario Management

Abstract

Scenarios are widely used to specify the behavior of software due to their informality and accessibility. However, their informality makes them difficult to analyze and manage. We address these difficulties with two complementary approaches, one syntactic and one semantic, that add a small amount of structure to scenarios to allow automated analyses and support. The syntactic approach represents a scenario as a set of attribute-value pairs, some of which may also be viewed as events, each of which is an actor-action pair, that are arranged in a sequence. This representation supports the use of episodes (shared subsequences of events) to show dependency relationships between scenarios and to help maintain those relationships as the scenarios evolve. The representation also supports automated measures of similarity between scenarios, to find duplicates or near-duplicates, searching in a collection of scenarios, and assess requirements coverage and completeness of the collection. The representation can be analyzed for consistency of various attributes within individual scenarios. The semantic approach integrates the scenarios that describe a system into a network that expresses which scenarios can follow each other. The network expresses the context expected by the events of each scenario and the temporal relationships between the scenarios. This information is either implicit or incomplete for an ordinary collection of scenarios. Construction of a scenario network provides process guidance for assessing and improving completeness and consistency of the scenario collection. A scenario network represents equivalence relationships between scenarios, and these relationships can be used to organize and classify the scenarios and to maintain the temporal relationships between scenarios as the scenarios evolve. A scenario network can be analyzed to evaluate completeness of the scenario collection and several kinds of consistency between scenarios in the collection. Together the syntactic and semantic approaches form an effective approach for addressing the scenario management problem, which has not been effectively addresses heretofore.

Description

Keywords

scenario relationships, scenario equivalence, similarity measure, scenario management, software engineering, requirements engineering, Scenarios

Citation

Degree

PhD

Discipline

Computer Science

Collections