Scenario-based specification languages such as Message Sequence Charts (MSCs) and UML sequence Diagrams (SDs) are popular means for eliciting, documenting and validating software requirements. However, each scenario represents a partial story of a system and needs to be combined for a complete system behaviour. Combining scenarios may introduce unexpected and implied scenarios in the system. These unexpected behaviours, commonly known as Emergent Behaviours (EB) and Implied Scenarios (IS), are often not evident during the design phase but could lead to severely unexpected and irreparable damages during run-time. Identifying and fixing these unwanted flaws during the early design phase can significantly reduce deployment costs and runtime hazards.

This dissertation proposes two novel approaches to detect unexpected behaviours in the early design model using Machine Learning (ML)-based approach and probabilities. Both approaches provide feedback on where the possible flaws exist, guiding developers toward potential solutions. These methods do not synthesize any automata-based behaviour model and can handle comparatively large scale scenario specifications without state space explosion.

While scenario-based specifications are widely accepted in the industry and are well suited to developing initial approximations of intended behavior, natural language such as English, is still considered to be the preferred method of documenting requirement artifacts. To incorporate software requirements written in English within the automated verification process, two independent strategies are discussed. An automatic extraction of Use case Scenarios (UCS) from NL documents followed by generating behaviour models from Use case scenarios in terms of Sequence Diagrams (SD). Additionally, a systematic literature review on the area of handling EB/IS in scenario-based specification is presented to summarize the state-of-the-art of EB/IS detection approaches, as well as to highlight past and current trends, identify open issues, focus on improvement areas and to provide a future roadmap for study in this area.