Do Rich Media Scenarios Support Requirements Discovery? T.H. Pang, N.A.M. Maiden, K. Zachos & C. Ncube Abstract. This paper reports an exploration into the effectiveness of rich media scenarios on requirements discovery. It reports recent extensions to the ART- SCENE scenario environment to generate scenarios that combine structured text with graphic, image, video and audio documents to support requirements discovery. It presents results from a controlled empirical study and industrial application in which ART-SCENE was used to discover requirements for a new medical system in a hospital to explore 3 questions about the cost-effectiveness of rich media scenarios. Findings revealed that rich media scenarios can lead to the discovery of new requirements during scenario walkthroughs, but the cost of capturing data for and setting up the scenarios was high, and the totals of requirements documented with rich media scenarios was not higher than with structured text scenarios. The paper ends by offering possible explanations of the results and outlining future work. 1. Rich Media Scenarios Walking through scenarios is an effective technique for discovering requirements [Weidenhaupt et al. 1998]. However, determining the right form of scenarios for different requirements tasks remains an open question that we are investigating with ART-SCENE. ART-SCENE is an internet-based environment for generating and walking through structured text scenarios that we have successfully applied to discover requirements for air traffic management systems [Mavin & Maiden 2003]. However, we believe that adding rich media representations of scenario events – visual, video and audio – to the structured text scenarios can lead to more complete requirements discovery. Scenario walkthroughs are effective because, in simple terms, people are better at identifying errors of commission rather than omission [Baddeley 1990], that is they are better at recognising rather than recalling events that a system will need to handle. The ART-SCENE environment generates scenarios that provide recognition cues, in the form of automatically generated text events, which stakeholders can use to discover new requirements. We conjecture that rich media scenarios will offer more recognition cues than text scenarios to stakeholders. Whereas text scenarios describe Centre for HCI Design, City University Northampton Square, London EC1V OHB, UK 1 Hammersmith Hospital, London, UK