VoiceToModel

Abstract

Requirements are typically modelled using some graphical notation with the aid of CASE tools. However, these tools make this activity exclusive to requirements engineers and stakeholders with no accessibility problems, but disabled professionals often have accessibility problems to use tools to build requirements models. Also, accessibility problems can happen to stakeholders in general in a particular context where the use of voice is more convenient (e.g., in a public transport). In this paper, the VoiceToModel framework is proposed to improve the accessibility of the requirements process by effectively integrating a requirements engineer or stakeholder with disabilities during requirements modelling. Through speech recognition, the engineer should be capable of voicing the requirements, which are automatically used to generate requirements models, more specifically goal-oriented models, object models and feature models. Also, the automated framework generates requirements models from speech recognition by firstly defining a grammar where we have the commands to create and modify a model. These commands return a feedback via voice. We show the applicability of the framework through an example and discuss the results of an experiment with 14 participants, including 2 blind people.

Survey on Challenges and Achievements in Context-Aware Requirement Modeling

“…Model-based methods always support the modeling of multiformat requirements by establishing a self-defined template or some pre-existing description standard. Based on this foundation, the requirements with different formats, such as requirements specifications, manuscripts, and results of questionnaires, are recorded, analyzed, validated, and maintained in a unique way automatically or semiautomatically [40]- [43]. For example, in [42], a requirement abstract model is proposed to integrate the conceptual graph, use case diagram and Vienna Development Method Specification Language (VDM-SL) in order to model requirements in different formats; in another study [43], a selfdefined model named VoiceToModel is presented to model requirements in the audio format.…”

“…Format diversity: Because the requirements of context-aware systems originate from different sources, the format of requirements also has great variety. For example, the requirements may hide in audio, video, text, and various monitoring data [40]- [43]. How to elicit and model the requirements in different formats is also a difficult task.…”