Suchergebnisse

Context: In order to ensure usability, it is necessary to schedule activities and methods to be applied throughout different stages of the development process. There exists a substantial number of usability methods to be applied in user-centered software development. However, the application of each usability method largely depends on specific constraints that should be closely considered. Even so, these constraints are not always known beforehand, remaining unidentified or under uncertainty at early stages of the project. Objective: This paper presents an approach to automatically recommend 43 usability methods depending on the project's stage and constraints. Our approach deals with uncertainty to recommend usability methods regardless of the completeness of the information available, which makes it suitable for enhancing initial scheduling. Besides, a supporting tool intended to schedule and guide on usability methods is presented in order to systematize the recommendation mechanism. Method: To validate our approach, we present two application scenarios demonstrating the suitability of the mechanism, including also an expert analysis to observe the recommendation appropriateness in terms of recommendation gap. Also, a user testing was accomplished to evaluate the usability of the approach with key users. Results: A low recommendation gap was observed (<2.5%) and, according to the results obtained in the user testing, high percentage values for usefulness (82.38%) and satisfaction (87.89%) were obtained. The user evaluation also reported high values concerning other dimensions such as ease of use (89.00%) and ease of learning (92.38%). Conclusions: Results obtained helped answer main research questions, demonstrating that it is possible to create a mechanism to recommend usability methods according to a software project's constraints, even under uncertainty, and also affirm that it is possible to systemize the recommendations with a scheduling tool being satisfactory for key stakeholders, denoting acceptable levels of recommendation appropriateness, usefulness, and overall usability. ; This work was partially supported by the Spanish Government [grant number TIN2014-52129-R]; and the Madrid Research Council [grant number S2013/ICE-2715]