RESUMOGrande parte das empresas desenvolvedoras de software criou seu próprio processo de trabalho. Devido à rápida expansão do mercado de software, a concorrência ocorre muito mais em custo do que em diferenciação. Para obter vantagem competitiva, as empresas devem atualizar-se continuamente na tecnologia, buscar a maturidade nos processos e eliminar a inefi ciência operacional. Isso requer um envolvimento das pessoas, dos processos e da organização como um todo. O artigo discute a implementação de melhorias nos processos de software segundo os principais modelos de qualidade e de maturidade. Com base em um Estudo de Casos Múltiplos, verifi ca-se que a melhoria dos processos de software requer que a melhoria ocorra primeiramente entre cada um dos desenvolvedores e, a seguir, envolva os grupos de desenvolvimento e por fi m, a organização como um todo. A pesquisa conclui que os modelos de qualidade e maturidade servem como orientadores do processo de melhoria. PALAVRAS-CHAVE

No abstract

Six Sigma emphasizes the variability causes search. The special causes are more evident, which favors their identification, while the common causes are more difficult to be identified and solved and, for that, they request a specialists' participation. Many software development process variability causes are common and it is not always possible to have Six Sigma specialists to identify them. An alternative might be the visibility criteria proposed by Lean Production. Instead of this criterion, software engineers are able to more easily notice the mistakes and their causes than Six Sigma specialists. This article verifies the advantages of using Lean Production practices, based on a Multiple Case Study on some software organizations that uses this methodology. The study shows that the projects were finished more quickly and they presented more effective results.

Aspects related to the users' cooperative work are not considered in the traditional approach of software engineering, since the user is viewed independently of his/her workplace environment or group, with the individual model generalized to the study of collective behavior of all users. This work proposes a process for software requirements to address issues involving cooperative work in information systems that provide distributed coordination in the users' actions and the communication among them occurs indirectly through the data entered while using the software. To achieve this goal, this research uses ergonomics, the 3C cooperation model, awareness and software engineering concepts. Action-research is used as a research methodology applied in three cycles during the development of a corporate workflow system in a technological research company. This article discusses the third cycle, which corresponds to the process that deals with the refinement of the cooperative work requirements with the software in actual use in the workplace, where the inclusion of a computer system changes the users' workplace, from the face to face interaction to the interaction mediated by the software. The results showed that the highest degree of users' awareness about their activities and other system users contribute to a decrease in their errors and in the inappropriate use of the system.