Cloud services categories identification from requirements specifications

In the Cloud Computing field, with the increasing number of Cloud Services available thanks to several cloud providers, looking for a particular service has become very difficult, especially with the evolution of the stakeholders' needs. At the same time requirements specifications have become more and more complex to define in a formal representation and to analyse, since the stakeholders' goals are typically high-level, abstract, and hard-to-measure. For these reasons it would be useful to automate, as much as possible, requirements analysis. In this work we propose an automatic classification and modelling of requirements that are expressed in a natural language form, and an automatic identification of cloud services categories from requirements in order to support the development of a cloud application. Automated requirements analysis is not an easy subject, due to the natural languages variability and ambiguity, that's why different machine/deep learning and natural language processing approaches are used and compared. The target data set is provided by the Open-Security tera-PROMISE repository.