Yazılım gereklilikleri öncelikleri için metin madencilik sınıflandırma uygulaması

Software product Priorities are needed in order to discover the most important parts of the product to be developed at first, as the capacity of the development team is constrained by the number of the workers involved, the experience of the individuals, the same as the good coordination of the whole team and the necessity to detect the conflicts among the requirements and planning the sequence of the requirements to be implemented and released. In the this paper , we have review the application of requirements prioritization automation by means of machine learning to the open source applications and software.As the development in the field of ML proceeds, there are many possibilities to utilize the advancements. One of them is the automation of analytic work performed up to now by humans. Requirements prioritization is one of them. It is time consuming and knowledge demanding activity and with the growing number of requirements it can get easily unfeasible for a human being to evaluate every requirement. The first part of this study is the literature review and related works in the field requirements prioritization automation. the second part we have reviewed some text mining background, the third discusses the relevant mechanics and problems of the open source software (OSS) projects.We have analyzed the options for solving the problems of OSS projects requirements management with particular emphasis on the ASF OSS projects, which are interesting with their open development approach. in the fourth part, we propose a system design, using the synthetic methodology of combining different known approaches to solve the particular problems of the OSS projects and the automation of the requirements prioritization process. the fifth part is devoted to the evaluation of the proposed system design with a prototype on apache software foundation (ASF) hadoop project.The proposed solution and evaluation is valid only in the context of one particular foundation of projects (ASF) and one project (Hadoop). The evaluation outcome cannot be generalized, since the fine tuning of the algorithms would require enormous effort. Using another project would mean the refitting of the whole proposed solution. Keywords:-Requirements-Prioritization-OpenSource-Software-Apache-Software_Foundation Machine Learning Text Mining, Support Vector Machine,Support Vector Regression.