A Machine Learning Approach for Identifying Expert Stakeholders

Abstract

Requirements gathering, analysis, and specification are humanintensive activities that rely upon finding and engaging a relevant set of informed stakeholders. In many projects initial requirements are captured through the use of wikis or forums, or through ini tial facetoface brainstorming meetings. In this paper we introduce a technique for analyzing stakeholders' contributions, extracting domain topics, and construct ing profiles which depict stakeholders' interests in each of the topics. Content and collaborative filtering techniques are then used to identify a diverse set of stakeholders for a given topic. The approach, which can be used to support requirements related activities throughout the software development lifecycle, is illus trated through an example of an Amazonlike student webportal.