METRICS-BASED RECOMMENDATION SYSTEM FOR SOFTWARE ENGINEERINGGašparič, Marko (Avtor) Heričko, Marjan (Mentor) Succi, Giancarlo (Komentor) Janes, Andrea (Komentor) recommendation systemsoftware engineeringmetricsThis thesis presents the state-of-the-art of recommendation systems for software engineering. The systematic literature review and the survey of authors of analyzed publications showed that the most successful characteristics of the systems are a modular architecture, that is not distributed, and a reactive behavior. Existing RSSEs are mostly standalone solutions or Eclipse plugins. When developers study their own activities they improve their estimating and planning skills and reduce the number of defects in their work. We developed a prototype that is an Eclipse plugin and visualizes an overall team effort spent per source code artifact, source code metrics, and recommendations for source code improvements. To evaluate the prototype we conducted nine interviews. Interviews showed that users find the functionalities of our system useful and would also like to get more information, particularly about an effort per developer. Additionally, interviews revealed that developers are willing to execute a non-repetitive task to use the system, while they are not willing to leave the development environment to obtain the same information and recommendations. [M. Gašparič]20132013-08-31 20:20:56Magistrsko delo/naloga41582sl