Thesis PDF

Abstract

Recommender systems help people ?nd information and make decisions where they lack experience or cannot consider all data at hand. It is only recently that recommender systems emerge to assist software developers in software reuse, and thereby, save time and improve quality. However, today?s recommender system approaches almost exclusively focus on code reuse and do not consider modeling tasks in model-driven development. Since modeling is time consuming and can introduce errors on a higher level of abstraction, reuse of models speeds up this process and is expected to improve model quality. This thesis proposes a generic recommender system architecture which lays the foundations for various recommendation approaches for modeling. Moreover, a software prototype of a basic recommender system is provided as an Eclipse feature, and operates in Ecore Model and Ecore Diagram editors of the Eclipse Modeling Framework. To this end, an Ecore library serves as a basis to generate recommendations.

Resources

• Thesis PDF