Ulteriori informazioni
Software engineering consists of human-based, knowledge-intensive activities in which new situations require new knowledge by the software engineers. An experience factory supports these activities through the collection, analysis, packaging, and dissemination of so-called experiences packages (i.e., knowledge, products, and processes). However, several explorative studies confirm that reuse-based approaches suffer from three problems in practice: bad understanding of reusable artifacts and experience packages in particular; no explicit support for the internalization of knowledge and no compliance with human information processing; and no explicit connection between experience management and technology-enhanced learning approaches. This work addresses the research question of whether the enrichment of experience packages with additional information (i.e., so-called learning spaces) improves the understanding and application of an experience package on the one hand and knowledge acquisition and perceived information quality on the other hand. The presented learning space approach extends the "project support" activity of the experience factory by automatically generating contextaware learning spaces by merging information from the experience base with learning content. Specified variabilities in generic learning space artifacts support adaptation on the level of structure, content, and presentation to context characteristics. The main contributions are a reference model consisting of a) a context model for describing situations in software engineering, b) a domain model for describing the body of knowledge in software engineering, c) a learning space model for defining learning spaces on different levels of abstraction (i.e., structure, content, and presentation), d) a variability model for defining variabilities in generic artifacts and their resolution, e) a role model for implementing the learning space approach in an organization, and f) techniques and tools for the systematic and automatic ondemand generation of learning spaces (i.e., resolution, adaptation, and presentation). A controlled experiment and a case study provide statistically significant results, which quantify the positive impact of learning spaces upon the understanding and application of experience packages, knowledge acquisition, perceived information quality of experience packages, as well as the use, acceptance, and software ergonomics of the developed tools. A power analysis and effect sizes provide a strong baseline for future evaluations and meta-analysis studies. Software engineering consists of human-based, knowledge-intensive activities in which new situations require new knowledge by the software engineers. Several explorative studies confirm that reuse-based approaches suffer from three problems in practice: bad understanding of reusable artifacts and experience packages in particular, no explicit support for the internalization of knowledge and no explicit connection between experience management and technology-enhanced learning approaches. The presented learning space approach automatically generates so-called context-aware learning spaces by merging information from an experience base with learning content. Specified variabilities in generic learning space artifacts support adaptation on the level of structure, content and presentation to context characteristics. A controlled experiment and a case study provide statistically significant results, which quantify the positive impact of learning spaces upon the understanding and application of experience packages, knowledge acquisition (i.e. learning), perceived information quality of experience packages, as well as the use, acceptance and software ergonomics of the developed tools.
Info autore
Prof. Dr. Dieter Rombach - und Prof. Dr. Peter Liggesmeyer - sind Lehrstuhlinhaber für Software Engineering an der TU Kaiserslautern. Sie leiten gemeinsam das Fraunhofer Institut für Experimentelles Software Engineering.
Prof. Dr. Peter Liggesmeyer - und Prof. Dr. Dieter Rombach - sind Lehrstuhlinhaber für Software Engineering an der TU Kaiserslautern. Sie leiten gemeinsam das Fraunhofer Institut für Experimentelles Software Engineering.
