Similarity for Analogical Software Reuse: A Computational Model

This paper describes a computational model of similarity developed to support analogical software reuse. Similarity is computed from conceptual descriptions of software artifacts of any substance (i.e code, design or specification artifacts). This computation is restricted by an axiomatic framework realizing properties of similarity assessments by humans and analogical reasoning and exploits the semantics of three common conceptual modeling abstractions, namely the classification, the generalization and the attribution. An operationalization of the model is presented together with a prototype implementing it. The consistency of the estimates of the operational model with respect to similarity assessments of software engineers and its recall performance are evaluated, in a preliminary experiment.

• Type: Conference
• Year: 1994
• Where: ECAI
• Authors: George Spanoudakis, Panos Constantopoulos
• Keywords: Analogical Software Reuse, Artificial Intelligence, Conceptual Modeling Abstractions, ECAI 1994, Similarity Assessments