Please use this identifier to cite or link to this item:
|Type:||Artigo de evento|
|Title:||Assessing The Relationship Between Size-complexity And Number Of Defects Of Software In Module Level [avaliando A Relação Entre Tamanho-complexidade E Número De Defeitos De Software Em Nível De Módulo]|
|Author:||De Lucca W.L.|
|Abstract:||A challenge in software testing research area is presented by the question: how do we reduce time and effort to test software without reducing the effectiveness? In this paper, we present the results of an experiment with a program that contains real defects. The data collection have been performed in module-level, rather than statement-level, such as what was done in a previous work. Our primary hypothesis is that the fault-revealing rate is greater when larger or more complex modules are tested first. We demonstrate that testing modules ordered by size or complexity metrics reduce the number of modules to be tested in 6,2% to 17,7%. A secondary hypothesis was formulated: A larger or more complex module has higher probability of containing a defect than smaller or less complex module. We observed that the probability of a module containing a defect is 43,8% to 78,7% greater in modules with size or complexity above the average. Our empirical results support both hypothesis and are an important reference to new experiments and studies that intend to develop new techniques and strategies to reduce the cost of software testing.|
|Appears in Collections:||Unicamp - Artigos e Outros Documentos|
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.