Uma abordagem para criação de casos de teste funcionais para sistemas de conhecimento

• Autor(a) principal: Merlin, José Reinaldo
• Data de Publicação: 2011
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Estadual de Maringá (RI-UEM)
• Texto Completo: http://repositorio.uem.br:8080/jspui/handle/1/2571
• topic: Sistema de conhecimento; Ontologia; Recuperação de informação; Teste de Software; Particionamento de Equivalência; Brasil.; Software Testing; Ontology; Equivalence Partitioning; Brazil.; Ciências Exatas e da Terra; Ciência da Computação

Software testing is an important activity of quality assurance. To ensure that tests be well conducted is necessary to use testing criteria. The criteria are systematic approaches to the design of tests. Each criterion uses different procedures for selecting test data. The criteria also should be adapted for each type of software under test. Thus, a critical system software requires a different test procedure that a software for electronic commerce control. The object of this work are software programs that manipulate knowledge bases expressed in OWL (Web Ontology Language). This type of software requires a different approach because, unlike conventional programs, knowledge is separated from the algorithm that handles. Most often, these two elements were developed by different people, because the ontology is defined by an ontology engineer or specialist in the domain, while the program is developed by a software engineer. Problems can occur when, although the ontology is correct, the programmer has a misunderstanding about what should be done. This work presents an analysis of the characteristics of these programs and proposes an approach to creating functional tests based on Equivalence Partitioning Criterion. The criterion divides the domain of input (or output) of a field or a function on sets of valid and invalid data. During the definition of test cases, representative data from different sets are used in order to avoid redundancy and prevent that important data be left out. The approach was applied to test a prototype that queries an ontology and the results showed that this criterion, considered limited to conventional programs, is appropriate to software testing subject of this research, because it was possible to write test cases based the criterion.