NLForSpec: Uma Ferramenta para GeraÃÃo de EspecificaÃÃes Formais a partir de Casos de Teste em Linguagem Natural

AUTOR(ES)
DATA DE PUBLICAÇÃO

2006

RESUMO

This work describes the NLForSpec, a Natural Language Processing tool to translate software test cases descriptions in English into a formal representation. NLForSpec is part of a larger project (CIn-Motorola BTC Test Research Project), which aims to automate test case generation, selection and evaluation for mobile phone applications. One of the projects main goals is to provide for automatic update of requirements documents from more up-to-date test cases (since test cases change more often than requirements). In this scenario, the NLForSpec tool is the first step in the translation process from test case descriptions into formal representations used to update or generate requirements documents from more up-to-date test cases. NLForSpec generates specifications in CSP (Communicating Sequential Processes) from English test cases descriptions. NLForSpec was built based on the traditional pipeline NL interpretation architecture. The input sentence is parsed and then mapped into case grammar structures (based on thematic roles). These structures are then mapped into representation in CSP formal language. The CSP representations are used as input by another tool of the major project. NLForSpec reached satisfactory results in a case study performed in Motorola Messaging domain

ASSUNTO(S)

symbolic artificial intelligence ciencia da computacao interpretaÃÃo de linguagem natural inteligÃncia artificial simbÃlica natural language interpretation test cases specifications geraÃÃo de especificaÃÃes de casos de teste

ACESSO AO ARTIGO

Documentos Relacionados