Recursos do processamento da língua natural aplicados na recuperação semântica de documentos de caso de uso

Abstract

The Requirements Engineering basically deals with how to discover, analyze, register and verify the functions and restrictions that software must consider. In this process the designer not only concentrates in understanding the necessities, goals and certainties of the interested users but also in changing them into software devices. This process is known as development cycle and it is carried out until the software covers all the requirements of the involved users. This study describes how the resources of the natural language processing were used in the construction for a solution of semantics recovery of use case document and it also presents the reached findings. For the construction of the solution, it is specified a method that organizes the preparation and recovery works in two phases. The first describes the form how the corpus must be prepared and how the terms used in the preparation phase can be used in the definition of the keys concepts of the domain. The second phase explains how the document recovery is carried out and shows how the described relationships in the ontology are used to improve the results of the recovery. The presented findings reveal the described method in this study is efficient, since it presented a covering of 100% in both tests. Related of measure of precision, that presented an inferior result of 50%, it was compensated by the ranking algorithm that sorted the documents of similar form of the manual classification done by the users.