Busca em repositório de artefatos de software utilizando raciocínio baseado em casos

Abstract

Reuse in software development has as its main goals promoting the saving of resources in the development of projects and increasing their quality. Software artifacts are defined as any of the storable products resulting of the activities pertaining to the software’s life cycle. The saving of resources is due to the non-necessity of developing artifacts that were already implemented in other projects. The quality comes of the previous use of the artifacts in other projects, ensuring that they have been more thoroughly tested. There are two relevant factors in relation to the proposal of solutions that promote reuse: the storage and retrieval of artifacts. The storage refers to the characterization that is associated to the artifacts’ metadata. The retrieval relates to the search’s techniques and procedures. In the automation of artifact repositories, those techniques and procedures are implemented and their algorithms are denominated search engines. Techniques characterized as Artificial Intelligence are being employed in order to promote increased effectiveness for such engines. In order to contribute in the attenuation of problems in the software development by the artifact reuse, this research presents the fundamentals of an engine for performing searches in a software artifacts repository using the Artificial Intelligence technique denominated Case-Based Reasoning (CBR). Such tool has the goal of providing an efficient mean allowing for artifacts reuse in the process of software development. The efficiency is in recovering artifacts that can be used for developing a software project. Keywords that define the problem characterize the project. The associated artifacts are also characterized by keywords and define the solution for the problem. Thus, a project and its associated artifacts constitute a case in the context of CBR. The proposal of the metadata for characterizing the project and the artifacts, as well as the organization of the structure of the search engine, is the goal of this work. That organization defined the use of the CBR technique for the implementation of a computational system characterized as a search engine. That engine has as its objective retrieving the most similar case to the defined search criteria. Attribute-value pairs represent these criteria. A weight associated to each pair indicates the importance of said keyword in the characterization of each artifact and as a search criterion.