Recuperação de visão estrutural de software combinando análise de semântica estática e latente

Abstract

Software Comprehension is still a challenge for the maintenance of unknown systems. This situation occurs because most of the information is not explicitly available. This work aims to support Software Comprehension through the use of a visual interactive tool, creating a structural vision of a target system. For this purpose, the technique of information retrieval Latent Semantic Indexing - LSI combinated with the static structure of the source code is uses. In this work, the informal hidden knowledge contained in the vocabulary of artifacts is combined with the static structure of source code for the construction of a structural vision of the unknown system. The method and its respective tool are validated with the analysis of the Prevayler system and the OpenJDK compiler. Specifically, the proposed method has shown that a structural vision of unknown system can be found through a simple way, providing a valuable aid to understanding such systems.