Information Retrieval Techniques for Software Design Visualization

Large software systems typically comprise many components written in many different languages that communicate through complex interfaces. The code base will consist of a mixture of source code, configuration files, test vectors and other artifacts. Tools are needed to help software engineers understand the architecture of the software and the role of all of the components in the code base. We propose the development of a tool that will support a wide variety of software engineering activities, and which will eventually scale to tens of millions of lines of code. Our approach is to provide two different features: an intelligent search engine for the code base that uses latent semantic indexing, and a hierarchical visualization tool that will present clustered views of items in an intelligent and interactive manner. These seemingly disparate features will both use the same underlying mechanism: information retrieval techniques that compute statistical measures of similarity between documents. We will leverage our experience and capabilities in static analysis to adapt these techniques so that useful facts are extracted from the code base.