In Pharo, as in many other programming languages, method/class comments represent a useful source of information to understand the responsibilities of a given class.
However, these comments tend to be often incomplete or inconsistent with the source code. Due to software evolution, class descriptions get out of sync with the source code.
The aim of the project is to analyze code and comments of methods/classes in Pharo to generate class descriptions in natural language and study the inconsistency between source code and its comments.
Comments are written in natural language. In recent research, the techniques to analyze the comments, are based on noun phrase analysis, dependency analysis, and semantic analysis. All of them could be used to extract significant information out of comments.