During software development and maintenance, human programmers read and modify source code that they and others produce. This code forms a means of communication between programmers, leading to code artifacts that are readable as well as runnable. While programming language syntax and semantics convey the algorithm to be executed, the identifier names and comments express the higher-level conceptual algorithmic steps and domain concepts behind the implementation. Our research focuses on capturing this conceptual knowledge of the programmer as expressed in both linguistic information and programming language structure and semantics, and leveraging this information in human-oriented software engineering tools. To that end, we have developed a novel Software Word Usage Model (SWUM) that captures linguistic word relationships and integrates them with program structure information.
