Abstract
Modern software development tools typically provide a syntax highlighting feature to enhance the readability of source code. Syntax highlighting means that different kinds of terms in the text of source code are rendered in visually distinct ways. A syntax theme is a collection of definitions that determine the look of a workspace. Editors and IDEs often let the user customize the syntax theme to their preference and can also provide services for users to share themes within the community. This paper presents a data mining study of a large collection of publicly available syntax themes (approximately 330) associated with a popular open-source text editor. For each theme we collected certain metadata as well as a cascading stylesheet definition. The metadata includes information such as the title of theme, a plain text description, and specific programming languages supported by the theme. For each syntax theme, a preprocessor script creates a feature vector that aggregates data from the theme definition and metadata. Themes are analyzed with both RGB and HSV color models. Analysis shows a variety of significant correlations in the dataset as well general patterns of syntax theme usage. We also discuss how our dataset bears on previously established models of code comprehension.