Details
- Supervisors
- Faculty
- Degree label
- Abstract
- Mining patterns in source code is a great subject in software engineering. Software engineers has a great interest in discovering regularities in source code, both in academia and in industry. Indeed, regularities can provide useful information to help the developer in a variety of tasks such as code comprehension, code refactoring, bad practices detecting, etc. However, traditional pattern mining algorithms are not always well suited for this purpose and discover few rich and interesting patterns. This thesis is based on a pattern mining framework that tries to overcome these limitations. This framework is parametric and allows the discovery of patterns in source code, independently of its programming language. The mining is based on a frequent tree exploration algorithm. The existing research on this framework considers source code written in Java and COBOL. Through this work, we will analyze how this technique can be applied to C# source code. Indeed, the functioning of this framework, as well as the configurations to be setup for C# source code exploration, will be explained. Plus, rich and interesting patterns discovered as a result of this exploration will be exposed during this thesis.