In this repository, you can find the dataset used in the research titled 'A First Look at the Usability of UML Complexity Metrics'.
This research is executed by Julian Meurer, Sjoerd Plat en Joep Overmars and supervised by Tannaz Zameni. Contact information of the researchers and supervisor can be found at the end of this file.
In this research, the usability of different UML diagram complexity metrics is analyzed using well-researched and widely used software complexity metrics.
In order to conduct this analysis, the two open-source codebases called Java Websockets and Mobile Media are used.
The first step in the research was to create class and state diagrams for all versions of the selected codebases. Then the selected diagram complexity metrics were used to determine the complexity of each diagram. Besides these diagram metrics, the selected software metrics were used to determine the software complexity of the same codebases. The last step of generating the dataset was then to calculate the correlation between the different diagrams and software complexity metrics.
This dataset consists of the following (type of) files:
This dataset contains the created class diagram (as an XML file) for each version of the Java Websockets and Mobile Media codebase. The tool Visual Paradigm was used to create these diagrams.
To help determine the correctness of these diagrams, a custom-developed tool was used to convert a UML diagram into a structured text file. This file could then be used, in combination with a difference report between the current and previous software versions, to help determine the correctness of the diagrams. The log file itself can also be found in this dataset.
Click Here to go to the class diagrams.
This dataset also contains the created state diagram (as an XML file) for each version of the Mobile Media codebase. The tool Visual Paradigm was used to create these diagrams.
The same tool, as used for determining the correctness of the class diagrams, is used for the state diagrams.
Click Here to go to the state diagrams.
This dataset also contains the calculated diagram metrics (as a CSV file) for each version of the Java Websockets and Mobile Media codebase.
The metrics are calculated based on the formulas provided by the paper the metrics are presented in. More information about these metrics and references to their original paper can be found in our conference paper.
Click Here to go to the diagram metrics.
This dataset also contains the calculated software metrics (as a CSV file) for both Java Websockets and Mobile Media.
The metrics are calculated by well-researched and widely used tools like SonarQube and JArchitect.
Click Here to go to the software metrics.
This dataset also contains the calculated correlation between the diagram and software metrics (as a CSV file) for both Java Websockets and Mobile Media codebases.
The metrics are calculated using the python package Pandas and are based upon the Software and Diagram Metrics.
Click Here to go to the metric correlation.
Below you will find the contact information of the researchers and supervisor.
| Julian Meurer | Sjoerd Plat | Joep Overmars | Tannaz Zameni |
|---|---|---|---|
| [email protected] https://github.com/ItsJulian |
[email protected] https://github.com/sjoerdp |
https://github.com/Ilomiswir | [email protected] |