thesis-se.md

@@ -5,23 +5,23 @@
 
 Here are the [slides](uploads/c973853f8ecebd1b26f25fbf0c6f9b7c/2025-04-08-thesis-topics-presentation.pdf) for the presentation from April 8, 2025, and some details about the topics:
 
-## Compare our approach with others
+### Compare our approach with others
 
 There are approaches that e.g., recommend how to split up a monolith into microservices, see, e.g., https://servicecutter.github.io/. A research question is how our approach compares to this one.
 
-## Integrate Byte Code Size as metric
+### Integrate Byte Code Size as metric
 
 For a modularization (say, Java projects into several `.jar` files), add the file size of the resultung `.jar` files into account. One useful metric here is the sum of the size of all jar files (where dependencies must be taken into account), which then must be weighed against constraints such as "`main.MainA` must be in a different file than `main.mainB`" that fix some separtion by business logic. (Of course, the constraints must not specify too much, otherwise the optimization space gets trivial).
 
-## Experiment with different mutations / recombinations
+### Experiment with different mutations / recombinations
 
 Add heuristic mutations and recombinations ones and compare them against the current set of operations; both on real and on synthetic examples.
 
-## Semantic Coupling
+### Semantic Coupling
 
 Building on the results of a previous bachelor's thesis, integrate semantic coupling as a metric into the optimization. Evaluate the metric manually on a known project.
 
-## Improve parallelization
+### Improve parallelization
 
 The current implementation of the optimization framework uses parallelization, but the support could be improved.