Revise related work

Author: alextopalova

paper/content/related_work.typ

@@ -35,5 +35,5 @@ Targeting a different area of software engineering education, Sölch and Krusche
 
 In contrast, Ramesh et al. explored the use of LLMs for automated code review in an industrial setting, combining static program analysis with prompt-based evaluation of code changes @ramesh2025ericsson. Although the authors don't focus on education, their approach generates concise and context-aware review comments for modified code segments, demonstrating how LLMs could generally support development teams by identifying potential issues and improving code quality. However, the system primarily targets code-level correctness and maintainability and does not consider collaboration aspects like contribution balance or team dynamics.
 
-Overall, prior work has explored a range of approaches to assessing student contributions in team-based software engineering projects, including repository mining techniques, code quality analysis and more recently, LLM-based summarization and feedback systems. While these approaches provide valuable insights into individual participation, code quality and project progress, they largely rely on either quantitative activity metrics or descriptive analyses. As a result, they fail to capture important aspects of collaboration, such as the semantic quality of contributions, the balance of effort among team members and the extent of shared ownership of code artifacts. Furthermore, existing approaches either focus on instructor-facing assessment or student-facing feedback, but rarely integrate both perspectives into a unified framework. These limitations highlight the need for a more comprehensive approach to assessing collaboration quality, which Harmonia bridges.
+Overall, prior work has explored a range of approaches to assessing student contributions in team-based software engineering projects, including repository mining techniques, code quality analysis and more recently, LLM-based summarization and feedback systems. While these approaches provide valuable insights into individual participation, code quality and project progress, they largely rely on either quantitative activity metrics or descriptive analyses. As a result, they fail to capture important aspects of collaboration, such as the semantic quality of contributions, the balance of effort among team members and the extent of shared ownership of code artifacts. These limitations highlight the need for a more comprehensive approach to assessing collaboration quality, which Harmonia aims to bridge.