This paper investigates the impact of experience-aware oversampling on automated code review models. The study aims to improve the quality of code reviews generated by large language models by focusing on reviews from experienced reviewers. The authors propose an experience-aware oversampling technique to enhance the model's ability to generate more accurate and meaningful code review comments. They evaluate their approach using a large-scale code review dataset and compare the performance of their models with the original model. The results show that experience-aware oversampling significantly improves the correctness, information level, and meaningfulness of generated code review comments. The study highlights the potential of utilizing high-quality reviews from experienced reviewers to enhance automated code review systems without requiring additional data. The findings suggest that current training strategies are not fully utilizing the high-quality reviews available in the training data, and that experience-aware oversampling can lead to better code review performance. The study also identifies that experienced reviewers are more likely to provide insights into critical issues such as logic, validation, and resource management. The results demonstrate that experience-aware oversampling can generate more accurate and useful code review comments, which can help improve the overall quality of code changes. The study concludes that automated code review models can be enhanced by focusing on experienced reviewers' feedback, leading to more effective and efficient code review processes.This paper investigates the impact of experience-aware oversampling on automated code review models. The study aims to improve the quality of code reviews generated by large language models by focusing on reviews from experienced reviewers. The authors propose an experience-aware oversampling technique to enhance the model's ability to generate more accurate and meaningful code review comments. They evaluate their approach using a large-scale code review dataset and compare the performance of their models with the original model. The results show that experience-aware oversampling significantly improves the correctness, information level, and meaningfulness of generated code review comments. The study highlights the potential of utilizing high-quality reviews from experienced reviewers to enhance automated code review systems without requiring additional data. The findings suggest that current training strategies are not fully utilizing the high-quality reviews available in the training data, and that experience-aware oversampling can lead to better code review performance. The study also identifies that experienced reviewers are more likely to provide insights into critical issues such as logic, validation, and resource management. The results demonstrate that experience-aware oversampling can generate more accurate and useful code review comments, which can help improve the overall quality of code changes. The study concludes that automated code review models can be enhanced by focusing on experienced reviewers' feedback, leading to more effective and efficient code review processes.