In recent years, the rapid rise of artificial intelligence technologies, particularly large language models(LLMs), has facilitated essay grading. However, current automated essay scoring tasks focus primarily on analyzing textual characteristics such as coherence, grammatical correctness, and clarity of expression to generate scores, while overlooking the critical relevance to essay writing requirements. In this paper, we present an overview of the Evaluation of Essay On-Topic Graded Comments(EOTGC) task in the NLPCC 2025 shared task, which focuses on assessing how well student writings align with writing requirements, maintain coherent central ideas, and use appropriate content. A total of 19 teams participated in this task, among which 9 teams submitted 127 submissions. This shared task sets up two tracks: (1)Relevance Scoring of Essays(RSE); (2)Generation of Relevance Comments(GRC). We detail the task definitions, dataset characteristics, evaluation metrics, and summarize the approaches adopted by participants. This work provides insights and assistance for future research in the field of essay on-topic graded comments.