Automated Writing Evaluation has shifted from score-centered assessment toward revision-oriented feedback. However, current Large Language Models (LLMs) often generate generic comments that overlook essays’ explicit structure and discourse-level flaws. We propose Graph-Reasoned Automated Feedback and Evaluation (GRAFE), which uses genre-specific structural graphs to guide score-feedback generation. GRAFE encodes discourse-level priors, including argumentation logic, evidence integration, and narrative progression, as continuous representations injected into the LLM. Furthermore, we introduce a training pipeline that combines two-stage Supervised Fine-Tuning for score-calibrated initialization with Group Relative Policy Optimization to align the model toward structural diagnostic accuracy, scoring consistency, and format compliance. We construct a graph-augmented ASAP++ feedback dataset with an expert-calibrated evaluation set. Experiments show that GRAFE outperforms text-only baselines on Structure-Aware Diagnosis and Structure-Grounded Actionability, demonstrating the value of explicit structural modeling.

The assessment of children’s narrative ability is crucial for diagnosing language disorders and planning interventions. Distinct from the typical automated essay scoring, this task focuses primarily on evaluating the completeness of narrative content and the coherence of expression, as well as the interpretability of assessment results. To address these issues, we propose a novel computational assessing framework NarGINA, under which the narrative graph is introduced to provide a concise and structured summary representation of narrative text, allowing for explicit narrative measurement. To this end, we construct the first Chinese children’s narrative assessment corpus based on real children’s narrative samples, and we then design a narrative graph construction model and a narrative graph-assisted scoring model to yield accurate narrative ability assessment. Particularly, to enable the scoring model to understand narrative graphs, we propose a multi-view graph contrastive learning strategy to pre-train the graph encoder and apply instruction-tuned large language models to generate scores. The extensive experimental results show that NarGINA can achieve significant performance improvement over the baselines, simultaneously possessing good interpretability. Our findings reveal that the utilization of structured narrative graphs beyond flat text is well suited for narrative ability assessment. The model and data are publicly available at https://github.com/JlexZzz/NarGINA.