[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-585379-105":59,"doc-detail-585379-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","from-flat-to-structural-enhancing-automated-short-answer-grading-with-graphrag","From Flat to Structural - Enhancing Automated Short Answer Grading with GraphRAG","","Automated short answer grading (ASAG) is vital for scaling educational assessment, yet large language models (LLMs) often generate hallucinations and fail to follow strict rubrics. Retrieval-Augmented Generation (RAG) can lessen hallucinations by supplying relevant references, but flat vector retrieval overlooks structural links and multi-hop dependencies in complex instructional content. This work studies Graph Retrieval-Augmented Generation (GraphRAG) for ASAG, evaluating Microsoft GraphRAG and HippoRAG on an NGSS dataset. Structural retrieval outperforms standard RAG baselines, with HippoRAG improving Science and Engineering Practices.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/from-flat-to-structural-enhancing-automated-short-answer-grading-with-graphrag/585379/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/from-flat-to-structural-enhancing-automated-short-answer-grading-with-graphrag/585379.png","ImageObject",300,407,{"name":92,"@type":93},"Aria Callaghan","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-29","2026-10-15",true,{"@type":102,"interactionType":103,"userInteractionCount":52},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why do LLMs struggle with automated short answer grading?","Question",{"text":112,"@type":113},"They can hallucinate plausible but incorrect justifications and often do not strictly adhere to rubric constraints, instead relying on generalized knowledge from pretraining.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does standard RAG help, and what is its main limitation for ASAG?",{"text":117,"@type":113},"RAG reduces hallucinations by grounding generation in retrieved references, but flat vector retrieval treats knowledge as isolated chunks and may lose structural relations needed for evaluating reasoning.",{"name":119,"@type":110,"acceptedAnswer":120},"What does the GraphRAG approach contribute to short answer grading?",{"text":121,"@type":113},"GraphRAG incorporates structural information by constructing and traversing knowledge graphs, enabling evidence retrieval that preserves relational structure and supports multi-hop reasoning chains.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},585379,1792101580,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":52,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":34,"language":139,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":67,"update_tm":143,"read_time":144,"preview_page_count":4},962084926284,"https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","From Flat to Structural: Enhancing Automated Short Answer Grading with GraphRAG  \nYucheng Chu Michigan State University [chuyuch2@msu.edu](chuyuch2@msu.edu)  \nHaoyu Han Michigan State University [hanhaoy1@msu.edu](hanhaoy1@msu.edu)  \nShen Dong Michigan State University [dongshe1@msu.edu](dongshe1@msu.edu)  \nHang Li  \nMichigan State University  \n[lihang4@msu.edu](lihang4@msu.edu)  \nKaiqi Yang  \nMichigan State University  \n[kqyang@msu.edu](kqyang@msu.edu)  \nYasemin Copur-Gencturk  \nUniversity of Southern California  \n[copurgen@usc.edu](copurgen@usc.edu)  \nABSTRACT  \nAutomated short answer grading (ASAG) is critical for scaling educational assessment, yet large language models (LLMs) often struggle with hallucinations and strict rubric adherence due to a mismatch between generalized pre-training objectives and the task-specific requirements of grading. While Retrieval-Augmented Generation (RAG) can reduce hallucinations by providing access to relevant reference materials, standard flat vector retrieval treats knowledge as isolated fragments, failing to capture the structural relationships and multi-hop dependencies inherent in complex educational content. To explore whether incorporating structural information is beneficial for ASAG, we investigate Graph Retrieval-Augmented Generation (GraphRAG) as a general paradigm and systematically evaluate two representative instantiations, Microsoft GraphRAG and HippoRAG, on this task. These approaches differ in how they construct and traverse knowledge graphs, providing complementary perspectives on graph-based evidence retrieval. Experimental evaluations on a Next Generation Science Standards (NGSS) dataset demonstrate that structural retrieval outperforms standard RAG baselines across all metrics. Notably, the HippoRAG achieved substantial improvements in evaluating Science and Engineering Practices (SEP), confirming the superiority of structural retrieval in verifying the logical reasoning chains required for higher-order academic assessment.  \nKeywords  \nLarge Language Model, Automated Grading,  \nRetrieval-Augmented Generation (RAG), GraphRAG  \n1. INTRODUCTION  \nAutomated short answer grading (ASAG) has emerged as a critical technology for scaling educational assessment, aiming to evaluate student responses with the same accuracy  \nYucheng Chu, Haoyu Han, Shen Dong, Hang Li, Kaiqi Yang, Yasemin Copur-Gencturk, Joseph Krajcik, Namsoo Shin, and Hui Liu. From Flat to Structural: Enhancing Automated Short Answer Grading with GraphRAG. In Anthony Botelho, Maria Mercedes T. Rodrigo, Adish Singla, Hiroaki Ogata, Hyojeong So, and Young Hoan Cho (eds.) Proceedings of the 19th International Conference on Educational Data Mining, Seoul, Republic of Korea, June, 2026, pp. 462–468. International Educational Data Mining Society (2026) .  \n© 2026 Copyright is held by the author(s) . This work is distributed under the Creative Commons Attribution NonCommercial NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license. [https://doi.org/10.5281/zenodo.21039760](https://doi.org/10.5281/zenodo.21039760)  \nand consistency as human instructors. While early approaches relied on keyword matching and latent semantic analysis, the advent of large language models (LLMs) has fundamentally transformed the landscape. LLMs possess remarkable semantic understanding and generation capabilities. These features allow them to not only interpret complex student answers but also, crucially, to understand and adhere to nuanced grading rubrics with human-level flexibility. However, despite their fluency, off-the-shelf LLMs frequently struggle with the strict constraints of academic grading. They are prone to “hallucinations” where plausible but incorrect justifications are generated. Furthermore, they often fail toadhere strictly to specific grading rubrics, relying instead on their generalized internal knowledge acquired during pretraining. This vast but generic training data often conflicts with specific course definitions","cbCaiiGJ769DE2pn","https://ap.wps.com/l/cbCaiiGJ769DE2pn","pdf",459563,"English","# Abstract\n# Introduction\n## Automated Short Answer Grading and LLM Challenges\n## Retrieval-Augmented Generation and Flat Retrieval Limits\n# Related Work\n# Method: GraphRAG Paradigm\n## Microsoft GraphRAG\n## HippoRAG\n# Experimental Setup\n## NGSS Dataset\n# Results and Analysis\n## Overall Metrics\n## Improvements for Science and Engineering Practices","[{\"question\":\"Why do LLMs struggle with automated short answer grading?\",\"answer\":\"They can hallucinate plausible but incorrect justifications and often do not strictly adhere to rubric constraints, instead relying on generalized knowledge from pretraining.\"},{\"question\":\"How does standard RAG help, and what is its main limitation for ASAG?\",\"answer\":\"RAG reduces hallucinations by grounding generation in retrieved references, but flat vector retrieval treats knowledge as isolated chunks and may lose structural relations needed for evaluating reasoning.\"},{\"question\":\"What does the GraphRAG approach contribute to short answer grading?\",\"answer\":\"GraphRAG incorporates structural information by constructing and traversing knowledge graphs, enabling evidence retrieval that preserves relational structure and supports multi-hop reasoning chains.\"}]","From Flat to Structural - Enhancing Automated Short Answer Grading with GraphRAG | PDF",1792054560,18]