[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-148014-105":59,"doc-detail-148014-en":123},{"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":116,"head_meta":118,"extra_data":120,"updated_unix":122},105,"en","embre-entity-aware-masking-for-biomedical-relation-extraction","EMBRE - Entity-aware Masking for Biomedical Relation Extraction","","Information extraction methods such as named entity recognition and relation extraction enable researchers to connect relevant facts across large volumes of unstructured text. This paper presents EMBRE, an Entity-aware Masking approach for biomedical relation extraction in BioRED Task 1, where gold entities are given. The method integrates entity knowledge by pretraining a backbone with an entity-masking objective, then uses multilayer perceptrons to predict relation and novelty. Experiments show improved extraction performance over a baseline.",{"@graph":69,"@context":115},[70,84,106],{"@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/embre-entity-aware-masking-for-biomedical-relation-extraction/148014/",{"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/embre-entity-aware-masking-for-biomedical-relation-extraction/148014.png","ImageObject",300,407,{"name":92,"@type":93},"Đào","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-27","2026-08-26",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",16,{"@type":107,"mainEntity":108},"FAQPage",[109],{"name":110,"@type":111,"acceptedAnswer":112},"What tasks does EMBRE predict during training and inference?","Question",{"text":113,"@type":114},"The model predicts relation classification logits and a novelty identification signal, trained together in a multi-task learning setup.","Answer","https://schema.org",{"og:url":83,"og:type":117,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":119,"canonical":83},"index,follow",{"doc_id":121,"site_id":62},148014,1787773577,{"code":4,"msg":5,"data":124},{"doc_id":121,"user_id":125,"nickname":92,"user_avatar":126,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":127,"file_id":128,"file_url":129,"file_type":130,"file_size":131,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":24,"language":132,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":133,"faqs":134,"seo_title":135,"seo_description":67,"update_tm":122,"read_time":136,"preview_page_count":4},1374402968488,"https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","EMBRE: Entity-aware Masking for Biomedical Relation Extraction  \nMingjie Li 1 and Karin Verspoor1, *  \n1 School of Computing Technologies, RMIT University, Melbourne, AU  \n*[Corresponding author:](Corresponding author: karin.verspoor@rmit.edu.au)[ ](Corresponding author: karin.verspoor@rmit.edu.au)[karin.verspoor@rmit.edu.au](Corresponding author: karin.verspoor@rmit.edu.au)  \nAbstract  \nInformation extraction techniques, including named entity recognition (NER) and relation extraction (RE), are crucial in many domains to support making sense of vast amounts of unstructured text data by identifying and connecting relevant information. Such techniques can assist researchers in extracting valuable insights. In this paper, we introduce the Entity-aware Masking for Biomedical Relation Extraction (EMBRE) method for biomedical relation extraction, as applied in the context of the BioRED challenge Task 1, in which human-annotated entities are provided as input. Specifically, we integrate entity knowledge into a deep neural network by pretraining the backbone model with an entity masking objective. We randomly mask named entities for each instance and let the model identify the masked entity along with its type. In this way, the model is capable of learning more specific knowledge and more robust representations. Then, we utilize the pre-trained model as our backbone to encode language representations and feed these representations into two multilayer perceptron (MLPs) to predict the logits for relation and novelty, respectively. The experimental results demonstrate that our proposed method can improve the performances of entity pair, relation and novelty extraction over our baseline.  \nIntroduction  \nThe BioRED challenge task (3) presented a rich biomedical relation extraction dataset, including 600 PubMed abstracts with gold standard annotations of 8 relation types connecting key entity types including drugs, diseases, genes, and chemicals. Additionally, the BioRED dataset includes an assignment of “Novel” or “No[t novel]” to each relation, indicating whether the relation comes from main content or background, similar to the focus/background entities of Jimeno Yepes and Verspoor (1) but extended to relations. BioRED provides a comprehensive testbed for researchers to build and evaluate their biomedical IE systems. Different from many existing RE datasets which focus on extracting relations from a single sentence, BioRED contains relations across multiple sentences, moving the research from sentence-level to document-level.  \nAlong with the dataset, Luo et al. (3) also performed benchmarking experiments with several stateof-the-art methods for RE based on deep learning techniques. Specifically, they utilized BERTGT (6) and PubMedBERT (5) as the backbone models to encode language representations. These encodings were then fed into classifiers to predict specific relation types. The experimental results demonstrate that using PubMedBERT – which is pre-trained on literature from PubMed – is more effective than the BERT (2) language model for biomedical RE. The PubMedBERT token vocabulary recognizes more words from the BioRED corpus than general BERT. In addition, it encodes more biomedical knowledge and generates robust and generalizable token embeddings for this task. Building on this dataset, several deep learning-based RE systems have been proposed. For example, BioREx (4) provides a data-centric approach that improves the benchmark  \nperformances by leveraging a rich dataset combining five RE datasets with different kinds of annotations. This approach injects diverse, synergistic, biomedical knowledge into the model.  \nIn this paper, we explore a method we dub EMBRE (Entity-aware Masking for Biomedical Relation Extraction) to incorporate knowledge into the benchmark model without the need for a large, heterogeneous corpus. PubMedBERT is trained on literature from PubMed with the masked language modelling (LM) objective. In a RE task","cbCainlaNRnLd5dz","https://ap.wps.com/l/cbCainlaNRnLd5dz","pdf",415075,"English","# Abstract\n# Introduction\n# Methods\n## Named Entity Masking Pre-training Strategy EMBRE","[{\"question\":\"What tasks does EMBRE predict during training and inference?\",\"answer\":\"The model predicts relation classification logits and a novelty identification signal, trained together in a multi-task learning setup.\"}]","EMBRE - Entity-aware Masking for Biomedical Relation Extraction | PDF",13]