[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-124574-en":3,"doc-seo-124574-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},124574,1099513958607,"Jiven","https://ap-avatar.wpscdn.com/avatar/100002390cf8733938c?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778829742770036399",8,"Research & Report","DAFS - A Domain Aware Few Shots Generative Model for Event Detection - Abstract","Event detection (ED) depends heavily on labeled trigger-word training data, which is scarce and uneven in domain-specific areas such as finance. Existing approaches that generate labels with generative models often require rich domain knowledge that cannot be inferred from only a few shots. DAFS (Domain Aware Few Shots) generates domain-based training data from a small labeled set by computing word transition probabilities with self-supervised sentence information across categories and retaining key triggers, then applying a joint method to balance diversity and effectiveness.","Springer Nature 2021 LATEX template  \nDAFS: A Domain Aware Few Shots Generative Model for Event Detection  \nNan Xia 1 , Hang Yu 1*, Yin Wang 1 , Junyu Xuan2 and Xiangfeng Luo 1  \n1 School of Computer Engineering and Science, Shanghai University, Shang Da Street No.99, Shang Hai, 200444, China.  \n2 Australian arti􀀌cial intelligence institute, University of Technology Sydney, 15 Broadway Ultimo, Sydney State, 2007,  \nAustralia.  \n*Corresponding author(s). E-mail(s): [hang.yu@shu.edu.cn](hang.yu@shu.edu.cn) ;  \nContributing authors: [shkklt@shu.edu.cn](shkklt@shu.edu.cn) ; [wangyin2018@shu.edu.cn](wangyin2018@shu.edu.cn) ; [junyu.xuan@uts.edu.au](junyu.xuan@uts.edu.au) ;  \n[luoxf@shu.edu.cn](luoxf@shu.edu.cn) ;  \nAbstract  \nMore and more large-scale pre-trained models show apparent advantagesin solving the event detection (ED), i.e., a task to solve the problem of event classi􀀌cation by identifying trigger words. However, this kind of model heavily depends on labeled training data. Unfortunately, thereis not enough labeled training data for some particular areas, such as 􀀌nance, due to the high cost of the data annotation process. Besides, the manually labeled training data has many problems like uneven sampling distribution, poor diversity, and massive long-tail data. Recently, some researchers have used the generative model to label data. However, training the generative models needs rich domain knowledge, which cannot be obtained from a few shots. Therefore, we propose a Domain Aware Few Shots (DAFS) generative model that can generate domain based training data through a relatively small amount of labeled data. First, DAFS utilizes self-supervised information from various categories of sentences to calculate words' transition probability under di􀀋erent domain and retain key triggers in each sentence. Then, we apply our joint algorithm to generate labeled training data that considers both diversity and e􀀋ectiveness. Experimental results demonstrate that the training data generated  \nSpringer Nature 2021 LATEX template  \n2 DAFS: A Domain Aware Few Shots Generative Model for Event Detection  \nby DAFS signi􀀌cantly improves the performance of ED in actual 􀀌nancial data. Especially when there are no more than 20 training data, DAFS can still ensure the generative quality to a certain extent. It also obtains new state-of-the-art results on ACE2005 multilingual corpora.  \nKeywords: event detection, domain-aware, joint algorithm, self-supervised  \n1 Introduction  \nAutomatic event extraction is a fundamental task of information extraction. Generally speaking, event detection (ED) aims at identifying event triggers which is a key step of event extraction. For example, from the sentence\"It0 s been ten minutes since I got home; and George called\", systems should detect the event of \"Movement : Transport\" triggered by \"got home\", and the event of \"Contact : Phone Write\" triggered by \"called\" .  \nMost of the ED methods before the year of 2018 applied a word-wise classi􀀌cation paradigm, which has achieved signi􀀌cant progress [1] . Afterwards, with the rise of new pre-trained model BERT [2], the method of representation learning can obtain semantic information in sentence more precisely, as it is known that word-wise ED models su􀀋er from the trigger word ambiguity and semantic loss problems [1] . For instance, we can't directly detect the event of \"bankrupt\" in sentence \"Will the bankruptcy caused by the financial crisis affect Ali?\". Although it has the trigger word \"bankruptcy\", it does not mean anything happened in a real 􀀌nancial situation . The pre-trained model can learn the language of this interrogative state through 􀀌ne-tune mechanism, but it needs more data of this type.  \nFuthermore, we summarize the similarities and di􀀋erences between training data and test data in real data in Table 1 . The 􀀌rst line in Table 1 can easily recognize because of similar trigger words in both training and test corpus. Additionally, the second line in ","cbCaiuzi5rjleSUG","https://ap.wps.com/l/cbCaiuzi5rjleSUG","pdf",1022622,1,18,"English","en",105,"# Abstract\n## Motivation and problem setting\n## Proposed approach (DAFS)\n## Experimental results\n# Introduction\n## Definition of event detection and triggers\n## Limitations of earlier methods\n## Data differences between training and test corpora","[{\"question\":\"Why is labeled data a bottleneck for event detection in specific domains like finance?\",\"answer\":\"Event detection relies on labeled trigger-word data for event classification. In finance, annotation is costly, leading to insufficient labeled samples and issues such as uneven sampling distribution and long-tail coverage.\"},{\"question\":\"What is the core idea behind the DAFS generative model?\",\"answer\":\"DAFS uses self-supervised information from sentence categories to estimate word transition probabilities under different domains and to keep key triggers. It then generates labeled training data using a joint algorithm that considers diversity and effectiveness.\"},{\"question\":\"How does DAFS perform when only a small number of training samples (e.g., no more than 20) are available?\",\"answer\":\"The results indicate that DAFS can still maintain generative quality to a certain extent with very limited training data. It also reports improved performance on ACE2005 multilingual corpora.\"}]","DAFS - A Domain Aware Few Shots Generative Model for Event Detection - Abstract | PDF",1785893051,45,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"dafs-a-domain-aware-few-shots-generative-model-for-event-detection-abstract","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/dafs-a-domain-aware-few-shots-generative-model-for-event-detection-abstract/124574/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is labeled data a bottleneck for event detection in specific domains like finance?","Question",{"text":75,"@type":76},"Event detection relies on labeled trigger-word data for event classification. In finance, annotation is costly, leading to insufficient labeled samples and issues such as uneven sampling distribution and long-tail coverage.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the core idea behind the DAFS generative model?",{"text":80,"@type":76},"DAFS uses self-supervised information from sentence categories to estimate word transition probabilities under different domains and to keep key triggers. It then generates labeled training data using a joint algorithm that considers diversity and effectiveness.",{"name":82,"@type":73,"acceptedAnswer":83},"How does DAFS perform when only a small number of training samples (e.g., no more than 20) are available?",{"text":84,"@type":76},"The results indicate that DAFS can still maintain generative quality to a certain extent with very limited training data. It also reports improved performance on ACE2005 multilingual corpora.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]