典型文献
Learning to Combine Answer Boundary Detection and Answer Re-ranking for Phrase-Indexed Question Answering
文献摘要：
Phrase-indexed question answering(PIQA)seeks to improve the inference speed of question answering(QA)models by enforcing complete independ-ence of the document encoder from the question encoder,and it shows that the constrained model can achieve sig-nificant efficiency at the cost of its accuracy.In this pa-per,we aim to build a model under the PIQA constraint while reducing its accuracy gap with the unconstrained QA models.We propose a novel framework—AnsDR,which consists of an answer boundary detector(AnsD)and an answer candidate ranker(AnsR).More specific-ally,AnsD is a QA model under the PIQA architecture and it is designed to identify the rough answer boundar-ies;and AnsR is a lightweight ranking model to finely re-rank the potential candidates without losing the effi-ciency.We perform the extensive experiments on public datasets.The experimental results show that the pro-posed method achieves the state of the art on the PIQA task.
文献关键词：
中图分类号：
作者姓名：
WEN Liang;SHI Haibo;ZHANG Xiaodong;SUN Xin;WEI Xiaochi;WANG Junfeng;CHENG Zhicong;YIN Dawei;WANG Xiaolin;LUO Yingwei;WANG Houfeng
作者机构：
Department of Computer Science and Technology,Peking University,Beijing 100871,China;Peng Cheng Laboratory,Shenzhen 518052,China;Baidu Inc.,Beijing 100193,China
文献出处：
引用格式：
[1]WEN Liang;SHI Haibo;ZHANG Xiaodong;SUN Xin;WEI Xiaochi;WANG Junfeng;CHENG Zhicong;YIN Dawei;WANG Xiaolin;LUO Yingwei;WANG Houfeng-.Learning to Combine Answer Boundary Detection and Answer Re-ranking for Phrase-Indexed Question Answering)[J].电子学报（英文）,2022(05):938-948
A类：
Phrase,PIQA,AnsDR,AnsD,ranker,AnsR,boundar
B类：
Learning,Combine,Boundary,Detection,Re,ranking,Indexed,Question,Answering,indexed,question,answering,seeks,improve,inference,speed,models,by,enforcing,complete,independ,document,encoder,from,shows,that,nificant,efficiency,cost,its,accuracy,this,pa,aim,build,under,constraint,while,reducing,gap,unconstrained,We,propose,novel,framework,which,consists,boundary,detector,More,specific,ally,architecture,designed,identify,rough,ies,lightweight,finely,potential,candidates,without,losing,perform,extensive,experiments,public,datasets,experimental,results,posed,method,achieves,state,art,task
AB值：
0.48053
相似文献
机标中图分类号，由域田数据科技根据网络公开资料自动分析生成，仅供学习研究参考。
