典型文献
Controllable data synthesis method for grammatical error correction
文献摘要：
Due to the lack of parallel data in current gramma-tical error correction(GEC)task,models based on sequence to sequence framework cannot be adequately trained to obtain higher performance.We propose two data synthesis methods which can control the error rate and the ratio of error types on synthetic data.The first approach is to corrupt each word in the monolingual corpus with a fixed probability,including replace-ment,insertion and deletion.Another approach is to train error generation models and further filtering the decoding results of the models.The experiments on different synthetic data show that the error rate is 40％and that the ratio of error types is the same can improve the model performance better.Finally,we synthesize about 100 million data and achieve comparable performance as the state of the art,which uses twice as much data as we use.
文献关键词：
中图分类号：
作者姓名：
Liner YANG;Chengcheng WANG;Yun CHEN;Yongping DU;Erhong YANG
作者机构：
School of Information Science,Beijing Language and Culture University,Beijing 100083,China;Beijing Advanced Innovation Center for Language Resources,Beijing Language and Culture University,Beijing 100083,China;Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China;School of Information Management&Engineering,Shanghai University of Finance and Economics,Shanghai 200433,China
文献出处：
引用格式：
[1]Liner YANG;Chengcheng WANG;Yun CHEN;Yongping DU;Erhong YANG-.Controllable data synthesis method for grammatical error correction)[J].计算机科学前沿,2022(04):66-75
A类：
grammatical,gramma,monolingual
B类：
Controllable,data,synthesis,error,correction,Due,lack,parallel,current,GEC,task,models,sequence,framework,cannot,adequately,trained,obtain,higher,performance,We,propose,two,methods,which,control,rate,types,synthetic,first,approach,corrupt,each,word,corpus,fixed,probability,including,replace,insertion,deletion,Another,generation,further,filtering,decoding,results,experiments,different,show,that,same,improve,better,Finally,we,synthesize,about,million,achieve,comparable,state,art,uses,twice,much
AB值：
0.522686
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。
