文档中包含网盘的地址,数据共319MNLP方向文本摘要,文本分类,等方向可采取!TheLCSTSdatasetincludestwoparts:/DATA:1.PARTI:isthemaincontentsofLCSTSthatcontains2,400,591(shorttext,summary)pairs.Itcanbeusedtotrainsupervisedlearningmodelsforsummarygeneration.2.PARTII:contains10,666humanlabled(shorttext,summary)pairswhichcanbeusedtotrainclassifiertofilterthenoisesofthePARTI.3.PARTIII:contains1,106(shorttext,summary)pairs,thispartislabledby3personswiththesamelabels.Thesepairswithscore3,4and5canbeusedastestsetforevaluatingsummarygenerationsystems./Result:1.sumary.generated.char.context.txt:containsthesummarygeneratedbyusingRNN+contextonthecharacterbasedinput.2.sumary.generated.char.nocontext.txt:containsthesummarygeneratedbyusingRNN+nocontextonthecharacterbasedinput.3.sumary.generated.word.context.txt:containsthesummarygeneratedbyusingRNN+contextonthewordbasedinput.4.sumary.generated.word.nocontext.txt:containsthesummarygeneratedbyusingRNN+nocontextonthewordbasedinput.5.weibo.txt:containstheweiboofthetestset.6.sumary.human:containsthesumariescorrespondingto'weibo.txt'writtenbyhuman.Thispartisthetestsetofthepaper.7.rouge.char_context.txt:therougemetriconsumary.generated.char.context8.rouge.char_nocontext.txt:therougemetriconsumary.generated.char.nocontext9.rouge.word_context.txt:therougemetriconsumary.generated.word.context10.rouge.word_nocontext.txt:therougemetriconsumary.generated.word.nocontext
2018/10/23 6:40:09
66B
nlp
1