?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=KINERJA+METODE+CONVOLUTIONAL+NEURAL+NETWORK+(CNN)+DAN%0D%0ALONG-SHORT+TERM+MEMORY+(LSTM)+PADA+KLASIFIKASI+DATA+%0D%0AJUDUL+BERITA+COVID-19%0D%0A+&rft.creator=+LUTHFIA+NUR+AZIZAH%2C++1817031065&rft.subject=000+Ilmu+komputer%2C+informasi+dan+pekerjaan+umum&rft.subject=001+Ilmu+pengetahuan&rft.description=%0D%0AThe+performance+of+deep+learning+methods+on+the+classification+of+text+data+with%0D%0Adifferent+imbalance+ratios+is+an+important+discussion+because+the+existing+data+is%0D%0Ainherently+imbalanced.+This+study+is+looking+for+a+reliable+deep+learning+method+to%0D%0Aclassify+data+on+Indonesian+news+headlines+about+COVID-19+with+several+data%0D%0Aimbalance+ratios.+Several+data+imbalance+ratios+were+made+by+taking+samples+from%0D%0Anews+events+using+simple+random+sampling+of+30%25%2C+20%25%2C+10%25%2C+and+1%25.+The%0D%0Aperformance+of+the+CNN+and+LSTM+methods+was+tested+using+10-fold+cross%0D%0Avalidation+and+compared+based+on+accuracy%2C+precision%2C+recall%2C+and+f1-score.+The%0D%0ACNN+model+architecture+built+in+this+study+generally+consists+of+an+input+layer%2C+word%0D%0Aembedding+layer%2C+two+convolutional+layers%2C+one+pooling+layer%2C+flatten%2C+two+hidden%0D%0Alayers%2C+an+output+layer.+A+batch+normalization+layer+and+a+dropout+layer+after+each%0D%0Alayer.+The+LSTM+model+architecture+built+in+this+study+generally+consists+of+an%0D%0Ainput+layer%2C+a+word+embedding+layer%2C+two+LSTM+layers%2C+two+hidden+layers%2C+an%0D%0Aoutput+layer+and+a+dropout+layer+after+each+layer.+The+performance+of+CNN+and%0D%0ALSTM+with+the+Bag+of+Words+(BoW)+model+as+word+embedding+in+this+study+is%0D%0Aquite+competitive+because+CNN+outperforms+LSTM+on+all+evaluation+measures+at%0D%0A37%25%2C+20%25%2C+and+10%25+data+imbalance+levels%2C+while+LSTM+outperforms+CNN+on+all%0D%0Aevaluation+measures+at+30%25+data+imbalance+levels.+Although+CNN+and+LSTM+have%0D%0Acompetitive+performance+results%2C+LSTM+consumes+significantly+longer%0D%0Acomputational+time+than+CNN.%0D%0A+%0D%0AKeywords%3A+Classification%2C+Imbalance+Data%2C+Deep+Learning%2C+CNN%2C+LSTM%2C+K-Fold%0D%0ACross+Validation%2C+Bag+of+Words+(BoW)%2C+COVID-19+news+headlines.%0D%0A%0D%0A%0D%0A%0D%0A+%0D%0AKinerja+metode+deep+learning+pada+klasifikasi+data+teks+dengan+rasio%0D%0Aketidakseimbangan+yang+berbeda+merupakan+diskusi+yang+penting+karena+data+yang%0D%0Aada+pada+dasarnya+tidak+seimbang.+Penelitian+ini+mencari+metode+deep+learning%0D%0Ayang+dapat+diandalkan+untuk+mengklasifikasi+data+judul+berita+berbahasa+Indonesia%0D%0Atentang+COVID-19+dengan+beberapa+rasio+ketidakseimbangan+data.+Beberapa+rasio%0D%0Aketidakseimbangan+data+dibuat+dengan+mengambil+sampel+dari+berita+event%0D%0Amenggunakan+simple+random+sampling+sebanyak+30%25%2C+20%25%2C+10%25%2C+dan+1%25.+Kinerja%0D%0Ametode+CNN+dan+LSTM+diuji+menggunakan+10-fold+cross+validation+dan%0D%0Adibandingkan+berdasarkan+akurasi%2C+presisi%2C+recall%2C+dan+f1-score.+Arsitektur+model%0D%0ACNN+yang+dibangun+pada+penelitian+ini+secara+umum+terdiri+dari+input+layer%2C+word%0D%0Aembedding+layer%2C+dua+convolutional+layer%2C+satu+pooling+layer%2C+flatten%2C+dua+hidden%0D%0Alayer%2C+output+layer+serta+batch+normalization+layer+dan+dropout+layer+berada%0D%0Asetelahnya+pada+setiap+layer+tersebut.+Arsitektur+model+LSTM+yang+dibangun+pada%0D%0Apenelitian+ini+secara+umum+terdiri+dari+input+layer%2C+word+embedding+layer%2C+dua%0D%0ALSTM+layer%2C+dua+hidden+layer%2C+output+layer+serta+dropout+layer+berada+setelahnya%0D%0Apada+setiap+layer+tersebut.+Kinerja+CNN+dan+LSTM+dengan+model+Bag+of+Words%0D%0A(BoW)+sebagai+word+embedding+pada+penelitian+ini+cukup+bersaing+karena+CNN%0D%0Amengungguli+LSTM+pada+semua+ukuran+evaluasi+pada+tingkat+ketidakseimbangan%0D%0Adata+37%25%2C+20%25%2C+dan+10%25%2C+sedangkan+LSTM+mengungguli+CNN+pada+semua+ukuran%0D%0Aevaluasi+pada+tingkat+ketidakseimbangan+data+30%25.+Meskipun+CNN+dan+LSTM%0D%0Amemiliki+hasil+kinerja+yang+saling+bersaing%2C+namun+LSTM+menghabiskan+waktu%0D%0Akomputasi+yang+jauh+lebih+lama+daripada+CNN.%0D%0A+%0D%0AKata+kunci%3A+Klasifikasi%2C+Ketidakseimbangan+Data%2C+Deep+Learning%2C+CNN%2C+LSTM%2C%0D%0AK-Fold+Cross+Validation%2C+Bag+of+Words+(BoW)%2C+Judul+Berita+COVID-19.&rft.publisher=FAKULTAS+MATEMATIKA+DAN+ILMU+PENGETAHUAN+ALAM+&rft.date=2022-08-05&rft.type=Skripsi&rft.type=NonPeerReviewed&rft.format=text&rft.identifier=http%3A%2F%2Fdigilib.unila.ac.id%2F65255%2F1%2FABSTRAK.pdf&rft.format=text&rft.identifier=http%3A%2F%2Fdigilib.unila.ac.id%2F65255%2F2%2FSKRIPSI%2520FULL.pdf&rft.format=text&rft.identifier=http%3A%2F%2Fdigilib.unila.ac.id%2F65255%2F3%2FSKRIPSI%2520TANPA%2520BAB%2520PEMBAHASAN.pdf&rft.identifier=+++LUTHFIA+NUR+AZIZAH%2C+1817031065++(2022)+KINERJA+METODE+CONVOLUTIONAL+NEURAL+NETWORK+(CNN)+DAN+LONG-SHORT+TERM+MEMORY+(LSTM)+PADA+KLASIFIKASI+DATA+JUDUL+BERITA+COVID-19.++FAKULTAS+MATEMATIKA+DAN+ILMU+PENGETAHUAN+ALAM+%2C+UNIVERSITAS+LAMPUNG.+++++&rft.relation=http%3A%2F%2Fdigilib.unila.ac.id%2F65255%2F