@prefix epid: .
@prefix void: .
@prefix xsd: .
@prefix cc: .
@prefix ep: .
@prefix eprel: .
@prefix dc: .
@prefix bibo: .
@prefix geo: .
@prefix owl: .
@prefix event: .
@prefix skos: .
@prefix foaf: .
@prefix rdfs: .
@prefix rdf: .
@prefix dct: .
<>
foaf:primaryTopic ;
rdfs:comment "The repository administrator has not yet configured an RDF license."^^xsd:string .
dct:hasPart ;
ep:hasFile ;
rdf:type bibo:Document,
ep:Document;
rdfs:label "ANALISIS PERBANDINGAN KINERJA BERT DAN INDOBERT DALAM MENDETEKSI ULASAN PALSU PRODUK E-COMMERCE BERBAHASA INDONESIA (Text)"^^xsd:string .
rdf:type bibo:Document,
ep:Document;
rdfs:label "ANALISIS PERBANDINGAN KINERJA BERT DAN INDOBERT DALAM MENDETEKSI ULASAN PALSU PRODUK E-COMMERCE BERBAHASA INDONESIA (Text)"^^xsd:string .
rdf:_1 .
bibo:abstract "The rapid growth of e-commerce in Indonesia has increased the use of product\r\nreviews as an important consideration for consumers before making purchases.\r\nHowever, the emergence of fake reviews has become a serious issue because it can\r\naffect consumer trust and reduce the credibility of e-commerce platforms. This\r\nstudy aims to compare the performance of Multilingual BERT and IndoBERT\r\nmodels in detecting fake reviews written in Indonesian.\r\nThis research employed an experimental method consisting of data collection,\r\nlabeling, preprocessing, model training, evaluation, and result comparison. The\r\ndataset was obtained from Kaggle and consisted of 18,242 Indonesian e-commerce\r\nproduct reviews. Approximately 10% of the dataset was manually labeled and then\r\nused in a pseudo-labeling process with the Multinomial Naive Bayes algorithm to\r\ngenerate additional training data. Both Multilingual BERT and IndoBERT were\r\ntrained using the same parameters to ensure a fair comparison. Model evaluation\r\nwas conducted using accuracy, precision, recall, F1-score, and confusion matrix\r\nmetrics.\r\nThe results showed that IndoBERT achieved the best performance with an accuracy\r\nof 87.12%, precision of 85.81%, recall of 83.01%, and F1-score of 84.39%.\r\nMeanwhile, Multilingual BERT achieved an accuracy of 86.30%, precision of\r\n83.23%, recall of 84.31%, and F1-score of 83.77%. These findings indicate that\r\nIndoBERT is more effective in understanding the characteristics of the Indonesian\r\nlanguage, especially informal e-commerce review texts. This study also\r\ndemonstrates that the pseudo-labeling approach can help increase the amount of\r\ntraining data, although the quality of additional labels greatly affects model\r\nperformance. Based on the research findings, IndoBERT is more recommended for\r\nimplementing fake review detection systems on Indonesian e-commerce platforms\r\ncompared to Multilingual BERT."^^xsd:string;
bibo:authorList ;
bibo:status ;
dc:hasVersion ,
;
dct:creator ;
dct:date "2026-06-03";
dct:isPartOf ;
dct:issuer ,
;
dct:subject ;
dct:title "ANALISIS PERBANDINGAN KINERJA BERT DAN INDOBERT DALAM MENDETEKSI ULASAN PALSU PRODUK E-COMMERCE BERBAHASA INDONESIA"^^xsd:string;
ep:hasDocument ,
;
ep:hasPublished ,
;
rdf:type bibo:Article,
bibo:Thesis,
ep:EPrint,
ep:ThesisEPrint;
rdfs:seeAlso .
rdf:type skos:Concept;
skos:prefLabel "004 Pemrosesan Data, Ilmu Komputer, Teknik Informatika" .
rdfs:label "22106050035_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf"^^xsd:string .
dc:format "text/html";
dc:title "HTML Summary of #76862 \n\nANALISIS PERBANDINGAN KINERJA BERT DAN INDOBERT DALAM MENDETEKSI ULASAN PALSU PRODUK E-COMMERCE BERBAHASA INDONESIA\n\n";
foaf:primaryTopic .
dct:hasPart ;
foaf:name "UIN SUNAN KALIJAGA YOGYAKARTA"^^xsd:string;
rdf:type foaf:Organization .
dct:isPartOf ;
foaf:name "FAKULTAS SAINS DAN TEKNOLOGI, UIN SUNAN KALIJAGA YOGYAKARTA"^^xsd:string;
rdf:type foaf:Organization .
foaf:familyName "Marsha Kamila"^^xsd:string;
foaf:givenName "NIM.: 22106050035"^^xsd:string;
foaf:name "NIM.: 22106050035 Marsha Kamila"^^xsd:string;
rdf:type foaf:Person .