(Similarity) Application Of N-Gram On K-Nearest Neighbor Algorithm To Sentiment Analysis Of TikTok Shop Shopping Features

Lestari, Riska Dwi Ayu and Rintyarna, Bagus Setya and Dasuki, Moh (2022) (Similarity) Application Of N-Gram On K-Nearest Neighbor Algorithm To Sentiment Analysis Of TikTok Shop Shopping Features. Institute of Computer Science (IOCS).

Text 1910651020_Artikel.pdf Download (1MB)

Abstract

This study contains sentiment analysis on Twitter data with the
direction of sentiment on the TikTokShop feature. In this study,
the k nearest neighbor method is implemented in which the
metric distance cosine similarity is used with the value of the
nearest neighbor distance k = 3, 5, 7, and 9. In the modeling, a
k-fold cross-validation scenario is used with a value of k = 10
fold. This study also uses unigram, bigram, and trigram selection
features to handle imbalanced data using undersampling
techniques. From the modeling results, it is found that the best
modeling is the model with unigram feature selection with
nearest neighbor k = 3. From this model, the average accuracy
value is 89.92%, the average precision is 90.54% and the recall
average is 87.37%. In the test, the results showed that the
unigram feature selection had the best performance with 91%
accuracy, 92% precision, and 89% recall.

[error in script]

Item Type:	Peer Review
Keywords/Kata Kunci:	Analysist, KNN, N-Gram, Sentiment, Undersampling
Subjects:	600 Technology and Applied Science > 620 Engineering
Divisions:	Faculty of Engineering > Department of Electrical Engineering (S1)
Depositing User:	Bagus Setya Rintyarna | bagus.setya@unmuhjember.ac.id
Date Deposited:	23 Dec 2022 01:16
Last Modified:	23 Dec 2022 01:16
URI:	http://repository.unmuhjember.ac.id/id/eprint/15711

Actions (login required)

View Item