ASPECT-BASED SENTIMENT ANALYSIS PADA RESPONS SURVEI OPEN-ENDED MENGGUNAKAN LDA, BERT, DAN SVM

Rahmawati, Dian (2025) ASPECT-BASED SENTIMENT ANALYSIS PADA RESPONS SURVEI OPEN-ENDED MENGGUNAKAN LDA, BERT, DAN SVM. Undergraduate thesis, Universitas Pembangunan Nasional Veteran Jawa Timur.

	Text (Cover) 21082010100_Cover.pdf Download (1MB)
	Text (Bab 1) 21082010100_Bab1.pdf Download (127kB)
	Text (Bab 2) 21082010100_Bab2.pdf Restricted to Repository staff only until 10 April 2028. Download (408kB)
	Text (Bab 3) 21082010100_Bab3.pdf Restricted to Repository staff only until 10 April 2028. Download (495kB)
	Text (Bab 4) 21082010100_Bab4.pdf Restricted to Repository staff only until 10 April 2028. Download (1MB)
	Text (Bab 5) 21082010100_Bab5.pdf Download (8kB)
	Text (Daftar Pustaka) 21082010100_Daftar Pustaka.pdf Download (91kB)

Abstract

Aspect-based sentiment analysis on open-ended survey responses is important in understanding the opinions and experiences of participants in an event. This research aims to classify sentiment and identify key aspects in open-ended survey responses from the 2021 and 2022 FKBM-IK national seminar events. The methods used include Latent Dirichlet Allocation (LDA) for topic modeling, as well as Bidirectional Encoder Representations from Transformers (BERT) for word embedding. A sentiment classification model was developed using Support Vector Machine (SVM). The results showed that the SVM model with IndoBERT word embedding and Synthetic Minority Over-sampling Technique (SMOTE) resampling technique achieved optimal accuracy with an average accuracy score of 94%, precision 89%, recall 93%, f1 score 91%. Topic modeling analysis with LDA successfully identified 7 main topics that reflected important aspects of the participants' responses. The majority of sentiments found were positive, especially in the topics of event success and event expectation.

Item Type:

Thesis (Undergraduate)

Contributors: