Analisis Sentimen Terhadap Kasus Pembuangan Bayi Melalui Media Sosial X Menggunakan Naive Bayes

SERAN, Roswita Marlina (2026) Analisis Sentimen Terhadap Kasus Pembuangan Bayi Melalui Media Sosial X Menggunakan Naive Bayes. Undergraduate thesis, Universitas Katolik Widya Mandira.

	Text ABSTRAK.pdf Download (328kB)
	Text BAB I.pdf Download (73kB)
	Text BAB II.pdf Restricted to Repository staff only Download (144kB)
	Text BAB III.pdf Restricted to Repository staff only Download (182kB)
	Text BAB IV.pdf Restricted to Repository staff only Download (552kB)
	Text BAB V.pdf Download (12kB)
	Text DAFTAR PUSTAKA DAN SURAT BEBAS PLAGIAT.pdf Download (118kB)

Abstract

The case of baby abandonment is a social issue that triggers strong emotional reactions from the public in the digital space. The social media platform X has become a public arena where various responses emerge, ranging from condemnation to expressions of empathy, as well as opinions related to economic pressures that may underlie such incidents. This study aims to analyze public sentiment toward baby abandonment cases on social media X based on humanitarian and economic aspects using the Naive Bayes method. Data were collected through a crawling technique on the X platform within the period of January–October 2025, along with manual data collection to enrich the variation of sentiment expressions. After the preprocessing stage, 1,100 clean tweet data were obtained. Sentiment labeling was conducted using a modified InSet lexicon, after which the data were transformed using TF-IDF weighting and classified using the Naive Bayes method. The results show that negative sentiment tends to dominate as a form of public condemnation, while positive sentiment reflects empathy and social concern. Model evaluation using a 70% training data and 30% testing data split resulted in an accuracy of 71.30%, with a precision of 70.86% and recall of 73.81% for the negative class, and a precision of 71.79% and recall of 68.71% for the positive class. The F1-score obtained was 72.30% for the negative class and 70.22% for the positive class, indicating that the model has a fairly optimal capability in classifying public sentiment related to this issue.

Actions (login required)

View Item