KARUNIAWAN, ALLAN (2025) DETECTING SPAM COMMENTS ON INSTAGRAM USING SVM AND NEURAL NETWORK. S1 thesis, UNIVERSITAS KATOLIK SOEGIJAPRANATA SEMARANG.

Preview	Text 18.K1.0043-ALLAN KARUNIAWAN_COVER_a.pdf Download (620kB) | Preview
	Text 18.K1.0043-ALLAN KARUNIAWAN-BAB_I_a.pdf Restricted to Registered users only Download (718kB)
	Text 18.K1.0043-ALLAN KARUNIAWAN-BAB_II_a.pdf Restricted to Registered users only Download (717kB)
	Text 18.K1.0043-ALLAN KARUNIAWAN-BAB_III_a.pdf Restricted to Registered users only Download (688kB)
	Text 18.K1.0043-ALLAN KARUNIAWAN-BAB_IV_a.pdf Restricted to Registered users only Download (1MB)
	Text 18.K1.0043-ALLAN KARUNIAWAN-BAB_V_a.pdf Restricted to Registered users only Download (716kB)
Preview	Text 18.K1.0043-ALLAN KARUNIAWAN-DAPUS_a.pdf Download (721kB) | Preview
	Text 18.K1.0043-ALLAN KARUNIAWAN-LAMP_a.pdf Restricted to Registered users only Download (775kB)

Abstract

Due to the time when the social media applications are abundant, Instagram has significantly grown as a fertile ground for spam in so many ways that famously affect user experience as well as the sanctity of the platform. In this project we propose an in-depth presentation of SVM- and Neural Network-based spammer detection techniques on Instagram. We show how the two conventional machine learning techniques and the modern Neural Network models can be combined to further enhance the accuracy and bring stability in detecting spam. The used dataset comprises a large number of Instagram posts, preemptively classified as spam and non-spam. Features extracted were based on textual content and patterns of user activity, as well as metadata. The reason to apply the SVM model is that spaces are very efficient in high dimensions, making it a very good choice for the small- to medium-sized dataset where classes are separated with a clear margin. On the other hand, a Neural Network has been used to extract complex patterns and semantic subtleties hidden in the text data.

Actions (login required)

View Item