Article Details

PENAMBANGAN TEKS UNTUK MENGENAL PEMBERITAAN PALSU DAN IMPLIKASINYA DI MASYARAKAT INDONESIA

Oleh   Yana Dwi Putra Nugraha [29016026]
Kontributor / Dosen Pembimbing : Dr. Manahan Parlindungan Saragih Siallagan;
Jenis Koleksi : S2 - Tesis
Penerbit : SBM - Sains Manajemen
Fakultas : Sekolah Bisnis dan Manajemen (SBM)
Subjek :
Kata Kunci : hoaks, analisis neural network, penambangan teks, deteksi kebohongan
Sumber :
Staf Input/Edit : Neng Kartika  
File : 8 file
Tanggal Input : 10 Mei 2022

Indonesia became one of the countries with the biggest number of internet users worldwide in which over half of them accesses the social media platforms. Social media is currently one of the most favored means to exchange information with friends and colleagues. Regrettably, there are also irresponsible users who transmit numerous deceptive information. Hoax information is one example that is getting a lot of attention and concerns. The current phenomena of hoax occur in many areas, resulting in fear and anxiety within the society. Due to its deceptive nature, hoax information can trick the recipients into believing the deceit information instead of the actual truth. This may lead into confusion and turmoil, caused by the false truth. However, stakeholders may attempt to reduce chaos by minimizing the possibility of hoax being delivered to the masses. There are methods to distinguish deceitful messages from the legitimate ones. By employing text mining on hoaxes, unrevealed patterns may be exposed to help produce an effective model. The purpose of this study is to generate a model of identifying hoax shared on the internet by text-mining contents from the internet through writing style analysis using neural network. The process may then be constructed to propose an early detection of hoax according to a validated pattern from the analysis. The study finds that the resulting model is capable to predict hoax information in the social subject. Though, to be able to accurately predict many different subjects, the indicators need not to be limited to a certain set of writing style features. Furthermore, the study recommends involving contextual indicators to better predict hoax information, hence producing a reliable model for an early hoax detection.