Dwiyan Putra, Teguh (2026) Analisis Sentimen Opini Masyarakat Indonesia Pengguna Twitter Terhadap Layanan Indihome Menggunakan Metode Naive Bayes Dan Support Vector Machine (SVM). Other thesis, Universitas Islam Riau.

Text skripsi_203510040_watermark.pdf - Published Version Restricted to Registered users only Download (4MB) | Request a copy

Abstract

The development of technology today is very rapid, especially related to information technology, namely the internet. Indihome is a digital service that provides internet, landline telephone and Interactive TV (Indihome TV) with a variety of package options and additional services that can be selected according to needs, IndiHome is one of the internet providers in Indonesia with quite a lot of users. This study collects review data about IndiHome sourced from the Twitter platform and uses the Naïve Bayes Algorithm and Support Vector Machine, and also which will later compare the results obtained from the two models. Thus, it is expected to understand the pattern of sentiment, attitudes, and views of the community more systematically and objectively. The research was carried out through methodological stages such as collecting research data obtained from Twitter using the keyword "IndiHome lang:id". Data Pre-Processing: Cleaning and processing data to remove noise, such as links, punctuation, and unimportant words. Dataset Formation: Forming a dataset that includes positive, negative, and neutral sentiments with Lexicon Based and TextBlob labeling based on tweet content analysis and using 3 data split ratios. Implementation of Naïve Bayes Algorithm and Support Vector Machine: Applying Naïve Bayes and Support Vector Machine algorithms to classify sentiments from the dataset that has been formed. Model Evaluation: Evaluating the performance of the sentiment classification model using relevant metrics, such as accuracy, precision, recall, and F1-Score. Result Analysis: Analyzing the results of sentiment classification to gain a deeper understanding of public attitudes and views towards IndiHome. The Support Vector Machine method with Lexicon Based labeling at a ratio of 80% : 20% shows very good performance with accuracy, precision, recall, and f1 score of 94%.

Actions (login required)

View Item