Abstract
Online videos platforms such as YouTube is the most popular social media platform in terms of user numbers in Indonesia. YouTube is also one of the most popular online platforms for accessing religious lectures.
Users can provide feedback on the videos through comments, likes, and shares. On the other hand, sentiment analysis in the Indonesian language is getting popular, but few have tapped the vast unstructured data source on YouTube.
Comments and reviews from viewers are valuable feedbacks for improvements. The review on YouTube is an essential resource to be analyzed by a preacher.
However, manual analysis of YouTube reviews is complicated due to a large amount of review data. Therefore, this study aims to analyze sentiment on YouTube video reviews.
In this paper, we employed the Lexicon and Latent Dirichlet Allocation (LDA) to analyze a total of 2575 review data. In this case study, we mined YouTube user's review to understand the netizen's opinion on a famous Islamic Preacher in South East Asia, namely Ustadz Abdul Somad (UAS).
We employed the Google Apps Script (GAS) with Javascript coding language to crawl YouTube review data. Based on the results, the lexicon method successfully analyzed sentiments with an accuracy of 70%.
Furthermore, 98% of YouTube users gave positive reviews on the UAS videos lecture. This study is a stepping stone for more complex sentiment analysis regarding text pre-processing and algorithm robustness.