Dataset Indonesia untuk Analisis Sentimen

This paper present a text dataset which can be used in the field of text analysis, especially sentiment analysis. This dataset covers the primary data which consists of 10,806 lines of Indonesian text data originated from Twitter social media, which categorized into three categories that are positive, negative, and neutral; and the raw data which consists of 454,559 lines of unprocessed data. Other than that, on the labeled data, the data is cleaned by removing many kind of noises in the data, such as symbols or urls. In this paper, the presented dataset is tested using a sentiment analysis model to make sure that this dataset is suitable to be used in the field of text analysis. The testing is done by measuring the model accuracy which is trained using this dataset and then comparing it to other model which is trained using already published dataset. After testing the data using various algorithm, such as SVM, KNN, and SGD, the accuracy result between our data and the comparison data are more or less equal with around 4% to 12% differences in accuracy, and prove that the dataset presented in this paper is feasible to be used in sentiment analysis. Dataset can be downloaded from link at conclusion section