ID-SMSA: Indonesian stock market dataset for sentiment analysis

Abstract

Social media has impacted daily life, affecting people’s habits regarding accessing and sharing information. Among the platforms, X (formerly Twitter) gives users the freedom of speech to express their subjects and topics. Hence, users express their opinions on every topic, from light-hearted to heavy topics such as politics and the economy. This vast opinion from users creates a valuable resource for research. This paper presents the Indonesian Stock Market Dataset for Sentiment Analysis (ID-SMSA), a collection of 3288 tweets discussing the top 10 largest market caps in the Indonesian stock market as of March 2023. The dataset is in Indonesian and an English translated version is provided, making it the first Indonesian-language dataset discussing the Indonesian stock market. Human annotators labelled each tweet as positive, neutral, or negative based on baseline annotation characteristics criteria created and reviewed by an expert in clinical psychology. A voting system determines which tweets to include in the dataset. This creates a consistent dataset that reflects clear and agreed-upon sentiments and removes ambiguous and contradictory data. The voted tweets include 2339 positive, 999 neutral, and 1025 negative sentiments. This dataset supports research into Indonesian stock market growth and the role of social media in financial discussions.