The result of the current technological developments makes increasingly tight telecommunication provider competition. Various opinions expressed by customers about telecommunication providers are found in social media. Twitter is widely used by the public to share information and socialize; also, to share opinions, express opinions of a product or service, and provide reviews on communication providers. Many reviews are provided by users on Twitter, making it hard to classify manually. Therefore, to make it easy to classify the tweets, an automation system is needed to determine that a comment is positive or negative. Maximum Entropy can be used for sentiment analysis of Indosat Ooredoo tweets. On these grounds, this study explains how to define tweets into positive and negative classes with applications created using Java. Based on the research and after testing, the Maximum Entropy obtained an accuracy value of 86.21% and an AUC value of 0.968.