Application of text mining employing k-means algorithms for clustering tweets of Tokopedia

Abstract

In this current digital era, people tend to shop online. Because of that, there are currently many e-commerce companies that can satisfy the various needs of society in shopping. Each company certainly has a strategy to attract consumers to shop at their shopping place. One of the media commonly used to attract consumers is social media. Tokopedia is one of the biggest marketplaces in Indonesia and is also active in utilizing Twitter as their social media mean. Therefore, it is essential for Tokopedia to pay attention to tweet contents interesting enough to be publicized. By applying text mining using K-Means Clustering algorithm, it can be seen which types of tweet contents that are attractive for Tokopedia consumers. Out of 885 Tokopedia tweets that have been collected, a clustering is then done using K-means algorithm, resulting 48 cluster tweets. Then, from the 48 clusters, they are further grouped into 5 major groups. Based on the results of the grouping, it can be seen that the most interesting content deals with quiz prizes and the least attractive content is on lifestyle.