Customer Complaints Clusterization of Government Drinking Water Company on Social Media Twitter using Text Mining

Social media is considered one of the most effective platforms to communicate between companies and customers. Frequently, the customer of a product or service sends complaints via social media. Customers' complaint data serve as a good suggestion for companies and organizations to improve their products and services. With the increasing number of customer complaints that have entered through social media accounts, government-owned drinking water companies need a more efficient way to extract information from complaint data. In this research, text mining is used to extract information about customer complaints against drinking water companies from social media Twitter. Latent Dirichlet Allocation (LDA) and self-organizing maps (SOM) approach is applied to model complaint topics and find out which are most frequently complained. The test results indicate grouping the data into five classes is the most appropriate model. Pipes leakage are the most frequently reported topics, 27.8% of total datasets.