PEMODELAN TOPIK PADA KELUHAN PELANGGAN MENGGUNAKAN ALGORITMA LATENT DIRICHLET ALLOCATION DALAM MEDIA SOSIAL TWITTER

Abstract

Large scale social restrictions (PSBB) is a policy issued by the Government of Indonesia as one of the efforts to reduce the spread of the Covid-19 virus. The impact of the policy is that it requires people to conduct activities online . This makes the internet users in Indonesia in the year 2020 up to 73.7%. Each provider must be able to determine strategies in order to maintain the quality of service and customer loyalty. Good reputation for the company is also important, so customers want to use internet services through their company. One of them is by listening to the complaints of the customers towards the company. In this research, modeling the topic of customer complaints carried out using the Latent Dirichlet Allocation Algorithm. The Latent Dirichlet Allocation Algorithm was chosen because the method has good performance. The topic modelling process is carried out using the gibbs sampling estimation. The topic that is often complained to First Media is that internet was turns off while working, while for IndiHome is that the internet often turns off and disconnect. Based on the results of the interpretation, 70% for First Media and 81,81% for IndiHome that these topics had been in accordance with what is complained by customers through their tweets. From the topic that have been known, it can be used as an evaluation for their company in order to maintain service quality and customer loyalty