A Customer Complaint Analysis Tool for Mobile Network Operators

Abstract

Mobile Network Operators (MNOs) are eager to learn more about complaint behaviour of their subscribers. In this demo, we study topic modeling approach for extracting relevant problems experienced by subscribers of MNOs in Turkey and visualize the topic distributions using LDAvis data analytics tool. For building topic models using Latent Dirchlet Allocation (LDA), we have built customer complaint text dataset of subscriber complaints for each MNOs from Turkey's largest customer complaint website. The proposed analysis tool can be used as customer complaint analysis service by MNOs in Turkey to gain more insight. We have also validated our generated topic model using another dataset obtained from Turkey's largest online community website. Our results indicate similar and dissimilar topics of complaints as well as some of the distinctive problems of MNOs in Turkey based on their subscriber's experiences and feedback.