Abstract
By looking at the reviews of potential users, you can see the responses of other users who have used the application first, besides the application reviews can be used as input for developers. Clustering reviews can use text mining. In this study, the Kmedoids algorithm was used to cluster reviews with Fasttext word embedding to represent the review's word units into vectors. The data used in this study amounted to 2,729 taken from the PeduliLindung application comment column on the Google Playstore. The results of the evaluation using the Davies Bouldin index or abbreviated as DBI by comparing sixteen trials and the best experiment was obtained, namely, an experiment using data without stemming, cbow architecture, using the Manhattan distance metric, and an experiment using data without stemming, cbow architecture, and using the cityblock distance metric with the same DBI value of 2.93 which resulted in two clusters of reviews. In the first cluster there were 2,183 reviews while in the second cluster there were 546 reviews.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
