Twitter Classification of Public Service Complaints
Fatkhurrochman, Friandy Dwi Noviandha, Arief Setyanto

Universitas Amikom Yogyakarta
Yogyakarta, Indonesia

Abstract

Public service become one of the factors of public satisfaction level to government performance. Complaints from society become mediators to improve public services according the wishies and needs of the society. Twitter is a popular micro-blogging in society. Society can tweets about their activites, experiences, complaints to the internet easily and realtime. In this research the classification of society into water, electricity, and roads. The twitter classification is built using the K-Nearest Neighbor (KNN) algorithm. The feature selection in this research are using term frequency (TF), document frequency (DF), information gain, and chi square. In this research, combination of features that have been produced from the previous feature selection. From this research shows that K-NN is able to classify complaints well, it is proved with the accuracy of 83.75% and 77.08% for complaints and types of complaints. For the classification of types of water complaints, the average values generated for each parameter are 87.5%, 87.8%, and 87.49%.

Keywords: public service, twitter classification, K-Nearest Neighbor, F-Measure

Topic: Computer-based learning