Klasterisasi Penjawab Berdasar Kualitas Jawaban pada Platform Brainly Menggunakan K-Means

Abstract

Brainly is a Community Question Answering (CQA) educational platform that makes it easy for users to find answers based on questions posed by students. Questions from students are often answered quickly by many answerers interested in the field being asked. The number of available answers is the choice of students to be able to receive answers and give a good rating to the answerer. Based on the number of good ratings, an answerer can be said to be an expert in certain subjects. Therefore, this research focuses on finding expert answering groups who have quality answers. K-means clustering is possible to group the answering data into two different clusters. The first cluster is expert users with ten respondents, and The second cluster is a non-expert cluster with 474 respondents. The expert cluster data is expected to help the questioner to be able to ask questions directly to the experts and obtain quality answers. Meanwhile, the number of clusters is determined based on the test results using a silhouette score that obtains a value of 0.971, with the optimal number of clusters being two clusters.