Application of K-means Clustering Algorithms in News Comments

More and more netizens prefer to comment on social hot issues today and their views become very useful for government decision-making. Specially, news and related comments often influence decision behavior of officers. However, it becomes a key problem to analyze them automatically in order to provide references for decision-making. One of effective way is to cluster news comments. In this paper, we discuss the k-means clustering algorithm and how to cluster news comments in order to obtain types of a special news comments. And we do an experiment on a real dataset collected from the news recommender system we developed for government decision-making. Primary results are shown that our k-means clustering method is effective and can be taken as an analysis method used in our recommender system.