As the spread of information through the Internet accelerates, more and more people are exposed to information through Internet portals, and the impact of comments is increasing day by day. Comments are used as a means to express readers' opinions in ...
As the spread of information through the Internet accelerates, more and more people are exposed to information through Internet portals, and the impact of comments is increasing day by day. Comments are used as a means to express readers' opinions in relation to posts, but malicious comments such as slander or gossip against others are becoming a social problem. To solve this problem, Naver uses an artificial intelligence called Cleanbot to block malicious comments, but due to the nature of the comments, they change rapidly and may include bias because they contain social issues. This study is an artificial intelligence governance study, which is conducted in two stages, a monitoring study and an implementation study. In the monitoring study, we check the performance of the cleanbot, and in the implementation study, we implement an alternative model of the cleanbot and compare the performance. In the implementation study, a malicious comment blocking model was implemented using a convolutional neural network, gated recurrent unit, and attention as technical mechanisms, and it was confirmed that the performance was higher than before.