An advisory method for BBS users and evaluation of BBS comments

Abstract Recently, BBS has a problem which threatens the safety of BBS community: vandalism comments. Such comments occur by fault and by misunderstanding. Thus, showing the probability of becoming vandalism comment can reduce such vandalism comments. We proposed a method to calculate the probability of becoming vandalism comment. This method classified each comment into “Normal comment”, “Vandalism comment” or “Flaming comment” using word and pair of words based on Bayesian theorem. As the result of experiment, we showed that proposed method could classify normal comments into normal comment class with 64.7% accuracy and vandalism or flaming comments into vandalism comment class with 85.4% accuracy.