Comparison of Fuzzy Matching and Naïve Bayes for Dialogue Design in Visual Novel

This research compares two methods for Dialog Design in a Visual Novel Game. Fuzzy Matching and Naive Bayes are used in this method to analyze which method is appropriate for generating responses from the story dialogue. The designed story dialogue allows players to provide free text input, which matches an established in-advance response. The Fuzzy Matching method calculates the input and response similarity value. In contrast, the Naive Bayes method is used to classify player input into the appropriate response category. The evaluation results show that the results using the Naive Bayes method achieve an accuracy of 81.03 % while the Fuzzy Matching method results in an accuracy of 20.69 %. The result shows that the Naive Bayes method is more effective in designing dialogue systems in visual novels.