Abstract—In this paper, we propose a method to extract foreigner interests for Japanese culture from interactive digital contents, making use of quick exchange trait of instant messages (IM) on Social Network Service (SNS). It is difficult for foreigners who living in Japan to know their interest about Japanese culture because of language barrier and cultural differences. The method enables them to search how to enjoy Japanese culture based on their interests. From the experiment, we found that people from the same country tend to be more interested in the same topics. The result indicates we can provide excitements for foreigners from a specific country, preparing appropriate topics for individual countries in advance. It implies we can develop an automatic consultation system to introduce traditional Japanese culture to foreigners. Index Terms—Data mining, extract topic, machine learning, Naïve Bayes classifiers. I. INTRODUCTION Japan has numbers of tangible and intangible cultural properties which attract attention from many foreigners. This country is holding various cultural activities and cultural experience programs [1]. By boosting awareness and interests of Japanese culture among foreigners living in Japan, both of the foreigners and Japanese would be benefit. First, foreigners’ life will be enhanced from knowing there are various cultural properties around the place they are living in. Second, Japan’s cultural maintenance, inheritance and development will be improved from insights of foreigners for Japanese cultural properties. However, language barrier and cultural differences have prevented foreigners from knowing their interests in Japanese culture [2]. It is difficult to suggest potentially interesting topics to a foreigner. Foreigners usually have no knowledge on individual cultural properties, but they have unclear categories of cultural properties they want to experience. Their expectation for those categories is formed through reputations for the Japanese culture in their countries. Suppose we have topics on Japanese cultural properties to be presented to foreigners who have just come to Japan. We can assume foreigners of specific background (nationality, gender, etc.) will likely be interested in particular topics his peers from the same country are interested in. If we know topics popular to foreigners of a specific background, we can make the topics burst in conversation with them. It enables the Japanese culture to be Manuscript received November 3, 2016; T. N. Le is with Ritsumeikan University, Shiga, 5258577 Japan (e-mail: le@de.is.ritsumei.ac.jp). H. Shimakawa is with the Data Engineering Lab, Ritsumeikan University, Shiga, 5258577 Japan (e-mail: simakawa@cs.ritsumei.ac.jp). spread and received more efficiently. To provide appropriate topics for individual foreigners, we need a proper method to find suitable topics as clues [3] for each foreigner to be inclined to experience Japanese cultural properties. With the method, we will take a huge leap in helping them understand their interests for Japanese cultural properties. In this paper, we propose a method to extract foreigner interests of the Japanese culture from interactive digital contents. The method makes use of quick exchange trait of instant messages on SNS. It extracts interested topics in the Japanese culture. The method enables foreigners to search how to enjoy the Japanese culture based on their interests. From the experiment, we found that people from the same country tend to be more interested in specific topics. It implies we can introduce traditional Japanese culture which excites foreigners from every country, preparing topics suitable for each country. II. EXTRACTING USER INTEREST USING IM A. Familiarity for IM The instant message (IM) on social network service (SNS) sites is being widely used as a primary means of contact. It is free and supports group chat. Besides, it allows users to exchange of digital contents (text messages, pictures, videos, icons, and so on) quickly and conveniently. Nowadays, many people are so familiar to use IM on SNS. They can communicate their opinion easily, even among far places, if the time is convenient for both participants. B. Naïve Bayes Classifiers Naive Bayes classifiers belong to a family of simple probabilistic classifiers based on the Bayes’ theorem with strong independence assumptions between the features [4] [5]. They can predict class membership probabilities, such as the probability that a given sample belongs to a particular class. Using Naive Bayes’ Theorem,   P doc cat the probability that the user has interests in class cat when we get document doc , is computed as:        doc P cat P cat doc P doc cat P | |  where   P doc cat is the conditional probability of the document occurring when we know the user is interested in class cat . Since each document is composed of multiple words,   P doc cat can be decompose as: Extracting Foreigner Interests for Japanese Culture from Interactive Digital Contents Thi Ngoc Le and Hiromitsu Shimakawa 108 International Journal of Information and Electronics Engineering, Vol. 7, No. 3, May 2017 doi: 10.18178/ijiee.2017.7.3.671 revised May 6, 2017.