Host: The Japan Society for Management Information
With the recent spread of internet, the number of websites in where users can freely share information and own opinion is increasing rapidly. Therefore, user reviews are also increasing constantly and it is difficult for people to see all reviews. In this study, we propose a technique for automatic summarization of user reviews to analyze opinions of all users. Concretely, the automatic summarization problem is formulated by extracting important reviews from the viewpoint of combinational optimization, and the optimum solution is searched by genetic algorithm. The proposed technique is applied to an extraction problem of reviews on "jalan.net" that is one of online websites to show its effectiveness.