Effective extraction method of loss aversion utterances based on the expected utility

Many studies of the human behavior understanding are forwarding by using sensors and huge data on the Internet in recent years. On the other hand, the research of the behavioral modification is also preceded continuously. These make users to change their behavior and realize better society which is represented by prohibition of smoking and a route guidance of a car navigation system. This research aims to construct the model of the behavioral modification by the information technology. This paper examines the technique of extracting the knowledge about the behavioral modification from the question-answering sites on the Internet in which user's problems and opinions often appear. Specifically, the extraction of the utterances expressing the loss aversion is carried out from a series of question-answering text sentences. The loss aversion means that men strongly tendency to select to avoid a loss rather than get a profit. Therefore, it is possible to build a system which presents users a series of actions of maximizing a profit by investigating what kind of loss aversion actions men perform. This paper shows examples of the loss aversion utterances in question-answering text sentences and tries to classify loss aversion utterances on the basis of the expected utility index that is computed only from the text sentences.