Implementing Personalized Web News Delivery Service Using Tales Of Familiar Framework

Abstract

We have previously proposed the framework of Tales of Familiar (ToF), where an agent (called familiar) autonomously delivers information from various data streams as exclusively personalized tales for individual users. Based on the To framework, this paper implements a news delivery service, where a stuffed doll (as a familiar) tells a user the latest and personally selected news headlines, by matching user’s interests with Web news resources. In the implementation, we especially address three challenges: duplication of tales, value estimation of tales, and delivery timing of tales. We deploy the service in an actual household. The empirical result shows that the subject felt it useful that the familiar pushed his interesting news, automatically. We also evaluate how much the developed service was able to cover the technical issues.