2015년 추계학술발표대회 논문집 제22권 제2호(2015. 10) 지오소셜 정보 기반 개인 맞춤형 경보 시스템 원형 구현 부둑 티엡, 뉘엔 반 퀴엣, 김경백 전남대학교 전자컴퓨터공학부 e-mail: ductiep91@gmail.com, quyetict@gmail.com, kyungbaekkim@jnu.ac.kr Prototype Implementation of a Personalized Warning Notification System based on Geosocial Information Vu Duc Tiep, Nguyen Van Quyet, Kyungbaek Kim Dept of Electronics and Computer Engineering, Chonnam National University 요 약 Nowadays a disaster event such as a building on fire, an earthquake or typhoon could occur any time, and any where. In such event, a warning notification system is a vital tool to send warning notifications to at-risk people in advance and provide them useful information to escape the dangerous area. Though some systems have been proposed such as emergency alert system using android, SMS or P2P overlay network, these works mainly focus on a reliable message distribution methods. In this work, we introduce a full prototype implementation of a personalized warning notification system based on geosocial information, which generates a personalized warning message for each user and delivers the messages through email or an android application. The system consists of four main modules: a web interface, database, a knowledge-based message generator, and message distributor. An android application is also created for user to receive warning messages on their smart phone. The prototype has been demonstrated successfully with a building-on-fire scenario. 1. Introduction A warning notification system is a system that can quickly send out important emergency information targeted to specific area. The system is essential to warn the at-risk population about dangerous event in advance and provide them with survival information. Recently, researchers have proposed several warning notification system. For example, in [1] and [2], Short Message Service (SMS) were proposed to be a reliable method to deliver emergency warning message. Similarly in [3] and [4], the authors take advantage of the Cell Broadcasting Service (CBS) to distribute warning messages over mobile network. Another work as in [5] has developed an emergency alert system using android, which aim to solve the problem of sending warning message to a wide range of people in bad network traffic by splitting bulk messages. With similar goal, a notification system called GSFord was proposed in [6]. The GSFord built a robust geo-aware P2P overlays to successfully deliver notification message in extreme situations. These works mainly focus on message distribution methods in emergency case. However, only broadcast some general information about the emergency event is not enough. If the message was personalized specifically to an individual, it could be much more meaningful. In [7], a design of knowledge-based message generator for personalized warning notification system was proposed. The personalized warning message is generated based on geosocial information such as user current location, subscribed area and relationship as well as event information. So far a full prototype system has not yet been implemented. Therefore, in this work, we want to build a full prototype of a personalized warning notification system based on geosocial information. There are several key requirements that a personalized warning notification system must satisfy. First, it must have an interface such as a web interface to collect geosocial information about user and emergency event. Through the interface, user can update their contact information, social relationship, current location, and subscribe areas while administrator can create, update and notify emergency event. A database is required to store the collected geosocial - 332 -