Web Intrusion Detection Using Probabilistic Network with Undirected Links

• Main Authors: Chun-Yuan Chan, 詹純源
• Other Authors: Dong-Her Shih
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/05868753892707153716

碩士 === 國立雲林科技大學 === 資訊管理系碩士班 === 91 === In this paper we present Bayesian probabilistic network to web intrusion detection through anomaly detection. A Bayesian network (or Probabilistic network) is a graphical model that can encode prior state distributions among dependent objects into its structure and probabilistic distribution tables. BN is used to learn long-term profiles of normal activities in web service LogFiles, and to measure deviation of observed activities from normal profiles that can be used to detect intruders. We adapt a new symmetric structure for Bayesian networks with undirected links between nodes. Traditionally BN supports only directed links between nodes. We adapt BN by replacing directed links with undirected links, and using joint probability tables instead of conditional probability tables. Our experiments show that it is possible for web intrusion detection by using Bayesian networks.