Abstract
Owing to the drastic development of the information technologies and the popularity of WWW applications, the common users have a more efficient and convenient environment to communicate and exchange information with each other. In the cyberspace, a great number of digital documents and transaction data make the Internet a huge knowledge depository. In order to efficiently manage the various documents and explore the target knowledge users, automatic document and user classification mechanisms are required for the modern enterprises to provide effective knowledge service. Automatic classification mechanisms have gradually been developed to reduce the human efforts dedicated to document/user classification. Concerning the high variety of document contents and user behaviors over the Internet, it is not appropriate for the modern organizations to exploit the document and user characteristics simply by human decision. This thesis develops an approach to automatically and consistently determine the document/user categories according to the document keywords and browse history. Previously, only the single-level classification approaches of documents and users are concerned. However, the single-level classification mechanisms cannot meet the organization operation requirements. Due to the complexity of enterprise processes,products and services, automatic multi-level classification methodologies of enterprise documents and users ar e explored in this thesis to fulfill the realization of intelligent document/knowledge management. In order to evaluate the feasibility and effectiveness of the proposed methodologies, a web-based prototype system is developed and a demonstration case is provided. The decision support model as well as the technology aims at providing enterprises an effective classification approach that can be applied in CRM systems for efficient relationship marketing or KM systems for effective document security management.