Web-based Document Classification Using A Trie-based Index Structure

Abstract

An automatic document classification system is useful to manage the massive quantities of documents such as the Web document collection. However, its complicated process of classification has become a serious problem when applying it to general services. In this paper, we suggest an efficient data structure for the document classification and develop a classification system based on a trie-based index structure. This efficient data structure reduces overheads for the task of document classification using naive Bayesian probabilistic models and makes it possible to implement commercial applications. In our system, both learning and classification are performed in a Web-based user interface rather than by a remote application, which contributes to achieve easy control of the classification process and the flexibility of diverse document provision.