A hybrid neural network approach for automated classification of online documents using a domain nonspecific thesaurus

Information overloading has become a serious problem due to the exponential growth of the use of the Internet, emails and other online information resources. One of the solutions to this problem is the deployment of an automated classification system so as to provide an efficient means to manage the ever increasing amount of information and documents. A hybrid neural network approach for the automated classification of text –based articles is reported in this paper. In this study, the research has centered on the classification of newsgroup documents (postings) in accordance to the relevant newsgroups. The classification was initially based on the original documents. The documents are then reclassified with replacement of words from a domanin nonspecific thesaurus. Experiments based on over 40,000 news articles have been carried out and the results are found to be compatible in both cases. The technique can be extended to other online documents such as email articles and web pages.