Towards an intelligent text categorization for web resources: An implementation

We propose the concept and implementation of a software system, TCAT (Text CATegorization) system, for an automatic recognition of a topic of an Internet document. In the training mode the user provides the system with a list of topics and sets of documents representing each topic (supervised learning). In the recognition mode the system automatically classifies previously unseen document to a topic category. A simple learning algorithm is devised and implemented. The results of the classification are presented to the user in the form of a set of linguistic terms. Some new measures of correctness of the classification are proposed. The implemented system processes documents in several popular Internet-related formats.