Structured textual data monitoring based on a rough set classifier

Abstract

Text is frequently stored in structures that are frequently complex and sometimes too large to be fully understood and/or apprehended. This problem has concerned the data mining community for many years as well as the information's community. Many algorithms have been proposed with the objective of obtaining better answers to the queries made and to obtain better queries that can respond to the questions that are in the users mind. Some of those algorithms are based on the relations between the concepts. But some of those relations are also dynamic and are, themselves, relevant information. This paper describes and adaptation of one of those methods, based on the Rough Sets theory, in order to detect changes in the existing relations between the stored concepts and, through that, to detect new relevant aspects of the data.