A self appreciating approach of text classifier based on concept mining

A good text classifier is a classifier that efficiently categorizes large sets of text documents in a reasonable time frame and with an acceptable accuracy. Most of the text classification approaches are based on the statistical analysis of a term, either a word or a phrase. Though statistical term analysis shows the importance of the term, it is tedious to analyze when more than one term has the same frequency level but one may contribute more meaning than the other. When analyzing by concept based mining it is easy to identify the most contributable term of the document. The performance of the categorizer is mostly depends on how well the system is trained for different categories. This paper introduces a novel approach of self appreciating model in which each of the positive testing is redirected to the training system to make the training stronger and stronger at all possible test events.