Automatic Document Classification

Abstract

Enormous amount of documents are generated everyday life. There is always a need to retrieve the documents over any medium. This has come up with the solution of classifying the documents with appropriate label. A lot of research has been done on this topic to classify them using different classifiers. The classifier used in this research is Naïve Bayes classifier, due to its simplicity. The Naïve Bayes classifier classifies a document only under one class no matter by what fraction posterior probabilities of other classes are smaller. By considering the fraction by which other associated terms are smaller we rank a document more into a specific context but also little less into another context. The Apriori algorithm is used to find the frequent patterns out of the document which will give the context of the document and will help in labeling the document with more appropriate classification tag. The proposed approach is to classify the document with Naïve Bayes classifier at first level and then finding associated terms from documents and comparing them with the already mined frequent patterns from the train dataset. This two level classification gives the more precise label to the document.