An Improved Expectation Maximization based Semi-Supervised Text Classification using Naïve Bayes and Support Vector Machine

Abstract

With the development of Internet and the emergence of a large number of text resources, the automatic text classification has become a research hotspot. As number of training documents increases, accuracy of Text Classification increases. Traditional classifiers (Supervised learning) use only labeled data for training. Labeled instances are often difficult, expensive, or time consuming to obtain. Meanwhile unlabeled data may be relatively easy to collect. Semi-Supervised Learning makes use of both labeled and unlabeled data. Several researchers have given algorithms for Text Classification using Semi-Supervised Learning. But still improving accuracy of Text Classification using Semi-Supervised Learning is a challenge. In the iterative process in the standard Expectation Maximization (EM) based semi-supervised learning, some unlabeled samples are misclassified by the current classifier because the initial labeled samples are not enough. To overcome this limitation, an EM based Semi-Supervised Learning algorithm using Naïve Bayesian and Support vector machine is proposed in this paper to improve accuracy of text classification using semi-supervised learning.