Chinese text mining based on distributed SMO

The paper presents the good classification accuracy of Support Vector Machine (SVM) with optimized parameters by particle swarm optimizer (PSO) in Chinese text classification. Through the simulation we also see that its training speed is slow when we deal with large amounts of texts in dataset, and it affects classification performance. Platt´s sequential minimal optimization (SMO) is one of the fastest algorithms for training SVMs, so we introduce the distributed SMO using multiple core processors to process the training data as a fast classification of large amounts of texts in dataset.