The vast array of medical text data represents a valuable resource that can be analyzed to advance the state of the art in medicine. Currently, text mining methods are being used to analyze medical research and clinical text data. Some of the main challenges in text analysis are high dimensionality and noisy data. There is a need to develop novel feature transformation methods that help reduce the dimensionality of data and improve the performance of machine learning algorithms. In this paper we present a feature transformation method named FFTM. We illustrate the efficacy of our method using local term weighting, global term weighting, and Fuzzy clustering methods and show that the quality of text analysis in medical text documents can be improved. We compare FFTM with Latent Dirichlet Allocation (LDA) by using two different datasets and statistical tests show that FFTM outperforms LDA.
展开▼
