A classification model for a relevant document using the multinomial model will be proposed. Rules for classification are extracted from a training dataset and the rules are aggregated by using the multinomial model. Unknown parameters of the multinomial distribution are estimated by using the iterative proportional fitting algorithm. Bayesian classification method is used for classification after estimating the population distributions. Simulation experiments show that the proposed method can be competitive with the well known other classification models when two populations are similar. Applications to the TREC Conference are discussed.
Keywords: Classification; Entropy; Multinomial model
Biography: Prof. Lee has the B.S. & M.S. degree in Computer Science and Statistics from Seoul National University in Korea, and Ph.D. in Operations Research from Case Western Reserve University in USA. He was a professor of Mississippi State University in USA before he join the faculty of Soongsil University in Korea. He has been published in many journals including Journal of American Society for Information Science, Computational Statistics, Naval Research Logistics etc. He has been a member and council member of ISI since 1999 and served as the scientific secretary and council member of IASC. He has been a member of Korean Statistical Society since 1989 and served as the director General, chair of statistical computing section.