Contents automatic classification device, contents automatic classification manner and contents automatic classification program

摘要

PPROBLEM TO BE SOLVED: To improve accuracy of automatic content classification even when a procedure for giving a labeled sample is unbalanced. PSOLUTION: An automatic content classification apparatus for inputting a content whose category is unknown and classifying the category of the content includes an identification function generation part and a content classification part. The identification function generation part gives an identification function expressing the strength of dependence between the content and the category by weighted combination of a conditional probability model and a simultaneous probability model and simultaneously calculates the parameter of the conditional probability model, which is a model parameter of the identification function, the parameter of the simultaneous probability model and the weight of the weighted combination on the basis of statistical information of both of a labeled sample and an unlabeled model. PCOPYRIGHT: (C)2011,JPO&INPIT