To provide a content sorting apparatus and a program for improving and stabilizing accuracy of a sorter to be used for sorting contents.
A content sorting apparatus comprises: a structure learning means for learning structures of a plurality of Naive Bayes Trees (NBTs) on the basis of attributes of learning data; a probability parameter computing means for computing a probability parameter for each of NBs in the plurality of NBTs; a storage means for storing the structures and probability parameters of the plurality of NBTs; a simultaneous probability computing means for computing a simultaneous probability for each attribute of data contained in content data and for each category in each NBT; an average simultaneous probability computing means for computing an average simultaneous probability for each category that is an average value of the simultaneous probabilities in the NBTs computed by the simultaneous probability computing means; and a category determining means for determining a category of a sorting destination of the contents data on the basis of the average simultaneous probability.
COPYRIGHT: (C)2010,JPO&INPIT
US6182058 |
Geoffrey I. Webb, et al.,Not so naive Bayes: Aggregating one-dependence estimators,Machine Learning,2005年,p.5-24
Ron Kohavi,Scaling Up the Accuracy of Naive-Bayes Classifiers: a Decision-Tree Hybrid,PROCEEDINGS OF THE SECOND INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING,1996年,p.202-207
Masatake Shiga
Yoshifumi Saeki